First-Hand:Philips Telephone Exchanges and Denmark 1980-1990
Tales about Philips
There were of course several tales about Philips and their products. There probably are about any large company. One was about the invention of a shaving automat based on the Philishave. One should only put the head into the automat, throw in a coin and then you were shaved. The answer to the criticism that not all chins were alike was: “No, but they will become so!”
Another was about the employee, who died and came to the waiting room of St. Peter. While he sat there waiting for the interrogation he wondered over the clocks on the wall. Each went in its own speed and in very irregular steps, click - click, click - click. And under each clock was the name of one of Philips’ competitors, LME, Siemens, Sony etc.
So when he came before St. Peter he immediately said that he had a question: What did the clocks mean? Easily answered, said St. Peter, when an employee of one of those firms make a mistake, the respective clock moves forward one step.
But, said the former Philips employee, there was no clock with the Philips name, why?
Still easier to explain, said St. Peter, we use that in the kitchen as a ventilator.
Another tale about a dead employee: While on earth he had been a devout church-visitor and had each Sunday heard the preacher tell how awful a fate waited for those people not leading a proper Christian life. They would come to hell, be dipped in tar and rolled in feathers.
Well, he died and woke up on a green meadow. Around him were people looking happy and glad. He thought that he – as deserved – had come to heaven until he saw a person he knew from the firm, but how had he managed to come up here?
He had to go to him and ask.
The answer was that they were both in the deepest hell! But, said the church-goer, the preacher said we would be dipped in tar and rolled in feathers, but I only see glad people?
The veteran said: Look, this is just like when we were in the firm back on earth. Either they have no tar, or they have no feathers, and if they have both, there is no one to do the job.
Another tale is about the visitor to the headquarters in Eindhoven. He saw all the large buildings and was impressed. He asked how many people worked there? The answer was: About half.
“Do you drive Swedish?”
Maybe the companies fight intensely about orders. This does not prevent that their employees have a good mutual relation. You do of course not tell about what is in development or how prices are calculated, but you have the techniques in common, almost as a hobby, so it is easy to talk with engineers from the competitors.
A good reason to be open and friendly to them is also the ongoing fusions between companies. Who knows in which organisation you are tomorrow? Will he be my boss, or will I be his?
At the Technical University there were often lectures about new initiatives within telecommunications. I went there when the lectures were on switching and met people from LME. ITT, Siemens etc. and talked with them. They knew of course that we were the major supplier of large PABXes to Copenhagen Telephone and had replaced LME in this in the late 1970es. We knew that LME would do almost anything to regain their position. But you can talk nicely together anyhow, and that was what we did.
After such a lecture we followed each other out and it appeared that the people from LME (in Sweden) should go to the station in Lyngby and take the train. I offered to take them there (it was a small detour) and they accepted. When we came to my SAAB, one of them said surprised “Do you drive Swedish?” I could only answer “Yes, you make good cars!”
1980: “Problem” with EBX 800
Funen Telephone had several EBX 800 exchanges in operation and told me one day that they had a problem. There were so few faults in them that their technicians almost had to start from scratch when there was something wrong.
I had two suggestions: One was that they could buy and install more exchanges, this would increase the risk of failure and keep the technicians up to date. The other was based on the fact that Hilversum had back in the 1950es had a letter with the same type of complaint from the Dutch Telephone Administration about the public exchange system UR49. We would welcome a similar letter from Funen Telephone!
I did not really expect such a letter, but lo and behold: One day a letter arrived from them, in English and with exactly this content. It was of course passed on to Hilversum with our compliments.
Shortly after the letter was printed in facsimile (i.e. as a picture of the original letter with the letterhead of Funen Telephone and the text) in Telecontact, a publication sent to all possible customers and business relations. It was a valuable statement about the quality of the system from a respected customer.
This made Funen Telephone react: I got a call stating that it had not been the intention that the letter might be used in that way. Of course I apologised and promised that we would never do such a thing again. An easy promise: It would probably last several years before we got such a letter again!
But I thought that by writing the letter in English the sender in Funen Telephone had shown that he reckoned with exactly the use of the letter we had made. He only had to react towards us after criticism from others in his company.
1980: Problem in EBX 8000
One day I got a phone call from Copenhagen Telephone: They had a serious problem in one of the big exchanges of the Public Administration. The customer did not feel the problem, the exchange served him perfectly, but every evening during some time the technicians were called because one of the main computers had broken down, the exchange operated “on one leg”. The computer automatically started up again, either while the technicians were underway or the next morning.
EBX 8000 had two main computers operating in synchronism with their results compared all the time. Orders to the rest of the exchange came from only one of them at a time. If the comparing circuit detected a difference, both computers ran a program to detect if it was wrong (the faulty one in this case was completely dead and was therefore detected to be the wrong one). Then the healthy one ran the telephone traffic. It was also built in that the exchange would regularly test if the fault had disappeared. In that case the acceptance of new traffic stopped, the whole content of the data store was copied from the good to the bad computer and the exchange started in synchronism again. Copying took only a few seconds. The alarm was reset, but in the “history buffer” the fault was noted and could be read out.
The technicians from Copenhagen Telephone had looked for the fault during some time but had not found a reason. They wanted help. We called for assistance from Hilversum, who sent one of their best people, Pols, to help. He went with the local technicians to the exchange, investigated what he could and took part in the fault finding when the fault appeared. To no avail. After a week he gave up and called for assistance from the laboratory. They sent the man who had developed the computers.
We were lucky. I had fetched him in the airport and driven him to the exchange an evening, where the fault had not occurred yet. But he had only just entered the room before the exchange gave an alarm! He ordered us back along the wall while he alone started looking for the voltages supplied to the computer. He detected which voltage was missing and while he was at it the exchange started again.
After more searching and guessing at possible and impossible causes we looked closer at the cover of a fuse giving one of the voltages to the computer. Indeed, it missed the little spring which should press the fuse into its holder. As a result, when the ventilation was stopped in the evening and the room got a little warmer, the fuse missed contact in the holder. When the ventilation started again (or the room got colder for other reasons) the contact was re-established and all looked normal.
The cover was replaced and that was the end of that fault. And the cover without a spring? It ended like the pea in the fairy tale by Hans Christian Andersen: It came at the museum. Or at least it came into a plastic bag and was hung up on display in the exchange room of Copenhagen Telephone itself with a note about the cost it had meant for both them and Philips.
1980: Teletechnical Society in Stockholm
The competition to EBX 8000 came nearer. The future was digital. When the Teletechnical Society arranged a trip to Stockholm to visit LME both Øberg and I took part.
Max Hansen had retired in 1979 and was followed by Niels Øberg as head of Philips Telecommunications in Denmark. He came from Philips-BOFA, our company for public address systems and closed TV-systems.
One point was as expected their new PABX MD 110, but of course LME did not tell too much about their new system. It was, however, clear for us that they came with a system more compact than EBX 8000 (and by the way more compact than the first digital PABX in Denmark, SL1 from Northern Telecom, which was introduced in Jutland). This meant i.a. that it could probably be delivered at a lower price than EBX 8000. We could hardly avoid that Copenhagen Telephone would introduce this system!
We reported of course what we had heard to Hilversum. There was still no news about their plans for the next generation of PABXes.
But otherwise we enjoyed the trip to Stockholm and the hospitality of LME. They invited i.a. to an evening at Berns Salonger for a nice dinner and performance of Ulrik Neumann and his two grown-up children. What a night!
The 25 years jubilee of Oksholm
It must have been about this time, when Øberg had already been boss of Philips Telecom in Denmark for some time, that Oksholm could celebrate his 25 years anniversary at Copenhagen Telephone.
I do not know if it was an aftermath of the time when Jørgen Lindegaard on his insistence had become the primary contact to them or if it was later situations, where I was sceptical concerning the way they marketed EBX 8000. In any case I felt no urge to come and congratulate him. Thus, when the day came I just put on my daily outfit, not the “meeting dress”, and refused to go with him when Øberg left for the reception.
When Øberg came back he said that Oksholm had wondered why I did not come and I understood that Øberg also did not find that I had behaved correctly.
In hindsight they were right. One has no right to play insulted (or whatever I did) for just as I aimed at what was best for Philips, Oksholm aimed at their goals. And how that took place I was in no position to judge. In the end Copenhagen Telephone was our customer, so we (or in this case I) had to adapt to them!
March 1981: Disaster!
EBX 8000 (and EBX 800 to Funen Telephone) were sold in rather small numbers because they were large PABXes (for their markets). We supplied e.g. 17 EBX 8000 exchanges with about 17000 extensions to Copenhagen Telephone in the years from 1976 to 1982, or 2,5 exchanges per year. It was nice big orders, to a totally sufficient sales price, but there were too big variations from year to year, dependent on the individual orders.
Thus it suited us well that the next system from Hilversum would be a system in the middle range. It would sell much better on a rolling basis, maybe 100 exchanges per year. It would not be a chock if an order did not materialise.
Discussions about the system, TBX for Time-division Branch Exchange, started in the late 1970es. Due to the use of time-division it was very compact and cheap. It did not use digital time-division but used PAM, pulse amplitude modulation, in a synchronous system transferring 24000 values per second. This meant that the two circuits between which a connection existed were in a direct contact at this rate. It gave a small delay of the signal and a better adaptation to the analogue network.
There were some who looked with disdain at such a system, as it was not digital. But Copenhagen Telephone realised – as we did – that the analogue telephone network would exist for several years to come, and that TBX was a good offer to their customers. Thus, the discussions were almost at an end and within Copenhagen Telephone there had been meetings with other involved groups telling them about the system when we reached February 1981. Deliveries were due to start from the same autumn.
One month later Philips asked for a new meeting: Now the picture was quite different, the program development had been bought from other parties and when all program modules were brought together for test the system did not work at all. The whole program development had to be done from scratch (“There is never time to do it right, there is always time to do it over”) and there was as yet no time planning for that. This plan would only be ready after the summer holidays.
This was a serious back-down for Copenhagen Telephone (and us) and Oksholm reacted very understandably by saying that he from now on would hear nothing about TBX until we could supply one to them. Thus, after presentation of the new plans, which showed a possible delivery from the middle of 1983, we kept strictly to the line that during meetings (which were about EBX 8000) we only used a few minutes to state where we were in relation to the time planning.
Luckily the new plans were realistic. We supplied a TBX for their examination in January 1983, the first TBX for a customer in May 1983 and serial deliveries started in September 1983. Up to 1987 we supplied about 400 exchanges with about 60.000 extensions.
Spring 1981: With Teletechnical Society to Canada and USA
In spite of the delay of TBX I took part in a trip to Canada and USA with the Teletechnical Society in the spring of 1981.
The trip went via Chicago, where we only saw the airport, to Toronto. We had left on a Saturday and arrived therefore on Saturday evening. The next day was thus a day off. The secretary of the society, Bente Kjøller, was our guide and had arranged a trip by bus to Niagara. Impressive, especially to go in the small boats “Lady of the Mist” close to the Horseshoe Falls clothed in big raincoats and southwesters.
In Toronto we visited the local telephone company Bell Canada and we then went to Ottawa to visit the development centre of Northern Telecom. They were on the Danish market with their deliveries of the SL1 PABX to Jutland Telephone.
Then to New York, landing in Newark, N. J., and entering Manhattan via the tunnel under the Hudson River. We stayed in the Holiday Inn close to Central Park.
In New York we visited IBM about whom we were still in doubt. Would they enter the exchange market in a serious way? Their PABX had never really been a hit and it was still some years before they bought Rolm (which was later sold to Siemens). Even if you have the technical base it is difficult to adapt to the market! We did not get any clear answers, as expected! But just to show how far microelectronics were in 1981: We received key rings with chips in them. Nice to see that IBM also can make bad components, for if they were good, they would never end up in a key ring! The chips were 3 mm by 3 mm and had on one side in a square 2 mm on each side 11 by 11 or 121 contact points.
We also made a trip to Connecticut, to the development centre of ITT for their System 2000 public exchange. Here we met Poul Friis, who was an employee of Jutland Telephone. They showed an impressive film and dias show, with film projector and 8 dias projectors in front of the screens and as many behind them. Yes, screens: the front one was transparent to make the show three-dimensional. System 2000 was of course the answer to all problems and mightily advanced. I could not help asking Poul afterwards if it should not rather be called System 2001 after the film by Kubrick? He seemed to agree with me, he was not satisfied with the development. When the system later came to Jutland it did also not live up to expectations.
We also visited the studies of CBS, where we i.a. saw the news room where Walter Cronkite during many years had told Americans how the world looked. I was surprised over how low-ceilinged the room was and they told that it was a problem to keep the cameras tilted low enough so the ceiling did not appear in the picture! They knew of course the joke about the meaning of the acronym for the US colour-TV system NTSC, Never Twice the Same Colour, but the significance of the acronym for the French system was new to them: SECAM, Systeme Evolué Contre les AMericaines, as was the acronym for the other European system, PAL, Peace At Last!
We also had a weekend in New York. I choose to walk from the mid-town, where we stayed, to the Southern tip of Manhattan, a walk of 8 km. Then I went by boat to the Statue of Liberty and later by subway back again. The next day I paid a visit to the AT&T exhibition with many interactive demonstrations of telecommunications methods. In company with Ihle from Copenhagen Telephone and the head of PABX sales from LME I was in the Radio City Music Hall for the big show with Rockettes, cars and horses on the scene etc. It was during Reagan’s first year in office, the hostages in the embassy in Teheran had recently become free and the feelings were obvious when the song about the yellow ribbon was sung from the scene.
I had of course also to go to the Eastern Manhattan, to the (then) IEEE headquarters and the United Nations.
We had a dinner in the Rainbow Room on top of the RCA building before we again flew back to Denmark.
Summer 1982: SOPHO-S task force
The delay of TBX gave us a problem: What should I do in the meanwhile? There were not much to do about the EBXes and we were not allowed to talk about TBX until we could realise the system.
Øberg took this up with Hilversum and they had indeed a job for me. A job of short duration, one month in total, but very important for them.
Around 1980 the development of the successor to EBX 8000 had been started. It should be a digital system, based towards the system sets on the extension side on the 2B+D interface recently decided for in the CCITT for the ISDN network (i.e. 2 channels of 64 kbit/s and one of 16 kbit/s for the user, simultaneously in both directions, CCITT is the forerunner of ITU-T and ISDN means Integrated Services Digital Network, a concept overrun by other developments like putting digital channels on top of the analogue speech, ADSL). Now, two years later, there was a need to collect the results of this development which was spread over several people in the laboratories. Thus a task force was made, consisting of F. Kluizenaar from the laboratory (he had worked with PABX development from before I started at Philips), Peter Tanis from the Dutch sales organisation (who had worked in the laboratories until the spring of 1982) and me.
Our task was, during three weeks in June 1982, to make a summary of the development in a report, which was to be the basis of the further development of what was later to be called SOPHO-S 2500. The report should then circulate in the laboratory during the summer months and in the end of August it should be discussed at a meeting before we finally edited the report. For this two weeks were considered necessary.
We did more than that and made all sceptics feel ashamed, all those who did not think that the job could be done in so little time. We made the report up in both Dutch and English within the appointed time! The reason was especially that we could divide the job over the three of us, they wrote in Dutch, which I later translated, and I wrote in English, which they translated. And we had an efficient secretary (we had not yet reached the PC age, where each person writes his own reports). The careful treatment you have to give each sentence during translation was also basis for extensive discussions between us, and this added to the quality of our report.
Well, such a report shall not be all dry technicalities, so I also got an Italian sentence on the front page which I had read somewhere. I have later found that it is a quote from Giordiano Bruno. It fitted well to the report: “Se non e vero e ben trovato” or “if it is not true, it is well put together”.
When I left Philips in March 1991 I received a greeting from Kluizenaar, who still worked in Hilversum, i.a. editing “Philips Telecommunication Review”. He reminisced about our common task nine years earlier and wrote that our report was still used as a reference if there was doubt about some topic in SOPHO-S!
Summer 1982: Commercial Requirements Specification
During the three weeks of writing and two weeks of editing I should not only participate in the work on the technical description of SOPHO-S. I should also with van Heck in the commercial section write a Commercial Requirements Specification for the system, that is a description of all the facilities the system should offer to the customers and those it was a wish that it could also offer. It was to be one of the ground pillars of the contract between the sales section and the development section: What did the sales group expect that customers would ask for in four years time when the system should be put on the market, and what should therefore be put into the system by the developers?
Compared with the work with Kluizenaar and Tanis I rather took this job as a left hand task. I did as much at it as the others in the sales section, but for me it had much smaller importance than the technical report.
But our work resulted also here in a very useful paper. So useful that several years later, when I had almost completely forgotten it, it was this aspect of my work in the summer of 1982, Øberg remembered. We met accidentally in Hotel Danica in Horsens in 1994 after we both had left Philips. He left in 1989 to become head of the business division of Copenhagen Telephone (after Oksholm had been replaced in that position by Bent Nilskov and later Jørgen Michaelsen), but left again in 1982 after a controversy with the CEO of Copenhagen Telephone, Jørgen Lindegaard (the same who started at Philips in 1975). Øberg then became CEO of Alcatel in Denmark who had a factory in Horsens. I left Philips in 1991 for a job in Øberg’s business division within Copenhagen Telephone (CTBD), but had moved to the public telephony division of Copenhagen Telephone in the autumn of 1992 when all business divisions of the Danish telephone administrations were joined together. While at Philips I had joined the PABX-group, preparing standards for the connection of PABXes (in the liberalised market) to the public telephone network and had remained in the group during the later employments. The PABX-group was in Horsens for a meeting at Alcatel about the standards (and I had brought my wife, so we could have a couple of holidays in central Jutland afterwards). We talked about the good old days, but when Øberg mentioned my work for the CRS I had to dig deeply in my memory. I only clearly remembered the technical report!
Summer 1982: Oosterschelde
It was not just work during the three weeks we wrote the report. Tanis’ uncle was a surveyor at the big final project of the Deltawerken in Holland, the blocking of floods in the Oosterschelde. So Tanis invited me on a trip on a Saturday to the building site.
Deltawerken was begun after the big floods of 1953 in Zeeland and deep into central Holland. The most exposed places were protected first, at Veere in Zeeland and at Gouda in Holland. Then the outlet of the Rhine to the North Sea was protected at Haringvliet (we had visited it in 1963 while we still lived in Holland) and now came the last chapter, the protection of Oosterschelde, a bay deeply into Zeeland.
Typical for the Dutch they had thus made a thirty year plan and kept to it for the protection against the North Sea. The flooding in 1953 was due to water levels only occurring once per hundred years, but that is no guarantee against their occurring again both in 1960, 1970 and 1980! One can imagine the panicky solutions politicians would go for in other countries. But not in Holland. Here it was first evaluated where the danger was the greatest and the possible solutions, and then the work started.
At Oosterschelde this had implied that the dams around the bay had been made higher so they could withstand normal floods while the final solution was discussed. A proposal was to make a high dam across the mouth of the bay, slowly making the bay change into a fresh water lake. There were large fishing interests against this and also the shell fishing would be spoiled. Thus it ended with a solution with a row of gates across the bay, normally open to the sea, but if there was an extreme flooding they could be lowered and protect the bay.
The gates were hung in pillars built in concrete. They were built on the seabed within a dam and Tanis’ uncle surveyed the pumping of water from this dam during the building period.
Thus, on a Saturday I drove to The Hague where Tanis lived, fetching him and his wife for a trip to the South. First we went to his uncle in Zierikzee for a cup of coffee and then with him down to the artificial island. A fantastic sight! Here were maybe 30 pillars, each with a footprint of about 100 by 50 meters in one big building site. Each pillar had a conical base, from which a tower went up about 50 meters with tracks for the gates.
We could drive in between the pillars and enter one of them. It had had fine visitors a few days before from the Dutch Royal Family and the princes had made grafitti inside, writing their names. The room itself was as big as a cathedral and with the same echo! When all casting was finished, water would be let into the site, the pillars would float and be pulled to their place. There water would be filled into them, later to be replaced by sand and stones, and then the gates would be hung up.
Years later I drove on the motorway over the pillars when the whole dam was finished. I was on my way to Brugge with a group of technicians from CTBD, who were (with me) on a SOPHO-S course in Hilversum and enjoyed a weekend off during the course. And saw to my astonishment that the careful Dutchmen had calculated wrongly: behind the dam stood a superfluous pillar!
It was late before we again went back, first bringing the uncle home, then dinner at Tanis’, so I only came back to Hilversum when it was near midnight.
PSTN, FDM, TDM, PCM, PDH, POTS, PABX, PANS, IDN, ISDN, IN, SDH, ADM, STM, ATM
It is time for a summary of the development within telephony from 1959 to 1997, the years I was a (small) part of it. You will find a similar survey in the LME publication on the development of the AXE-system “A Switch in Time”.
Up to the 1980es exchanges and transmission equipment for the PSTN, the Public Switched Telephone Network, were developed separately and at the customers they were bought by different departments. The interfaces between exchanges and transmission equipment were separate lines for each individual conversation. In addition to the speech connection it was of course necessary to transfer signals from exchange to exchange to announce new calls, digits, releases etc. but the task of the transmission equipment was only to transfer speech and signals with the slightest possible distortion from exchange to exchange. And to do it as cheaply as possible.
During the 1960es the transistor really came in use in the transmission equipment, which was still based on FDM, Frequency Division Multiplex, i.e. each call occupied its own 4 kHz wide band on the common line. There were two lines of development, for the long distance lines (Copenhagen – Aarhus – Kolding) still more calls via the same (coaxial) line until it was possible to send 15.000 calls in a 60 MHz band through one line. Each call cost less and less to transfer, but the customer prices did not fall as fast so the telecommunication side of the public administrations (often by tradition run together with the postal service) was a gold mine for the states. The other line aimed at shorter distances and was to transfer more calls on twisted pairs, which were originally meant for just one call each. Here electronics meant that it became more economical to install such equipment than to lay more cables, first for distances of over 30 km, then over 20 km, but still not within the cities.
In the 1970es TDM, Time Division Multiplex, came in use, where PCM, Pulse Code Modulation, was used to sample 8000 instantaneous values per second for each call, encode each sample into 8 bit and join these 64 kbit/s with bits from 29 other calls to a total of 2 Mbit/s on he line (including 128 kbit/s for synchronisation and signalling). More 2 Mbit/s transmissions could be joined into a stream of 8 Mbit/s, 34 Mbit/s or 134 Mbit/s (each higher step joining 4 systems from the lower step and adding some synchronisation) in the PDH, Plesiochronous Digital Hierarchy. Plesiochronous means that each low step system ran at its own clock (within certain bounds, of course) such that a 2 Mbit/s system entered and left at its own sender’s clock in spite of the use of higher steps during the transmission. PCM made the distance where a transmission system was cheaper than a new cable still shorter, especially within cities. But the interfaces were still individual for each call. If one e.g. had a 134 Mbit/s system along an exchange, in which 2 of its 2 Mbit/s systems should be terminated, one had to demultiplex from 134 to 34 to 8 to 2 Mbit/s, take out the two systems required, possibly replace them with other 2 Mbit/s systems and multiplex up to 134 Mbit/s again. The two 2 Mbit/s systems taken out had to be demultiplexed down to individual 64 kbit/s calls and be decoded to analogue speech before the calls could enter the exchange.
In the 1960es exchanges were still controlled by relays and the calls used individual paths through the selectors, rotary or crossbar. In Denmark register systems had been used since the 1930es where the most complicated control circuits were in use for only a few seconds at the start of each call. Thus, only a few of them were required and they might therefore be both complicated and expensive. But the exchanges could only deliver POTS, Plain Old Telephone Service, where one could go off-hook, dial a number, talk together and go on-hook again. In PABXes, Private Automatic Branch eXchanges, serving a company, there was one service more: You could during a call to the public network give a signal to the PABX, causing it to hold the external call and give you a dial tone. You might then dial another extension number, talk with its user and transfer the external call to him by going on-hook. This facility was not introduced in public exchanges, it would be too expensive to attract customers.
In the 1970es computer techniques were finally reliable enough to be used in telephone exchanges and the most complicated control circuits could be replaced by electronics. But the selectors were still mechanical, with contacts for each call, whether it was in reed relays as in Philips’ PRX or code selectors as in LME’s AKE or other systems installed during these years. The computer control opened for quite new services which had been totally uneconomical in relay exchanges, such as automatic call back, where you – if you get a busy tone – can choose to let the exchange keep an eye on when the busy user becomes free. It then makes a call to you and when you answer it makes a call to the now free party.
The new services were soon called PANS, Pretty Amazing New Services, in opposition to POTS, and were first introduced in PABXes like ETS 8000. Although such a PABX provided many new services, the users only exploited a few, mostly because one had to dial digit combinations to use them and customers could not remember them. The use of facilities in Copenhagen Telephone’s own PABX was measured and published in an article in Philips Telecommunication Review in 1979. Only common abbreviated numbers (called just like extension numbers), automatic call back and call forwarding (moving calls to one’s telephone to another extension) were used on a regular basis.
During the 1980es the mechanical selectors were replaced by electronic switches, in analogue and later digital time multiplex exchanges. Let us imagine a digital exchange with 30 analogue interfaces. For each interface the exchange has an encoder changing the analogue speech into a 64kbit/s bitstream. The streams from the 30 interfaces are collected into the 30 channels of an internal 2Mbit/s system. Likewise there is an internal 2Mbit/s system towards the interfaces and the 30 channels are decoded into the 30 analogue speech signals on the individual interfaces of the exchange. The selector is a memory for 30 times 8 bit. The incoming signals are written into the memory in a place for each channel. When writing has taken place for channel 30 the place for channel 1 is rewritten again and this occurs 8000 times per second. If interface 2 and interface 17 shall be connected, place no. 2 is read into timeslot 17 in the outgoing direction, and likewise place no. 17 is read into the outgoing time slot 2. In stead of a selector of 30 by 30 positions having 2 contacts each (1800 contacts) there is now a single memory chip of less than 1000 bit in almost no space. Naturally the interface circuits are more complicated than in an electromechanical exchange, but this is mass-produced electronics and therefore cheap. Thus the digital exchanges were introduced, using about one third of the space for an electromechanical exchange of the same capacity. If they cost the same per kilo, they should also cost one third of the price. However, an exchange does not consist of only hardware. The programs, software, and the adaptation to the wishes of the customer must also be paid for. These tasks could only be automated to a certain extent (the collection of customer requirements not at all), and work got still more expensive. But anyhow the digital exchanges could be priced at 80% of the price of a similar “old” exchange.
And now came a jump cutting a further 20% off the price: Until now the 2 Mbit/s of the transmission system was decoded to the 30 individual analogue interfaces to the exchange. In a digital exchange these analogue signals were again coded and multiplexed into the 2 Mbit/s system entering the selector. The whole coding and multiplexing operation could be saved if the 2 Mbit/s system from the transmission equipment was led directly into the selector, letting the selector split it into the individual channels. This was the idea behind the IDN, Integrated Digital Network, which integrated exchanges and transmission systems into a whole. With a full introduction of IDN a call is only coded once, at the entrance to the network, and only decoded once, at the exit.
IDN was introduced at a large scale during the 1990es, and included ISDN, where S means services. In ISDN the interface between analogue and digital technique has moved to the individual terminals. Coding into 64 kbit/s and decoding to analogue speech takes place in the telephone sets themselves. With 64 kbit/s (and even two of these signals to each user) the channel between user and exchange can alternatively be used for data with this speed, far more than could (at that time) be transferred via the analogue network.
At the same time more and more services appeared, which at the price they could be sold for were attractive for the users. Free call and overtaxed services (calls were either free of charge as a part of the marketing of a company or calls were more pricy than normal, but the charge was put on the telephone bill by the operator who made it up with the service provider), countrywide numbers (where you called the same number throughout the country, but the operator routed it to the nearest service centre of the company or to different centres dependent on the time of day (during the night only one centre for the whole country)) etc. It would be too much if each exchange should have the capacity to handle all these services, especially if they were changed all the time. Therefore these services were concentrated to a few computers in the network in the IN, Intelligent Network. The many exchanges should only recognise digits characterising a call to be an IN call, ask the IN computer what to do, and act according to the answer, e.g. routing the call to a certain terminal.
What else happened during the 1990es? There was especially a development at the transmission side, where the SDH, Synchronous Digital Hierarchy, was introduced. The whole network is synchronised to the same clock rate and in a block of bits for many connections one can find the bits belonging to a certain call, take them out and replace them with bits for another call without having to demultiplex the whole SDH signal. This takes place in an ADM, Add and Drop Multiplexer. The method is called STM, Synchronous Transfer Method, and includes in addition to the user channels data about the transfer, which the equipment along the transmission line will need. SDH works with rates of 150 Mbit/s or 4 or 16 times this rate, far more than coaxial lines can transmit. It is only because copper has been replaced by transmission in glass fibres that SDH has become possible. SDH on fibres may even be extended to individual customers who need these rates.
In the SDH user bits can be transferred using ATM, Asynchronous Transfer Method, in some of the or in the entire place reserved for the users. ATM does not mean that each user works at his own rate on the transmission line, but that the user’s bits are collected in cells of 48 times 8 bits and transferred with an introduction of 5 times 8 bits when the cell is ready. For speech, coded into 64 kbit/s, there is a cell each 6 milliseconds (167 cells per second), while e.g. a TV signal coded into 10 Mbit/s has a cell ready each 38 microsecond (26.000 cells per second). The introduction tells exchanges underway what the destination is. Transmission only takes place when there is something to send. During speech pauses one can e.g. stop sending completely. This transmission method is well suited for data, where a user often sits looking at a picture for a long time before a new small packet of bits shall be transmitted. But even for speech ATM can put more users on the same line than if each user all the time used a certain share of the capacity. Will the end result be that the transmission systems are STM with the superior possibilities for control of the transmission, but that user bits are transferred using ATM from exchange to exchange with the exchanges handling the cells?
One question is whether the users need the large number of bit/s? Is it cheaper to rent a video film in the local store or get it via the network? Is it cheaper to have the full data handling capacity in your PC or to have a simpler terminal and let the network do the data handling?
Anyhow, the use of mail addresses ending in @ieee.org has only become possible as transmission costs almost nothing. Imagine the costs in the “good old days” if every letter to you had to be sent first to New York to be examined for fraud and then retransmitted to your local address. And imagine the delays this would entail!
1983: The TBX exchange
Finally TBX was so far that it could be supplied to Copenhagen Telephone. In the end of 1982 the first one arrived in Denmark and was passed on to the laboratory of their business section, where Mogens Thomsen was their contact. He was – as far as I know – originally a technician, but had shown flair for the commercial side. An ideal mix (applying also to me, I think) when sales people have a technical insight. I was in Hilversum several times during the autumn to learn about the system and on January 9th, 1983 (the day a storm blew the copper roof off the building between Christiansborg and the castle’s church, killing some people at the bus stop below) I was in the laboratory to put on power and teach the people in the business section about the system.
The exchange worked as it was supposed to and it led to more courses during the spring and the first proper order when the people in the business section were convinced that all was in order. It was for the social services of the Frederiksberg Community, in the neighbouring house to the business section. It arrived in the middle of May and was installed and started up without any problems. Then sales to other customers could start.
What was special for TBX? Above all that it worked on the time multiplexing principle and what does that mean?
Older exchanges used space multiplex, i.e. each call had its own path through the exchange and all ongoing calls were present simultaneously and at the same frequencies. Early transmission systems used frequency multiplex, where all calls existed all the time on the same line, but each call used its own frequency band (like the stations at a radio). In time division all calls use the same line and the same frequencies, but each call only gets a short time access to the line, interleaved with the access of the other calls. If a call has access 8000 times per second (each 125 microseconds) to transfer the instantaneous value of the speech voltage, frequencies up to 4000 Hz (the normal speech band for telephony) can be transferred.
This was proved already in the 1940es by Claude Shannon at Bell Laboratories in the USA and LME had conducted lectures in 1958 at the Technical University in Copenhagen (where I was spotted by Max Hansen from Philips). If each transfer takes only 1 microsecond there could be sent 125 calls over one path.
Time multiplexing was already introduced as PCM, Pulse Code Modulation, in transmission systems and in PABXes like SL1 from Northern Telecom in Canada and MD110 from LME. Philips considered that PCM in exchanges was still too expensive and it was only introduced in SOPHO-S in 1985.
But the instantaneous values of the speech voltages can be transferred by other methods than PCM, and this could be done in both a compact and cheap way using transfer of the analogue values, also called PAM, Pulse Amplitude Modulation. This was the method used in the LME exchange of 1958, but there were now components available to realise it in a serious way, in TBX. This exchange, for about 50 to 600 extensions, was a really competitive product in this interesting range of capacities.
In PAM each external 2-wire line (on which speech goes in both directions) ends in a hybrid circuit, a circuit which divides the connection in one path for each direction (as it has always been done in transmission systems). Incoming speech goes to a gate, controlled by the exchange. When this gate is opened, the speech voltage gets access to a common input line, a “highway”. From this highway the signal can through another gate, opened simultaneously, enter an output highway and finally, through yet another open gate enter the outgoing side of the hybrid circuit of the other party to the call. Similarly – but in another time slot – instantaneous speech voltage samples are switched through from the circuit of the other to the circuit of the first party.
In TBX it was chosen to transfer 24.000 instantaneous values per second for each direction of each call. This is three times more than strictly necessary, but was chosen such that cheap filters could be made to cut off higher frequencies. An incoming signal must not contain frequencies higher than half the sampling speed, as such frequencies will give rise to aliasing, i.e. if you sample a 6000 Hz signal at 8000 times per second, it will result in an output of 2000 Hz. In PCM systems (exchanges or transmission systems) the filter must cut off rather sharply, from a small attenuation at 3400 Hz to a large attenuation at 4000 Hz. In TBX the large attenuation only had to be obtained at 12.000 Hz, meaning a less steep filter characteristic and thus a cheaper filter. Each transfer lasted 0,9 microsecond including time to discharge the highway between two samples, so a highway could serve 48 simultaneous calls (48 times 24.000 is 1.152.000 samples per second). Having about 100 external lines (to extensions or public exchange) to a highway, about half of them could be in use at the same time, and this is a more than adequate traffic capacity. It also fitted well with the possible number of external circuits in each mechanical unit, a shelf, in the exchange.
If there was just one shelf, as in TBX 100, the two highways could be connected directly. But already in TBX 250 with two shelves a permanent connection between the four highways would mean that only 48 external circuits or about one fourth of the maximum possible number could inject a signal into the highways, limited by the total number of time slots. In stead it was chosen to introduce gates from each incoming highway to each outgoing. Thus a certain time slot could be used in both shelves for a connection within that shelf. Or it could be used for two connections between the shelves, one from shelf 1 to 2, the other from 2 to 1. The program for making connections should only find a timeslot out of the 48, which was not used on the incoming highway of one external gate and on the outgoing highway of the other. And find a similar timeslot for the connection from the other gate to the first.
In TBX 1000 this was extended to 8 shelves, with 8 gates from the incoming highway of a shelf to the outgoing highways of the 8 shelves.
In each shelf there were 18 places for circuit cards for external lines (to extensions or lines to the public exchange) and local services (e.g. receivers for tone signals from extensions, keytone signals). A circuit card could serve 8 extensions or 4 lines to the public exchange. An exchange fully equipped with 16 cards per shelf for extensions (and two places for keytone receivers) would have a capacity of 256 or 1024 extensions in 2 or 8 shelves. Thus came the number in TBX 250 or TBX 1000. At Copenhagen Telephone, with rather many lines to the public exchange, we never reached more than about 700 extensions in a TBX 1000, but in cruising ships – without lines out of the ship – we reached 1000 extensions.
Copenhagen Telephone did not introduce TBX 100 and TBX 250 as they were suggested by Hilversum, i.e. installed in one or two shelves in a cabinet for wall mounting. They did not trust the walls in modern office buildings. They did not at all introduce TBX 100, while TBX 250 was installed in the same type of floor cabinet as TBX 1000. The additional price was negligible.
As regards the mechanical set-up there was another essential advantage compared with EBX 8000. In both exchanges all cables between shelves (in one or in more cabinets) were mounted from the front, such that the cabinets could be placed against a wall or back to back, and that was an advantage. But in EBX 8000 many cables were plugged into sockets on the back panels of the shelves, and with the narrow space in the shelves one often had to remove circuit cards just to get access during expansions with their rearrangements of cables in an exchange. In TBX all cables between shelves were connected to the front of the circuit cards and they were consequently considerably easier to work with.
Time multiplexing meant a much more compact exchange. While EBX 8000 had a foot print of 6 m² of floor space for 1500 extensions or 0,4 m² per 100 extensions (which was one third of the foot print of an electromechanical exchange) TBX 1000 had a foot print of 1 m² for 700 extensions or 0,14 m² for 100, or a further reduction to one third of the floor space. It was of course not just time multiplexing which caused this reduction (SL1 took up as much floor space as EBX 8000 for the same number of extensions) but also the new components.
Now to the computer part of TBX. The smallest one, TBX 100, did not have more to do with its few extensions than that one processor could do the whole job. It was a requirement to make the exchange easy to install and the main part of the data were therefore burnt into a read-only memory (a PROM).When the exchange was powered up, the data were read from this memory and the exchange started working. Some data could afterwards be changed during operation, but it was too few according to Copenhagen Telephone who did not introduce this system.
In TBX 250 the processing was divided between two processors on the same circuit card, a CPU (central processor unit) and a PPU (peripheral processor unit). The latter took care of all routines like scanning all external circuits all the time to detect new calls. If there was a new call it notified the CPU, which took care of all the more special things in the exchange like which timeslots to use for a connection, digit analysis etc.
In TBX 1000 there was one CPU and up to four PPUs, one for every two shelves.
At the start of a TBX 250 or TBX 1000 exchange one connected a suitcase with a processor and compact cassette station to a socket at the CPU card. From the cassette the data for the exchange were loaded and then it could start operation. This gave the flexibility wanted by Copenhagen Telephone in the set-up of the exchanges. Several data were of course repeated from exchange to exchange, but it was e.g. essential that the placing of circuit cards could be controlled individually per exchange.
Analogue or digital?
As already mentioned TBX used analogue time multiplexing (PAM), while there was already from about 1980 digital PABXes (PCM) on the market. In the public network digital exchanges appeared also, the first AXE from LME was ordered in 1980 for Aalborg. The first PCM PABX in Denmark was SL1 from Canada, where the first one was installed in Jutland in 1980, while LME’s MD110 came on the program of Copenhagen Telephone in 1982.
Therefore the people of Jutland Telephone considered the decision to introduce TBX in Copenhagen to be a step backwards. The future was digital! True enough, but especially in the interesting range up to 600 extensions digital exchanges were still something to come in the future. With TBX Copenhagen Telephone introduced at an early moment a compact, cheap and reliable product.
SL1 took up as much space as EBX 8000 for the same number of extensions and was much more expensive. But it was still monopoly time, where customers were bound to the local telephone company, so if Jutland Telephone would not introduce EBX 8000 their customers could not get one installed. I remember from 1981 that it was with a short interval reported in the newspapers that the local administrations in Odense and Aalborg had got a new PABX each, resp. EBX 8000 and SL1. Both had about 1500 extensions and an equal number of lines to the public network, but the systems differed much in their lay-out. EBX 8000 could not have remote units placed far from the exchange itself, so every extension in the suburbs was connected by its own wire pair to the exchange at the town hall. SL1 could be divided in more units with digital transmission between them, meaning that the extension lines were shorter. According to the press Odense had paid 10 and Aalborg 14 million kroner for their new exchanges.
In the beginning SL1 was only digital within the exchange itself, while all extension lines were analogue. It had system sets (telephone sets unique to the system) but they used two wire pairs to the exchange, one for analogue speech, one for digital signals from the function keys of the set. It was a lot easier to use the new facilities in the exchange from these sets. EBX and TBX did not have system sets, and to use facilities one had to dial complicated codes from the keys of the ordinary sets. This was a major draw-back.
MD110 was no success in the first years. Just read the pitiful story about the exchange for “Privatbanken” in the publication from Copenhagen Telephone Business Division (CTBD) when this subsidiary of Copenhagen Telephone was formed in 1983. It lasted until 1986 or 1987 before LME could deliver a reliable program package. The last EBX 8000 had been delivered in 1982 and from then on CTBD only supplied MD110 exchanges to their major customers. In 1987 Philips could deliver SOPHO-S, but except for very special cases it had to wait until Philips after the liberalisation could deliver directly to the customers.
While I find it very right of CTBD (or its forerunner) to promote EBX and TBX in stead of the old reliable exchange systems, I can on the other side not reproach the telephone companies that they settled for SL1 and MD110 before they were mature enough to be installed at the customer sites. It was in fact more reasonable to settle for these systems than for systems from Philips, as the suppliers, Northern Telecom and LME, were considerably larger and well established as exchange suppliers than Philips.
MD110 had digital system sets with one wire pair. Both speech and signals for function keys and display were sent digitally over the wire pair. 64 kbit/s for speech and 32 kbit/s for signals. It had to be economically justified, so for the parting of the two directions of transmission the ping-pong method (or TDD, time division duplex, as the official name is) was used. The sending of 96 kbit/s in either direction was realised by sending short bursts at 312 kbit/s alternatively in one or the other direction. This fitted well with the normally short extension lines of a PABX.
In SOPHO-S Philips chose from the start to send digitally to and from the system sets just as in the ISDN, 160 kbit/s in both directions simultaneously on one wire pair (of which 144 kbit/s was user data and speech). This required echo-cancelling in components which were not economically justified before around 1990. Echoes of the signal sent could be 10 to 100 times more powerful than the signal from the other end, which is what one is interested in. An echo-canceller calculates which echoes the sent signal causes from the line and subtracts them from the received signal. Only the signal from the far end is then left over!
Both price and properties spoke for the digital exchanges in the 1990es. The EBX exchanges were replaced in large numbers. Of the 17 EBX 8000 and 9 EBX 800 which had been delivered from 1976 to 1982, there was at the end of 1996 only 3 EBX 8000 exchanges (of which 2 in hospitals, it was still a reliable system) and 1 EBX 800 in operation. In the 1970es it had been foretold that the new exchanges would only be used for about 10 years, although they could work for a much longer time. I had not believed it, but was proved wrong.
Summer 1983: Expert system
Some words about the techniques of TBX.
All programs in the CPU and PPU were stored in PROM memories (Programmable Read Only Memories), i.e. memory chips where the programs were loaded once and could afterwards only be read. This was also called firmware, in opposition to hardware (programs in wiring) and software (programs in memories, the content of which can be changed). In spite of this TBX was very flexible as data was loaded in volatile memories and controlled almost everything. Data were divided in project data, i.e. data with a small probability of change during the lifetime of an exchange, and administrative data, i.e. data subject to repeated changes. Project data are e.g. the frequency and periodicity of dial tone, busy tone and other tones in the exchange, the periodicity of the ringing voltage, the placement of extension and exchange line cards in the shelves, whether a given fault should raise a major or minor alarm and the result of number analysis (the range of extension numbers, access number to exchange lines, numbers for invoking call forwarding or automatic call back). Administrative data were e.g. which circuit had a given extension number, which circuits were in operation or were blocked, to what facilities an extension had access etc.
Project data had to be loaded into a TBX from the MPU, Maintenance Processing Unit, a suitcase with a processor, a keyboard and a compact cassette station. Data were prepared on a cassette tape and when an exchange was started up both project and administrative data were loaded from the MPU. Later changes in the administrative data could be done from the MPU or from the operator set.
Data in TBX were encapsulated, there was only access to them via the protected loading programs. You had no information about where certain data were placed in the memory or what the bits signified. This was of course done so as to protect the exchange against possible changes which could stop its operation. But it also meant that if project data were to be changed a new tape had to be prepared as well as a new tape with actual administrative data. Then you waited until it suited the customer, stopped the exchange and loaded the new data. This took at least a couple of hours, and as one of the project data was the time before a call should be automatically forwarded and this was a point where each customer had his own opinion (and changed it as time went by), it was a minus for TBX and CTBD considered it rightly to be less flexible than other available systems. At this time they also introduced the OCS from Standard Electric Kirk, which could have its time limits changed without breaks in the operation, so we heard much about this.
There was always just one CPU in an exchange. Its functions were limited, so it could be on one circuit card (plus smaller memory cards) and this made its reliability large enough to have no need for duplication, just as it was the case in EBX 800.
In every shelf there was to the right a power supply from 220 V AC or 48 V DC. The latter was used if one wanted an external rectifier and a battery, so the exchange could work even if the power network failed. In the shelf with the CPU there was also an NiCd-battery, which in case of a power failure would keep the data in the memory intact for 96 hours. If power came on again within this time the exchange would start up automatically.
To the left in each shelf there was room for processor cards and for connections between the shelves. First there were three places with cards of half height. In the CPU shelf the top positions were used for memory cards. The first generation needed all three positions, but already the year after only one position was needed. The two other positions could then be used for processor cards for optional functions like ticketing (recording of called numbers, metering pulses, start time and duration of calls) or hotel functions like wake-up calls. The bottom positions were used for cards for connection to other shelves, for the speech connections or processor connections. Then there was a place for the processor, for the CPU in one shelf and for PPUs in up to four other shelves. In the remaining shelves this place was empty. The next card was for one operator and for adaptation between the PPU and the other cards in the shelf. Then the 18 positions came for cards for extensions, external lines and tone receivers.
All cards of one type were identical, thus there was just one type of PPU card. But a PPU had to know its number in order to be loaded with the right data. This was done with an address plug on its front edge, a plug with 1 of 4 different wirings indicating the PPU number.
The first TBX at a customer of CTBD was put into operation in May 1983 and they started selling the system. Some standard configurations had been worked out, but they could always be reduced if the customer did not need all the circuits in the beginning. The orders from CTBD only called for “a TBX with x extensions, y lines to the public exchange, z operator sets and battery or power grid supply”. Our order to Hilversum had to translate this to a list of type numbers for each part of the exchange. I took care of this order handling.
Soon several orders arrived and with a relatively short delivery time it was not very appropriate that I was due for a summer vacation of three weeks. Others had to be able to pass on orders. Luckily the group had just got its first Philips PC for home use to see if we could use it in our daily work. It could be programmed in Basic, and I had tried this a few times before on small problems (e.g. once on a terminal under the Danish communities data centre in our public library where I had taken our girls with me to show how their school mathematics could be solved). This was just what I needed for the order handling. I wrote a program asking questions, which could be answered based on the order from CTBD, and when all questions had been answered the PC could print the draft of a telex to Hilversum with all type numbers, their amount and both piece and total prices. And then I went on vacation after having shown the others in the group how to use the program.
When I came back they had handled and forwarded 17 orders and my review of them showed them to be correct in all respects. The program was used for every order for a TBX after this, it was translated to other PC types available to the group by J. H. Bojsen, and he also wrote – when the time came for this – a similar program for the SOPHO-S exchanges.
Several years later, when “expert systems” became a fashion, I discovered that this was what I had developed back in 1983!
Foundation of Copenhagen Telephone Business Division
I have already called our customers CTBD a few times, but it was only in 1983 that they changed from being a section under the main company to be a subsidiary with their own management and supervisory board. It was still owned 100% by the main company. At the same time Oksholm was replaced as head of the division by Bent Nielskov.
Oksholm changed to the supply section of the main company and later to the personnel section (or was it the other way round?). He then left the company for a while, but when Øberg took over as head of CTBD in 1989 he took him on as a consultant and made, I am sure, great use of his experience.
Bent Nielskov had a quite different way of tackling the business than the technically based way, which was mine. But this was a problem for Øberg, my contacts to people lower in the division continued rather unchanged.
KBX 100 to “Filipinas”
In 1983 we received a request for a telephone exchange for a school ship for the Philippines, which was being built at the shipyard in Aalborg. It was financed by the Danish official support organisation Danida and the construction of the ship had begun in Elsinore. This shipyard went broke and the further building of the ship moved to Aalborg. It was a rather big ship (for that time) and the intention was that it should sail as a normal freighter while navigators learned the trade on board. It turned out to be a scandal, the ship was never used as intended. One reason might be its relation to the politics of the former rulers. In any case Imelda Marcos came to Copenhagen to give the ship its name. Then the ship sailed away and appeared again 10 years later in the news on the TV as a heap of rust, offered for sale at 10% of what it had cost to build (including the excess price the shipyards were accused of taking because the state paid).
Our part should be an exchange for about 50 extensions in cabins etc. TBX was a possibility in its smallest set-up, TBX 50, but we had an even cheaper solution in this case: the KBX100 from Pye-TMC in the United Kingdom. It was created as a line selector system based on the use of system sets with keys and lamps, but could also serve normal telephone sets. A set-up with one system set, two external line circuits (for connection to the shore when in a harbour) and a number of keytone sets from Siemens (bought via Hilversum, who at that time sold these sets if a customer wanted everything to be included in an order) was offered and accepted.
KBX 100 was as TBX a computer controlled system but used electronic contacts in an “old-fashioned” selector network. This made the system cheaper than TBX. It was very compact, with the whole system installed in a wall-mounted cabinet.
This delivery marked the delivery of more ship’s exchanges.
Assistance to the sales activity in Hilversum
In the summer of 1984 I was with some of the technicians from CTBD on a course in Hilversum about the further development of TBX. It grew not only as to hardware, new program packages every year gave still new possibilities. We had started with package 610 in 1983and its program and data memory took up the space on three half-size circuit cards. In 1984 package 615 came with more possibilities, i.a. calling number display for extension to extension calls using keytone signals, and with program and data memory on one half-size card. In 1985 package 620 and finally in 1986 package 630 came. The reduction in space required for the memory made space for ancillary computers, e.g. for a card receiving data from the exchange on the outgoing calls (calling extension, dialled number, time and duration of call). The ancillary computer could then make up a bill for the single extension. CTBD sold this along with TBX exchanges to several hotels.
The facility “calling number display” required an extension set with a keytone receiver and a display. Hilversum could deliver such a set, but we had to disregard it, as its use in Denmark would first require that it was approved as a telephone set, and it deviated too much from the Danish specifications for that. The market for an adapted set would also be too small. We needed a unit with receiver and display, which could be connected to the extension line in parallel to the normal Danish apparatus. Hilversum did not have such a unit.
The Danish factory TFA Automatic had a set with a display (called danMark2). The display was used to show the called number, there was no keytone receiver in the set. We tried to make TFA Automatic interested in making a set with such a receiver build in, but did not succeed. With TBX as the only exchange able to have calling number display (and that only for internal calls) they also considered the market to be much too small.
Shortly after this I saw an advertisement from the firm Hasselriis about a DTMF (Dual Tone Multi Frequency, keytone) tester, designed to test the signals sent from a telephone set. It was exactly what we needed, if only the display could be turned on and off according to the signal protocol used by Hilversum. We succeeded in getting a description of the protocol from Hilversum (some people there would rather keep it to themselves so customers had to use their set) and I sent this description to CTBD and to Hasselriis. Then I withdrew. We should not become an expensive go-between and the more CTBD was involved, the more they would sell the facility to their customers.
It never became a large selling object. Each unit was rather expensive and it could only show calling extension numbers. The public exchanges could not transfer a calling number to the called subscriber or PABX, this only came with the ISDN and was later introduced also on analogue lines much as in TBX in 1984.
One day during the course we made a walk on the heather besides the office building during our lunch break. Aad de Winter from the technical-commercial section, who took care of the Danish guests, told me about a problem he had to confront the next day: They would receive a visit from Kenya and how did you convince them that it was a reliable system when it had only one main computer?
As a former British colony the telecom administration was naturally under a strong influence from British Telecom, and they had specified that if a PABX could have more than 120 extensions, it should have a duplicated main processor to protect against operational failures.
I could only do one thing and advised Aad to discuss it with one of the pupils from CTBD, Niels Hald, and draw on his experience with TBX. He was a very capable technician, knowing our systems well, as he had started with EBX 8000. He later worked with SOPHO-S and then with LME’s MD110, before he left CTBD and was employed by Philips when PABX deliveries were liberalised. Aad did as proposed and got a lot of ammunition for the discussions the next day. The operational reliability of TBX was on top, even if there is only one main processor. If only the processor has a sufficiently long MTBF (Mean Time Between Failures) and the operational experience confirms this MTBF, one might even reduce the reliability if it is duplicated. The circuits surveying the operation and choosing which processor should carry the load might also fail!
I never heard if Aad succeeded in convincing the Kenyans about the blessings of TBX, but he was glad for our help.
My 25 year anniversary at Philips
On February 1st 1984 I had been 25 years in Philips and it was marked with a reception and a lunch in accordance with the rules for such an occasion.
It was nice to see all the people I had had relations to during the reception. And it was nice to see those I especially had worked with at the lunch. Invitations had gone to all those, who had been or were in the telecommunication section during the 25 years, with Max Hansen at the head.
In my speech I reminisced over how easy the first 15 years had been before we received the first real order for an exchange. When I thought about the lot of work this had carried with it, I could almost regret my efforts!
It was received as intended: nobody took it in a serious way. They knew that also at this point I agreed with Peter Townsend, who made Avis big in the course of a few years: “If you are not in business for fun or for money, what the hell are you doing there?”
“You have got new glasses”
I was born short-sighted at the right eye. This is a family weakness, my father had it the same way. During many years I found it very practical to have a natural division of work between my eyes, reading with the right, looking into the distance with the left! I had to accept being a left-shooter in the Home Guard and in the Navy, which was more difficult while guns should still be charged manually between shots.
But in the 1980es my age was felt. My good left eye got long-sighted and I did not see sharply at middle distances, e.g. the TV at home. I had to give in and get glasses.
When I fetched them I got the good advice to only take them in use when I had come home. Being used to drive a car without them it would be dangerous to use them immediately. And then the adaptation began, from only seeing a foggy picture with the right eye (which was suppressed by the brain) I now had to combine two sharp pictures which did not in the first instance overlap.
I looked forward to the reaction among my colleagues at Philips. I drove to work as I was used to, without glasses, and then put them on. As I was at my desk I took them off again (and I still do not use them while reading in 2009). And put them on again when I went for lunch.
On my way out our Publication Officer, Henning Bjerno, stopped me. We had worked together for years in the committee for the personnel newsletter “Philiskopet”. “You are different”, he said and looked me in my eyes, “oh, now I see it, you have got new glasses!”
“No, I have got glasses” I answered. I could hardly have received a better proof that they fitted with my looks!
ISSLS 84 in Nice
Through my membership in IEEE I received many invitations to international conferences. In 1984 an invitation came to the “International Symposium on Subscriber Loops and Services” which was to be held in Nice in Southern France. I asked for permission to participate and got a green light.
It took place in “Acropolis”, an enormous conference centre in downtown Nice. There was much to learn, especially about ISDN, which had now passed much of the standardisation and looked like a growth area for the next years. SOPHO-S was already adapted to the basic rate for user connections of ISDN, with 2 channels for user data of 64 kbit/s each and a signalling channel of 16 kbit/s, independent of the user channels.
But there was also time in the evening to visit the old town with its almost Italian atmosphere (Nice was Italian until about 100 years earlier and there was a large statue of Garibaldi in the town). I postponed my return trip to Saturday evening to visit the opera (Maria Stuart) Friday night and rent a car to drive along the Cote d’Azur on Saturday. First I went eastwards to Monaco and Menton, then westwards past Nice on the Promenade d’Anglais to Cannes and Grasse. There was a lot to look at but too little time to enjoy it this time around. But I would like to return and 6 and 7 years later I got the chance to live down there during some months’ work at ETSI, the European Telecommunications Standards Institute.
Chairman of the personnel society
In the spring of 1984 I was elected chairman of the personnel society after several years as a member of its board and as vice chairman under chairs like Alfredo Pedini and Willy Elkjær. Now it was my turn.
In November we had the traditional Banko with good prices. Every year Philips gave a number of their products as main prices and they were much appreciated. In addition there were vouchers for a butcher, wine and chocolate. And there was, as tradition also required, prices to the neighbours of the winners.
My wife had – as usually – told me that if I did not return with a price, I would not be allowed to go to the Banko next year. So, careful as I am, I went to the personnel shop during the lunch break and bought a flashlight. Then I had at least the ticket for next year!
The Banko went its way as usual. First we had a good dinner and then we played while we had coffee. Some had as always drunk too much during the dinner to be able to follow up on the numbers called out. This made as always the evening lively. “Shake the bag” and other calls.
I had got no prices and was glad that I had bought the flashlight. But then we came to the last play and the main price of them all, a portable TV (14 inches and black-and-white, that was what was possible then) for all numbers on the plate. The neighbours would each get a steam iron.
And then I was the first to get a full plate. I had hardly said “Banko” before there were cries from all over the room: “Corruption”, “Loaded dice”, “Arranged play” etc. But the plate was OK and I got my TV. Our secretary, Betty Bjerre, sat next to me and was overjoyed by getting a steam iron. She had just said to me that she had hoped for it, as she should iron the dress of her daughter the next day for her last day at school.
When I came home I hid the TV and only placed the flashlight so it could be seen. But there was great joy when I revealed the price. It was at once seized by our daughter of 16 and put up in her room
Coffee in equipment for the blind
While there was a lot of activity around the new types of exchanges the EBXes still existed and needed an expansion from time to time. A wish from the state administration was that their PABXes should have the ability to be handled by blind operators. This was a possibility in EBX 8000 with a special panel under the normal operator’s desk.
Its principle was that the cables from exchange to desk went through this panel. All commands from the exchange were detected by the panel, which had a number of small pins, driven by coils so they were either level with the surface or protruded slightly. For individual lamps on the desk there were individual pins, for the digit display of the desk there were pins in the same pattern as for digits in the Braille alphabet, so the operator could read them. To the left there were four pins telling if there were raised pins in one of the four sections of the panel, corresponding to the groups in which calls to the operator were divided. A metal plate could cover the whole panel (like the lock over the keys of a piano) and the desk could be used by a seeing operator.
In total we only delivered one of these panels to CTBD. Only a few were at all produced and they were consequently quite expensive to make. It could be discussed if Philips should sell hem at a low price anyhow. My view is that such a social action ought to rely on the public authorities, it is i.a. therefore we pay such high taxes!
One day there was an alarm from CTBD: The panel for the blind did not work. They were open about the reason, a cup of coffee with sugar had been tipped over it. Just such an event is one of the most dangerous things for the contacts of electronic equipment and a good reason not to have any horizontal surfaces on an operator’s desk, on which a cup could be placed.
We sent the panel to Hilversum with a fault report clearly stating why the fault had occurred. It was quite obvious that this fault could not fall under any kind of warranty, and if it could not be repaired, Hilversum would be right in taking the price for a new desk for the replacement they would send.
But they did not. We got a new panel free of charge from Hilversum and sent it on to CTBD at no cost.
Summer 1985: “You are a salesman”
A total of four cruising ships for Carnival Cruise Lines, Miami, were built during these years, 2 at the shipyard in Aalborg, Denmark, 2 at Kockums in Malmö, Sweden. Philips supplied the TV-installation and public address systems for all four and via this path we in the telecommunication group were also asked for offers of telephone exchanges for the last 3 ships. We got the orders for all 3 and delivered TBX exchanges.
These six TBX exchanges were as far as I know the only Philips PABXes delivered to a customer in the USA during the 1980es. Admitted, via the shipyards, but still…
The whole projecting of the exchanges was done by me. It turned out to be an advantage for our cooperation with CTBD that we in this way got experience with having a direct relation to a customer.
The installation was done by the electricians of the shipyards based on the manuals from Hilversum (and the translations into Danish). It tells much about the quality of these manuals that the installation could be done in this way. The main distribution frame (where cables from the exchange meets cables to every corner of the ship) was wired up according to lists I had prepared based on the wishes of the shipyards regarding extension numbers and cable plans. Kockum hired assistance from the Swedish Televerket (today Telia) for this wiring up and the installation of telephone sets in the cabins etc.
It was by the way against the policy of Philips that we in this way went across the border to the area of another national organisation, but we notified Swedish Philips, who at that time were unable to take on such an order. One of our arguments for handling the case from Copenhagen was that we could see the customer from our office windows (16 km away across the Sund) – that was impossible from Stockholm! They would have liked a percent of the sales price, but neither the TV-section nor we paid them. They had done nothing at all in connection with these orders.
When the orders were signed the consultant of Carnival Cruises, an Englishman, said that they felt a little bit cheated by TBX. The first ship had got an electromechanical exchange from LME and the equipment including main distribution frame took up the same space as three cabins. TBX with all items took up the space of one. They had received much less equipment so they found that the price should also have been much smaller. I answered that they should realise that they had now two cabins more, which could be sold and earn money at each cruise. In fact TBX should have cost them more than the LME exchange. He just looked at me and said “Oh, you are a salesman!”
I would, however, not have managed the situation so fast if I had not had a warning from Bent Hasselflug from our TV-section that this question might pop up. I had had time to think of a good answer. And it confirmed how important it is to consider all aspects before a meeting and have an answer for whatever might turn up.
Cable Lists and Installation of the Ship Exchanges
I shall go more in detail about our assistance to the shipyards. It was not enough to supply the exchanges in their transport boxes and escorted by the documentation as we were used to when supplying exchanges to CTBD. We should not install the exchanges ourselves, but we had to deliver all preparations, i.e. exact plans for the connections in the main distribution frame and which data to load into the exchanges.
We did not supply the telephone sets. It was a condition that their keys were marked with letters corresponding to the different decks. All passenger cabins at deck 5 had numbers starting with 5, but their cabin designations started with M (for main deck) and so should their extension numbers (which should be identical to the cabin number) and the letter should be on the sets. It was the same for the other passenger decks. I rejected any thought of having Hilversum deliver sets with such a non-standardised marking and did not even ask them. As far as I remember the telephones for the ship built in Aalborg were supplied by Standard Electric Kirk in Horsens and the sets for the two ships built in Malmö were supplied by LME. The work on the main distribution frames in Malmö was done by a technician from Televerket. He was quite impressed by TBX and said one day that he found it a pity that Philips had not supplied the telephones. Yes, for when the passengers were satisfied with the operation of the telephone system they would only look at the sets and give LME the honour!
The preparations should make it possible for the electricians of the shipyards (who were clever, otherwise they would not have been employed for long at a shipyard) to do the installation. The basis was of course the documentation provided by Hilversum and the Danish descriptions were a good supplement. But this documentation only comprised what was the same in all exchanges, not the individual adaptation.
The shipyards sent as a basis for these preparations detailed cable plans for the ships, i.e. how the extension lines went from each cabin or other telephone location to the main distribution frame. It was a list of to which cable and to which pair of the cable each extension number in the exchange should be connected. These plans I used to list where each cable (and cable pair) should be connected to the cable side of the frame. I also made lists of how the cables from the exchange should be connected to the exchange side of the frame and finally how the two sides should be connected. It was a considerable administrative workload and required great attention to avoid errors. Each ship had about 1000 extensions at the passenger and about 300 extensions in the crew exchange.
Data for the exchanges should also be prepared. The numbers were of course laid down in the wishes concerning extension numbers in the exchanges, but there were many more data to be laid down. I made first a questionnaire with the data I expected them to want and sent it to the shipyard for confirmation. After a suitable time I asked if there were comments? There never were, I think everybody was so busy with other things that no one had time to look at my proposals. Then I could start preparing the project and administrative data and make the tape for loading them into the exchange.
I had for each ship included an MPU, the suitcase with processor and cassette station, in both offer and order, so the ship would be able to manage even a complete failure of an exchange. It would have been possible to prepare the tapes on the MPU alone, but then they would only contain the absolutely necessary data (no comments in words as the MPU only had a numerical keyboard) and that would be rather useless except for the loading itself. For humans to use a listing there should be comments for each line. Luckily MPU had a socket for connection to a terminal or a minicomputer (this was before PCs became common), and from here one could write both digits and letters and get the whole content of the tape printed on paper.
This was again a considerable administrative job to make the necessary four tapes for each ship and get it all to fit to the lists for the main distribution frames without errors. But we succeeded, the exchanges worked as intended when my tapes had been loaded and each cabin had got the right extension number. The latter point tells of course as much about the care with which the cables had been laid and connected in distribution boxes around the ship and in the frame.
Spring 1986: Expensive strap in TBX
The passenger exchange at the cruising ships was completely filled with the maximum eight shelves possible in TBX. I took care of the first two ships myself, but when the second ship from Kockums should be loaded we had in Tele got a technician, Peter Miedema, who after a course in Hilversum could do the work on board to start the operation.
But the exchange would not start up. It was impossible to get the left cabinet with PPU 2 and 3 (the two PPUs for the four shelves of this cabinet) to work. He worked hard to find out why. I was called in but I also failed to find the fault. We called for help from Hilversum and a clever technician came. It did also not help for a couple of days.
The third day I was also in Malmö. It was still impossible to get the exchange in operation. One of us came on the idea that maybe two of the PPUs reacted in the same way on orders? They had each its number and should only react on orders to its own number. The circuit cards were alike and the number was decided by a plug with straps on the front. I had earlier examined the principle, which was that PPU 0 had two straps, PPU 1 had one of them, PPU 2 had the other and PPU 3 had none. We looked at the plugs which were all marked correctly. However, current does not care about what the markings are, so we looked through a small hole in the plug at the straps themselves. They seemed OK in the plugs for PPU 0, 1 and 2. But there was also a strap in the plug marked PPU 3! Off with this plug (the function would be the same with or without the missing strap), another start of the exchange and now it worked perfectly!
Hilversum always complained over the costs of administration of each type number and therefore I never understood why there absolutely should be a plug without straps in PPU 3. Well it made all PPUs look alike and that is maybe pretty. But it cost the world in this case that the factory had marked a plug for PPU 2 wrongly.
It is by the way for me the most beautiful type of trouble shooting: You remove something – and then it works!
Inmarsat and a trip to Miami
The two ships from Kockums should have an extra facility for their telephones, the possibility to talk with telephones on shore from all passenger cabins. It was specified that the telephone exchange should have a connection to the satellite radio, working via Inmarsat.
In our offer we had of course described the connection. It was like a connection to the public network from a PABX on shore, you dialled a “0” and got an exchange line circuit which – towards the public exchange, or in this case towards the satellite radio – worked like a telephone set. That is, our circuit should receive DC supply and ringing voltage from the satellite radio.
One of the last days at the shipyard it was so far that this connection should be tested. It did not work. It turned out that the satellite radio also worked like a telephone set. Thus, in both ends of the connection there was a circuit requiring voltages from the other end and that did of course not work at all.
With the possibility of a satellite connection there was also bought a metering processor for the ship. It should for each call via the exchange line circuit note the calling and the called number, the time of day and the duration. This was to be used in billing the passengers. This equipment could only work together with an exchange line circuit so we could not change the interface in TBX. For the satellite radio there were no other 2-wire interfaces mentioned in the brochures. Further it came from Japan, so it would take time to introduce any kind of solution in this equipment. Something had to be inserted between the two interfaces which could supply the voltages to both sides.
I maintained that we had described our interface clearly in our offer and both shipyard and ship-owner agreed that it was not our fault. But this should of course not prevent us from assisting in solving the problem. I promised to try and find such a box which could connect TBX to the satellite radio.
At first I thought of a separate box, but when I sat at home that evening it struck me that TBX itself had all the necessary circuitry in place for a solution. It had extension circuits towards the telephone sets and two of these could be connected to the two interfaces, to the exchange line circuit in TBX and to the satellite radio, which both worked like telephone sets. It had the facility “hot line”, meaning that if an extension goes off-hook (if you lift the handset) the exchange would automatically call a certain number, in this case the other extension circuit, and when it answered (went off-hook) there was connection. And finally TBX had the possibility to have user groups which were barred from having mutual connections. This was important for the billing, for if you could just get a connection to the satellite radio by dialling its extension number, you would circumvent the whole billing device. Thus it was required that the two extension line circuits which were connected to the interfaces were in a separate user group with no connection possibility to other extensions in the exchange.
The next day I went through all the papers about TBX and determined what was needed in all details. It was indeed possible to do as I had imagined. But it would be necessary to load the exchange from scratch with new project data (as it was necessary in case of any change of project data) to introduce the separate user groups. The opening of the two new extension numbers and the making of “hot line” between them could take place during normal operation. I would have to take care of the loading, especially as the function should also be tested. I had to get on board again to introduce the new facilities.
Thus I called the shipyard to propose the solution. They quite agreed to it but the ship had just left for New York and Miami. We agreed that I should go to Miami and do the loading when the ship was for the first time in its home port. The shipyard paid for the trip.
I had all data about the project in my office so I could prepare a new project tape and bring it with me to Miami. And could plan it such that it also became a vacation trip!
I left Copenhagen a Saturday morning and about 2200 the same evening I was in my hotel in Miami. I felt it like 0400 Sunday morning, but a good sleep helped. Next day I went to Everglades, “a river 80 km wide and 25 cm deep” as it is said. I enjoyed the large national park with alligators, racoons, pelicans and many other animals. When I returned to Miami I saw that the ship had arrived.
Next morning I went on board and loaded my tape. It worked fine so the solution was found. Except for one thing and I do not know how that was solved afterwards: The connection in TBX was released when the calling or called extension went on-hook. Then the exchange line circuit went on-hook towards the two special extension line circuits and this caused a busy tone to be sent to the satellite radio. This equipment, however, did not go on-hook. It would go on-hook if the keytone signal for * was sent to it but it was impossible to get TBX to send this tone signal immediately before every release. From the other side, i.e. from shore via the satellite radio, there was no release, only a time release after several minutes. During this period one had still to pay for the connection. I could not solve this problem.
On Tuesday I had again a day off, going along the North side of Everglades to an Indian village and out sailing in the park on one of the special boats with a flat bottom and driven by a large air propeller. On the ship they knew where I lived and I had asked them to phone me if there were more questions. There were, so Wednesday I was on board again and settled the last questions. Then a trip to Fort Lauderdale, going by boat to an alligator farm and see the show. Finally back to Miami airport and back to Europe, to Hilversum where there was a meeting about metering equipment the next days. And then finally back to Copenhagen.
Later Peter Miedema was also on a service trip to Miami. And after I left Philips in 1991 J. H. Bojsen was also on a trip over there. As far as I know it was because a new satellite radio had been installed. Its interface worked as an extension line circuit, i.e. it could deliver the voltages to the exchange line circuit so it was not more necessary to go via the two hot-line extension circuits. This also solved the release problem.
Equipment for the blind in TBX
It was a natural requirement that also the TBX exchanges should accommodate blind operators, but a panel with pins as in EBX was considered much too expensive. It might cost as much to make as the rest of the exchange (due to the small number required), so this was out of the question.
In stead a system with stored speech was designed. Every command from exchange to operator’s desk went via a device, which based on the command (which lamp to light or digit to show) sent a spoken message to the telephone of the operator’s headset, injected on the speech connection from exchange to headset. The speech was stored as about 50 single words (e.g. the individual digits) and a command would call up these words in the right sequence.
The spoken words were loaded into the device from a tape and that tape had to be recorded according to strict rules in an echo-free study. The words had to be said in a certain sequence with a certain pause in between. This made it easy to load the device with speech in different languages. As to Danish, my wife spoke the words for the recording during a trip to Hilversum in the middle of the 1980es.
The Danish Society for the Blind arranged a meeting for many operators in a hotel, where we could present the system and LME could present a system with pins for their new PABX, the BCS. A nice opportunity to learn about a competing product! It was a pity that Hilversum had not loaded the speech into the device with any kind of quality, so we could not act very convincingly about the advantages of our system. The operators clearly preferred a system with pins.
BCS was digital but not based on the usual PCM. Just as we did with TBX LME had found that PCM would be too expensive for a small PABX. They coded the speech according to another system and sent this signal through the exchange. Along the road, after TBX, CTBD choose to rather sell BCS than our successor to TBX, the SOPHO-S system.
Cordless telephone from TeKaDe
Other products appeared in these years. One was cordless telephones. It had already for a long time been possible to get some cheap types from the USA and the Far East, but they were banned from use in Denmark. For one part they used frequencies set aside for other services (which might be disturbed), for another there was no security against other people listening to the conversation. They sent in clear speech on a certain frequency. In addition every handset could work with any base station, so if you did not like to pay your calls to other countries yourself you just went down a street with your handset, switched it on and saw if there was an open base station near by. Then you called and the owner of the base station got the bill! This system was called CT0 for Cordless Telephone 0.
A system was developed, called CT1, where some drawbacks were eliminated. They used legally provided radio frequencies. They choose at random one out of forty radio channels when they were switched on (a channel which was not already in use in the neighbourhood), but they sent clear speech over the channel. Finally base station and handset exchanged handshake messages all the time during a conversation making it impossible to use base stations belonging to others (and they could not use yours).
In Germany they went a step further, to CT1+ with 80 radio channels. TeKaDe developed such a cordless telephone and as a baby a “normal” CT1 telephone. We received one of the latter for a type approval. It was already approved in Holland and TeKaDe sent with it the report from the Dutch testing laboratory as to its properties at the radio channels.
I took this report with me to the Danish testing laboratory established by the State Telecommunication Administration, for we would try to pay as little as possible for the test, preferably only for test of the properties as a telephone (these properties were adapted to the Danish requirements). At that time it was only this official laboratory which issued type approvals, only later the system with independent testing laboratories and approval by the Telecommunications Board was introduced.
The laboratory gave a good example of flexibility. The report was in Dutch and I offered of course to translate it. But they would not require this. G. H. Hansen of the laboratory said very precisely that the figures were the same and the technical terms were easy to translate. So testing of the radio side was unnecessary.
After the test we received the type approval. But the price was high so there was not much sale. In addition there was also the problem that it was a telecommunications product and we were only used to sell to the administrations. Philips Radio who sold radio sets and TV sets via the shops had been the right channel, but it was not a product from their factories and therefore they did not want to take it into their program! We had the same problem with the digit senders from Pye-TMC.
Only 10 years later when the production of cordless telephones had moved to France CT1 telephones became a success at the market. This was when they had crossed the magic price line of DKK 1000 and when the DECT system had come with its digital cordless telephones. In other European countries there was also a system called CT2 which obtained a certain success. It used radio frequencies set aside for other purposes in Denmark, so they were never in legal use in Denmark.
An addition to TBX was the call metering system. When the space for the CPU memory had been reduced to 1 out of 3 (half) places there was room for additional functions in the 2 places made free. As said earlier one of these functions was call metering. There was not space for much functionality on the card (which was only half as big as the other circuit cards in the exchange), but in spite of this a number of these cards were sold with exchanges in the area of CTBD, especially to hotels.
If more functions were required data about each call had to be sent to a minicomputer housing all the necessary programs for calculation of cost etc. There were two solutions, a small and a large one. The small one comprised a PC with the control system “Concurrent DOS”. A PC is (or was) a system for one user at a time, so normally it could not accept new signals at arbitrary moments from more than one user. This was, however, required here, as it was impossible to control when TBX was ready with data for a call and would get rid of them. Luckily the concurrent version of the old control system DOS could do this. While the PC was busy calculating costs it could jump to a new signal at its serial port where new data about a call arrived from TBX.
The large solution contained more cooperating processors. It was a minicomputer system from Philips Data Systems in Apeldoorn. We had sold this solution to the cruising ships. Data arrived at one of the processors and it notified the main processor that it had new data. When it suited the main processor it fetched the new data and began calculating on them. All processor cards were joined by an SCSI, Small Computer System Interface. This happened in the middle of the 1980es, but even 10 years later it sounded in many advertisements as if SCSI was a quite new invention!
We hoped also to sell this call metering system to CTBD, but had to know it better first. So I joined a course of 14 days, one week in Apeldoorn on the minicomputer in general, one week in Hilversum on this special application of it.
In Apeldoorn I met other pupils with other applications in mind. One was an Englishman and he was impressed by the way I could get along in Dutch. He would prove that he also managed the language and when we were out eating he would see to it that we got individual bills to present to each our home organisation. He went to ask for “twee bills”. A few moments later the servant came with two beers, “twee pils”. We laughed and got things right while we enjoyed the extra beers.
40 year jubilee of PAP
The Philips personnel society PAP, for which I was chairman at the time, had its 40 years anniversary in 1986 and had prepared a large celebration in the NIMB restaurant in Tivoli in the middle of September. Luckily we had a good celebration group handling all agreements while I only signed the bills, for it collided to some extent with the above course in Holland. I was away the last 14 days before the celebration, only coming home Friday evening as the celebration was on the next evening.
But I went to NIMB the next morning to ensure that everything was in order (it was) and in the evening we had the feast. With performances by several people from Philips, i.a. Øberg who sang “Svantes song” by Benny Andersen with the line “Here I stand looking towards the coast of Sweden”, a very appropriate line after the joining of the Nordic national organisations in Philips Nordic.
Thus it had worked well that I had delegated all the work to our celebration group, but it had perhaps been too much for some of its members. In any case I received Monday morning after the feast a note from our chief telephone operator, Grete Biilmann, the chairman of the group, that she resigned from the board of PAP. I went of course immediately over to her to say that I was very sorry to get the note. I did not try to get her to change her opinion, I considered that a dead end. But it was typical for her that she did not jump ship during the preparations but first when the task was completed.
1986: Large and small SOPHO-S
In the task force of which I was a member back in 1982 we had specified a type of SOPHO-S to replace EBX 8000, thus a large PABX. It would be uneconomical at capacities under about 500 extensions, but there was a change underway, soon the only saleable PABXes of any capacity would be the digital ones.
That means digital both in their control (as EBX had been since 1976) and in the switching of the speech connections. Already in 1980 Jutland Telephone had introduced the SL1 PABX from Northern Telecom, so far, however, with only analogue connections to extensions etc. and in 1982 the first installation of the MD 110 PABX from LME took place at one of the customers of CTBD. TBX was a temporary solution and had to be replaced soonest with a digital PABX.
TBX was a fine solution both in the 1980es and later (if it was on the market) for customers only interested in speech communication. But requirements came up to integrate speech and data, and then the digital solutions are the right ones because they treat all types of communication alike, as bit streams. In addition digital connections can in principle be better than analogue because there is no attenuation between sender and receiver. The stream of bits generated at the sender will arrive unchanged at the receiver. In addition TBX had shown for the first time how compact it was possible to make a PABX using time multiplex. The other PABXes mentioned above did also use time multiplex but were hardly more compact than EBX 8000.
As a consequence Philips now developed in parallel digital PABXes for large and small capacities, where the small one luckily could be made through reuse of common circuits and programs from TBX and of switches and interface circuits from the large SOPHO-S.
The large SOPHO-S, called HR for High Range, was designed with three modules each placed in a shelf in the cabinet. It was the Peripheral Module, PM, the Switching Module, SM, and the Control Module, CM. While an exchange only had one CM and one SM, there could be up to twenty PMs in it.
PM housed all circuits for up to 128 external connections (to extensions or other exchanges), a peripheral processor to serve these circuits (it could in fact serve circuits in 2 PMs), and a digital switch to connect between these 128 circuits mutually, between them and 60 connections to the rest of the exchange, or to a tone generator in the PM for dial tone, busy tone etc. This switch was unblocking, i.e. whatever other connections existed, it was always possible to make a new one between two free circuits. The conversion of analogue signals (from normal telephones) took place in the interface circuit itself, so the rest of the PM handled only digital signals.
SM housed switches and was connected to all PMs in the exchange and to SMs in other SOPHO-S HR exchanges and to CM. The SM connections made it possible to have a number of exchanges work as one big exchange with up to 20.000 extensions, though each exchange could only accommodate 2000. When an exchange was started up a 64 kbit/s data channel was made via SM between CM and each PM in the exchange and between CM and each of the other CMs in the SOPHO-S network. All connections to SM were standard 2 Mbit/s transmission systems having 32 channels of 64 kbit/s each In each system to PMs or to other SMs one channel was used for the control from the CM, the other channels were for speech or data connections between PMs. In the system CM to SM all channels were of course used for control. Thanks to the use of standard transmission systems and that all channels to a PM came from the SM, a PM could be placed as a remote unit several kilometres from CM and SM. This saved cable capacity, in stead of 128 extension lines there was now 2 2Mbit/s systems, corresponding to 4 extension lines. Neither PM nor SM were duplicated, they were inherently reliable enough. However, the 2 2Mbit/s systems from a PM (each with a control channel) were connected to 2 switch cards in the SM and each switch card was an independent unit. A fault would only reduce the traffic capacity as the PM to CM connection could use the other transmission system and switch card. The power supply to the SM, which was common for all switch cards, was triplicated and so was the common clock generator, which was also placed in the SM.
The CM housed processors for the overall control of the exchange. It took care of all events not requiring a real time reaction. Analysis of dialled digits, allocation of resources and control of facilities were typical jobs for the CM. Its connection to the SM was two or more 2 Mbit/s systems, and there was always two channels to each PM or other CM, led through different systems. The processor with its power supply was quadruplicated with a common memory for all four processor slices (as they were called). The distributed memory contained 16 bit data coded into 32 bit with 8 bit per slice. If there was a fault in a slice the other processors could recreate the data from the remaining 24 bits, even if there was an additional bit error in these 24 bits. This 4/2 system, as it was called, was chosen due to the combination of a trend towards cheaper processors and towards more and more need for memory space. Thus one could use more processors without much consequence for the price, but the total memory space should be minimised.
SOPHO-S LR, for Low Range, which was based on TBX, had four members, SOPHO-S 50, 100, 250 and 1000. The latter one had by and large the same set-up as TBX 1000. From TBX it was kept that the CM (the CPM card and its memory card) was connected to the PMs (the PPM in every second shelf) outside the switch card. The switch card was only found in SOPHO-S 1000; in the smaller types the PMs were directly connected to each other. In the control part the same hardware was used as in TBX, only the programs for connection switching were different due to the PCM switching. This went so far that when you looked into an empty shelf you could clearly see TBX sockets to the left where the control had its place, and SOPHO-S sockets to the right. All external circuit cards were the same as in the HR exchange. This meant that maybe 80% of the cards used in an exchange could also be used in the other types of SOPHO-S exchanges, an important point for the stock of spare parts! This did not mean that 80% of all card types were common, that number was far smaller, for it was the cards, which were used in large numbers in every exchange, which could be used in all models.
A consequence of the reuse of TBX programs was that the low range had in the beginning more facilities than the high range!
SOPHO-S 50, 100 and 250 were an attempt to make a small, cheap digital PABX. In these the CPM and PPM were on the same card, they were even in SOPHO-S 50 and 100 combined in one processor just like in TBX, and this saved the conversion to and from serial transmission between them. In addition there was no need in SOPHO-S 100 for connections to another PM, so the two switch outlets reserved for this could be used for another two groups of telephony circuits. The same was true for SOPHO-S 50, but here there was no physical space for the extra circuits. In SOPHO-S 100 there was space for two cards in each of these groups, but it was only possible to have digital interface circuits in them.
Prices for TBX and SOPHO-S
When we were able to market SOPHO-S LR, CTBD would change as fast as possible to this system from TBX. And they required that the price should be unchanged for the same capacity. Hilversum wanted to sell an exchange for about 30% more than a corresponding TBX exchange.
Luckily I had been lazy during the years past. In step with rationalisations in the fabrication of TBX, Hilversum had lowered the export price of it a few percent each year. The price to CTBD was already one of the lowest ones for a PABX of this capacity, so we had not passed on these reductions to CTBD. They would not have influenced the sales figures. This was my opinion and I did not hide that it was also because I would avoid making new price lists to CTBD. As the Dutch say: Do not awake sleeping dogs! Øberg had agreed to this.
When CTBD now came with the demand to have unchanged prices, Øberg was happy that our TBX prices were on the high side (seen from Hilversum). About 20 of the 30% were in this way already taken care of and that made it much easier to give in (after many long discussions) to this demand.
Presentation of SOPHO-S LR in Hilversum
November 1986 it was so far: A bus filled with salespeople from CTBD should meet SOPHO-S and the people making and selling it on a trip to Hilversum. This should mark the beginning of the sale of the system in Denmark as a replacement for TBX.
A few days before I was in Øberg’s office as his telephone rang. It was Nielskov, the head of CTBD, and his message was that he did not want me to participate in the trip to Hilversum.
Øberg (and I) could only do as wanted, so I did not go. The reason for this very unusual wish was perhaps that Nielskov feared that I would ask “on behalf of CTBD” questions about items which he did not want his salespeople to know. It is a question about how to sell. My opinion: Tell also about problems and argue for why they are not that bad after all. The other approach: If the customer does not ask then do not tell him.
The trip went well, I heard afterwards. Off Friday morning with arrival in Hilversum the same night, a tight program Saturday and home again Sunday. Thus only one working day was lost (if that is the word).
No customer names
There was a typical change when we changed to deliveries of SOPHO-S LR in stead of TBX: CTBD did not any more indicate the names of the customers when they ordered exchanges. The reason for this change was the upcoming liberalisation of PABX deliveries. CTBD foresaw that their suppliers might choose to sell exchanges (or their service) directly to customers. Thus it was essential that the suppliers knew as little as possible about where the exchanges were used.
CTBD could go two ways. Either they could guarantee a certain sales volume and buy the corresponding number of exchanges whether they were sold or not. In this case the contract with the supplier might have a clause that sales would only take place via CTBD, but they had a risk of ending up with a stock of unsold exchanges. Otherwise they could indicate how much they expected to sell and order exchanges as orders from the users came in. Here there was no guarantee that the expected sale became a reality and a supplier who had based his factory program on the sales budget would require that he could also sell directly to the users. Philips was always in the latter group so the question about making only indirect sales in a liberalised market did never come up. Thus our access to the user file of CTBD should be cut off.
We had to accept this, but the system was not watertight and we had our fun over that. There was a problem with the adaptation to the public telephone network and all the circuits for these connections had to be modified. This took place in a Philips workshop and CTBD brought us the circuits and fetched them again. In order to follow up on the modifications all circuits were labelled with the directory number of the relevant customer. Due to this it was easy for us to find the user names!
Lower sales of PABXes
Philips was not the only supplier of digital PABXes with less than about 500 extensions. Alcatel Kirk delivered their OCS 300. This meant that the sale of SOPHO-S LR was significantly lower than the sales of TBX had been. Philips had to wait for the liberalisation in the summer of 1990 before they could try to market directly to the users or sell via other channels than CTBD.
Above 500 extensions CTBD had introduced the MD 110 from LME and would not start selling SOPHO-S HR. However, a few installations were made via CTBD after a good deal of arm-twisting and Philips themselves installed a network of three exchanges, one in the office building, one in the factory and one in our central stock in Glostrup. We had only operators in the office building and all three worked as one big exchange, i.a. with free numbering throughout the system. We did not after all exploit this to the full extent, in stead all extensions in the office building had directory numbers starting with 2, in the factory with 3 and in the stock with 4, so the number gave a hint about the location. I had had extension 333 since I came to Philips in 1959 and Max Andreassen had seen to it that this number had been left untouched during the 2½ years I had been in Holland. Now I got 2333 and some people wondered why I had not given the offices 3 as first digit, to give myself the number 3333! We got also 2 Mbit/s PCM links to the public exchange and got in this way finally direct dialling in with all extensions called from the public network with 8 digits just like normal subscriber telephones.
Arm-twisting: SOPHO-S HR came in where MD 110 either could not be used or where the customer directly forced CTBD to offer SOPHO-S. An example of the first was the court house in Copenhagen, a protected building where data connections should be extended to PCs in all offices, but where the laying of new cables was banned. LME had from the beginning (in 1982) promised that data units could be added to their system sets such that speech and data could be sent simultaneously over the extension line. The maximum speed was limited to16 kbit/s, but worse was that nobody had seen these data units! Here SOPHO-S was born with ISDN on the extension lines, 2 user channels of each 64 kbit/s and a signalling channel of 16 kbit/s. The two user channels could be used simultaneously and there were from the beginning units to use one or both for data traffic at the full capacity. In this case SOPHO-S won because the offices could get data connections with a sufficient speed without laying new cables.
The other example requires first a few words about the relation to the consultant Bodil Larsen. Bodil had been employed at CT but had chosen at an early moment to leave the company and become an independent consultant on PABXes. For the “monopolists” in CTBD she was as a red cloth, they rather wanted her hanged! I had from the start said to my contacts at CTBD that it was rather a note of distrust that customers were willing to pay her for services they could get free of charge from CTBD, and that they should treat her well, as she was the chosen representative of the customer. Never mind how much one considered her a plague, she should be treated as a customer! Bodil ran her firm to well into the 1990es, but got seriously ill and died in 1996 after having wound down the activity.
But we were still in the 1980es. I had already during the time of TBX lived up to what I preached for CTBD and treated her well, sometimes so well that CTBD should not know of it! Øberg was also after me some times when I had talked directly with a customer. One thing was when a customer had learnt that it was a PABX from Philips they had been offered and then phoned us. They should be treated well, as they were also customers at Philips in many other respects. But it was something quite different when Bodil had phoned and asked me to talk with one of hers and CTBD’s customers.
In the late 1980es the Technical University of Denmark should have a new PABX and had hired Bodil as a consultant. CTBD had made an offer of an MD 110, but Bodil would not accept that. She knew SOPHO-S and required that CTBD came up with an alternative offer of this system. The customer backed her and an offer was made. The customer’s choice was SOPHO-S.
PCM and the selectors in SOPHO-S
I have in the above told about analogue and digital signals and the conversion between them and about bit streams of 64 kbit/s and 2 Mbit/s. This calls for an explanation.
Sound is small variations in the air pressure with a frequency corresponding to the pitch. For speech transmission it is sufficient to transfer the frequencies between 300 and 3400 Hz (oscillations per second). The microphone transforms the oscillations in the air pressure to oscillations in an electrical voltage. The voltage is transmitted to the telephone which again translates the oscillations to oscillations in the air pressure. This is analogue transmission. During transmission the voltage can be attenuated by the resistance of the line and distorted by the influence of foreign voltages on neighbouring lines and different transmission properties for different frequencies. In order to ensure a good quality of the speech there are strict rules for the amount admissible of attenuation, frequency distortion and noise, with due regard to the economy. The quality shall not be better than “good enough”, i.e. the customers must accept the quality and be willing to pay for the connections.
The transmission does not have to be continuous. If a sufficient number of instantaneous values of the voltage are transmitted the receiver can recreate the original signal. A sufficient number means at least twice as many per second as given by the highest frequency (in Hz) which shall be transmitted. For speech 8000 instantaneous values per second has been chosen. If an instantaneous value could be transmitted in e.g. 4 microseconds (µs or millionth of a second) the 8000 values would occupy the transmission line for 32 millisecond (ms or thousands of a second) each second. Between the values for one connection there is room for 30 values for other connections. This is time multiplex.
It is not necessary to transmit each value with full accuracy. Remember that the analogue signal will always be distorted. If one makes a small distortion when coding the analogue voltage into a signal, which can be transmitted with great fidelity, the receiver shall only cope with the small deliberate distortion. This is done in PCM transmission by coding each value to the nearest of 256 fixed values and transmit this value. The value is coded into 8 bit, which can each be either 0 or 1. 8 bit can form 256 patterns, from 00000000 over 00000001, 00000010 etc. to 11111111. This signal is very reliable as the receiver shall for each bit only perceive whether a 0 or a 1 was received.. If it perceives correctly the original signal can be recreated exactly as it was sent. 8 bit 8000 times per second gives 64 kbit/s and 30 of these connections (plus 2 more for framing (for counting to the right bits for a certain channel) and signalling) gives 2048 kbit/s or 2 Mbit/s, the primary multiplex in PCM transmission.
The selector in SOPHO-S was a chip having 8 pcs. 2 Mbit/s inputs and 8 pcs. 2 Mbit/s outputs. Thus there was 256 pcs. 64 kbit/s speech or data channels in and out. The selector chip could connect an arbitrary input channel to an arbitrary output channel and was non-blocking, i.e. if there was 255 connections in it, the last free input could always be connected to the last free output. For each input the chip had two memories with 32 addresses corresponding to the 32 channels and room for 8 bit under each address. One memory was loaded with the incoming bits and was thus overwritten every 125 µs or 8000 times per second. The second memory was loaded when a call was switched through, at an address according to the output channel, with 5 bit for the address in the first memory to be read and 3 bit for the output to which the content read out should be sent in that output channel time. If e.g. input 2, channel 23 should be connected to output 5, channel 12, then in the second memory for input 2 on the 12th address there was loaded 23 and 5. Thus, in timeslot 12 address 23 in the first memory of input 2 was read and its content sent to output 5 in that timeslot.
4 of the in- and outputs of the selector chip were in the PM used for the connections to peripheral circuits for speech or data in the same PM. Thus a PM could serve up to 128 peripheral circuits with 64 kbit/s to and from each of them. 2 of the in- and outputs were used for connections to SM or – in SOPHO-S 250 – for connections to the other PM. Thus there were 64 channels to the SM corresponding to half the maximum number of peripheral circuits so all circuits could not be used simultaneously, the exchange was not non-blocking if the PM was filled up. As to the 2 last in- and outputs, 2 output channels were via one out of 30 pcs. 3-party circuits connected to 1 input channel. In a 3-party conversation, as e.g. when an operator was connected to both an extension and a line to the public exchange, 3 of these circuits were used, each receiving speech from two of the parties and sending the combined speech to the third. This leaves one input with 30 channels, and it was connected to a tone generator for 30 combinations of tones. Thus it was possible to send any tone to any peripheral circuit. Each tone signal was defined in project data as x seconds of tone y followed by z seconds tone w followed by… One of the tones was “silence” and this was used to define pauses in the tone signal. In Denmark ringing tone was defined as 1 second 425 Hz, 4 seconds “silence”.
In the SM all 8 in- and outputs were connected to PMs or – in SOPHO-S HR – to CM
PM had originally up to 8 peripheral circuits on each circuit card and 4 cards had a common 2 Mbit/s link to the selector chip. All 32 channels could be used for speech or data, as framing and signalling went via other wires. A PM could house 4 of these groups of 4 cards or 128 peripheral circuits.
The simplest SM was found in SOPHO-S 1000 with up to 8 PMs. It used 2 selector chips, each connected to all PMs on both sides. If a chip should fail the exchange could work on via the other, but of course only at half the traffic capacity. SOPHO-S 50, 100 and 250 did not use an SM.
The system was developed to have far more than 8 PMs in SOPHO-S HR (or 2500). However, each selector chip can only send to 8 PMs. This was managed by letting all incoming systems to SM go to more chips, one for each 8 PMs. The CM will then control which of these chips shall send a certain 64 kbit/s channel to which output channel. There were various capacities of SM. The smallest had in- and outputs for a total of 16 2Mbit/s systems, i.e. for 7 PMs and 1 CM. It used 4 selector chips. The largest had in- and outputs for a total of 96 2 Mbit/s systems, for up to 40 PMs or connections to other SOPHO-S HR exchanges and to a CM with more 2 Mbit/s connections. A total number of 40 times 128 or 5120 peripheral circuits would, however, be a too great load on the CM, so the actual maximum number of PMs in a SOPHO-S HR exchange was rather 20 or 25 (the smaller number if the exchanges formed a network), with maybe an SM for up to 64 pcs. 2 Mbit/s systems.
1987, ISS 87 in Phoenix, Arizona
In the early spring of 1987 the International Switching Symposium was held in Phoenix, Arizona, and I was permitted to take part. My wife would not go but I combined it anyway with a vacation such that I could visit EPCOT (ExPerimental Community Of Tomorrow) in Orlando, Florida, and have a closer look at New York.
The trip to Phoenix was by plane via Los Angeles. On the plane I read in a Danish newspaper about “Les Miserables”, which had just had its opening in New York, and decided to go and see it. In Phoenix I stayed at a small hotel in the center (in walking distance from the symposium). I strolled in the town Sunday. It was raining and we learned next day that this was a rare occurrence in this town. It is a sprawling town, we learned also that when the guide book said that there was 5 minutes distance to e.g. the state government buildings it meant 5 minutes by car!
The symposium presented many new things. ISDN was being introduced in the US and several companies demonstrated their goods in hotels in the town.
It was three years after AT&T had been split in the long distance company and the regional telephone companies and competition had started in the long distance business. Arizona was within the area of US West and they had a motto inspired by the cowboys. It was very appropriate in the situation they would meet in a few years when also local telephony was opened to competition: “You either make dust or eat dust”.
There were also arrangements in the evenings. The different companies invited to them, e.g. Siemens to a rodeo in one of the suburbs. We saw all the shows belonging to a rodeo, riding wild horses and bulls, attack at the stage-coach etc. Afterwards there were steaks from enormous pans, each could take the meat from a whole cow! We also made a walk in the Wild West town neighbouring the rodeo.
In Phoenix there were arrangements in the city theatre. Henry Mancini directed, David Copperfield made his magic.
After the symposium there was a trip made by bus to the Grand Canyon. A hole so big that if Denmark was dug away to the sea level and put into the hole, it would not even be filled up! Going to it we were in a cinema to see a film on a very wide screen about it and it gave a first class impression of the canyon. It was almost an anticlimax to see it in reality! Next morning it was foggy and the bus soon returned to Phoenix.
Then I went by plane to Orlando and spent a day in EPCOT. I had read about it in IEEE Spectrum some years before and had intended to go there the year before when I was in Miami. But I gave up as there is 300 kilometres between the two cities. Now I could realise it. The different shows by nations and companies were impressive, such as the pavilions (a too small word for them) of Japan and China, the ball of AT&T with all about communications, the trip through the jungle at Exxon, the trip under the sea at United Technologies. I ate supper at the latter place with big windows to the pool, where sharks and divers swam while I enjoyed my “Red Snapper”.
The trip went on to New York for three nights, and each evening a musical. “La Cage aux Folles”, “Cats” and “Les Miserables”. During the days I strolled in the city, visiting both the Metropolitan Museum and the Guggenheim Museum. I also visited the AT&T exhibition on communications.
The final trip was by Jumbo back to Copenhagen. It was one of the last trips SAS had with that type of plane, they were changing to smaller planes with more flights.
Summer 1987: Philips Tele Nordic
Philips Nordic was started in 1985 or 1986 as a “national organisation” for the Nordic countries. In Eindhoven they obviously thought that one could make business in Finland from Denmark and vice versa. OK, there was some logic in operating with reasonably large organisations and therefore uniting the Nordic countries to one, which was, however, still smaller than the Philips organisations in Germany or Great Britain.
As the first head of Philips Nordic our head of Philips Copenhagen was chosen. He had the very appropriate name Bos (although this just means forest in Dutch). I was chairman of the personnel society at that time and we should give him a farewell present at the reception when he left us. I found the right gift: a globe where Denmark takes up one hemisphere with all other countries jammed together on the other! So I could give it to him asking him to place it in his new office and when he looked at it remember how the world really looked!
Within Philips Nordic each business area was collected under one Nordic manager. Most of them were Swedish, as the turn-over in Sweden was far the largest. Tele and data were joined in one group with Lars Nyberg in Stockholm as CEO. Under him there were a row of Swedes as head of various data sections. However, within Tele, the only area which had not only a turn-over but also a profit, there was not much activity in Sweden. We were active in Denmark selling PABXes for already several years. The PABX market in Sweden was being liberalised and Philips wanted to be a player with SOPHO-K from Pye-TMC in Scotland (they had moved from near Bristol to near Glasgow) and with SOPHO-S.
Thus, Philips Tele Nordic was started in Copenhagen with Niels Øberg as head. He and Erik Schacht should see to it that the sales- and service organisations in the other countries were established, I should support them on technical matters and our secretary, Betty Bjerre, followed us to the new organisation. Tele Denmark got a new head, Finn H. Nielsen, a former military man.
We succeeded in Sweden. Øberg and Schacht left Philips in June 1989 when Øberg had been head-hunted to become CEO of CTBD, and at that time the organisation was in place. Finland was never very interested, so there was no activity. In Norway an organisation was also established, but it was broken down in the end of 1990 as it became clear that it would take years before it could give a profit. On the Norwegian market there was a war on prices and it could not pay to continue only having costs. At least it could not with the short-sighted attitude prevailing in Philips from 1990, which also caused my leave in 1991. At that time also Betty left for a secretary position in the newly established European Radiofrequency Office in Copenhagen.
At that time my work for Sweden was by and large finished as we had the type approval for SOPHO-K in December 1988 and of SOPHO-S in December 1989 after an intensive participation in the preparation of the Swedish standards. I can only commend the Swedish authorities, Televerket (today Telia) for their open attitude to the standardisation work. My experience was that one could always get through with factual arguments for changes of the text in the standards. Naturally one had to keep to the factual arguments and realise when proposed requirements were mandatory although they were awkward for us. One had also to accept that the weaknesses of Philips could not be kept secret for the other suppliers, LME, Siemens, Nokia, Alcatel, in the standardisation group but this was a mutual problem although I felt the others to be more in a waiting position than we were at Philips.
An example: The higher authorities demanded that the market for “telephones with a switching possibility” should be opened up very fast. Telephone sets had already been liberalised during some years and now a company wanted to have their set approved just like a normal telephone, although it could change between more external lines. This was accepted and the necessary standard was quickly made as a basis for the approval.
SOPHO-K did not comply with the standard for PABXes due to some special signals on the line (caused by the Swedish net only disconnecting when both parties had laid on hook. Adaptation would take a long time and cost much, but these signals were required in a proper PABX like SOPHO-S. Thus we asked for approval of SOPHO-K (a line selector system) according to the standard for telephones with a switching possibility.
By the way: It would have been unprofessional to argue against these special signals in connection with SOPHO-S, so I did not do this. But I was anxious whether Hilversum would come back at this point, as it was also new signals for SOPHO-S. They never did, and I like to think that my status was such that when I said how it had to be it was treated as the Gospel itself. The real reason was probably that also the people in Hilversum could see the unavoidability of coping with the signals!
Then Televerket came in the autumn of 1988 with a new definition of a PABX, which made SOPHO-K into a PABX: When the telephone sets were supplied with current from a local source (and not from the line) it was a PABX. This resulted in a big discussion turning all aspects of the problem over. The result was that the definition of a PABX was maintained but the special signals could be disregarded for PABXes complying with certain clauses like supervision of each line to town with lamps and keys on the sets. This was OK for SOPHO-K, which was approved shortly after.
Another example: The proposal from Televerket for speech transmission to and from special sets under a PABX was that the requirements had to be complied with for all speech sets provided, handset, headset or loud-speaking set. The special set for SOPHO-S could not comply with the requirement to a loud-speaking set, as the loudspeaker was in the back and the test microphone during approval testing should be in front of the set. I proposed that a set should be approved if it only complied with one of the speech modes. At our next meeting Televerket came back with an accept of this, with an addition: If there was a handset, the requirements had to be complied with for this mode. Everybody agreed and we had also no problems with it.
Summer 1988: Trouble shooting by telex
One day as I was in Hilversum Betty phoned from Copenhagen: Philips had received a telex from one of the cruising ships that their passenger exchange had broken down. They had taken circuits cards out for some unspecified reason and as one of these cards was the CPU, the main processor with project data in chip memories, all project data were lost and the exchange could not start up again. They asked for assistance three days later in Miami.
In the first instance I could only consider this to be impossible. It would require a proper written order from Carnival Cruises before Philips would let a technician go and these preparations were impossible to make in such a short time. However, there was another way: All necessary items for starting up were on board (if nothing had broken during the activities). There was an emergency tape with all data, the MPU suitcase with a tape station and communication to the exchange and documentation on how to use all the equipment. However, during the last hectic days at the shipyard, where Carnival Cruises accepted the ship, none of their people had looked to the telephone system. There were so many other items to look for. Navigation equipment, alarms, radio equipment etc. No one on board had ever used the MPU and tried to load data into the exchange. But it was worth trying if they could!
Thus I advised Betty to telex them back, refer to the relevant chapters in the manuals describing how to load an emergency tape and ask the people on board to follow these directions to get the exchange up themselves. Betty did as advised, not only referring to the directions as advised by the local technician, Peter Miedema, but quoting them in full in her telex.
The next day we got a telex from the ship: They had done as advised and the exchange was in operation again, the day before the ship arrived in Miami.
What really made this success possible was not us with our idea and a long telex. It was rather those people in the teaching and documentation section in Hilversum, who had seen to it that the documentation was so clear and instructive that even people, who had never tried an emergency loading before, could succeed in making one.
Loading EBX 8000 from a PC
When an EBX 8000 was started up, it had to be loaded with all the programs and data. This took place during the first many years with paper tape, loaded into the exchange from a paper tape reader from the Danish factory GNT Automatic. We talk about up to 15 paper rolls, each about 30 cm in diameter. It was a specially prepared tape, lightly oiled and strong, but it was several hundred metres of tape which had to be read. The first readers were mechanical, but they were soon replaced by optical readers in which photo cells detected the light through the holes in the tape. Data from the exchange, e.g. about traffic measurements, could be read out to a paper tape puncher, also from GNT Automatic.
The reader had to comply with rather strict requirements – and it could! It should be able to stop immediately, before reading the next character in the tape. It should nevertheless be able to read thousands of characters per minute or the loading was too slow. The reading had also to be error free, each character was protected by a parity code (there was an even number of holes in each character) and if the reading of a hole failed, such that the parity became uneven, the reading stopped immediately.
The technicians from CTBD learned some tricks from the technicians from Hilversum, e.g. how one should not start from the beginning if there was a fault in the reading of a program tape. One could start over with the same tape after adjusting the content at a few places in the memory. It was also necessary to keep the reader in order, clean and well adjusted. During loading one could not let the reader pull the tape off the roll, you had to pull it off yourself and see to it that the tape always hang in a soft coil to the reader such that it should not accelerate a too great mass for each character.
All this was in short rather complicated, but during the 1980es the PC had appeared and Hilversum developed a new unit to the exchange for connection to a PC, where program and data were on the hard disc of the PC. This also required an additional program to the exchange, which was only developed to the second generation of them (those with up to 1 Mbit DST). Around 1988 Hilversum stopped delivering expansions on paper tape and we informed CTBD in good time that the first generation could not be expanded any more. This was in agreement with our contract with CTBD, in which we had promised that an exchange could be expanded during at least 10 years after first delivery and we had delivered the last exchange with program package 11 in 1978.
Thus it was only exchanges with package 20, which could be expanded, and which were in that case changed to accept loading from a PC. It was easy compared with the old ways! After having prepared the PC one should now only call for the procedure in the exchange for start of reading, and all went in a smooth stream. One had only to prepare the reading of data and start it after the reading of programs was finished. It was also possible to prepare a file in the PC with a list of additional programs wanted, and then they were read in according to the wanted sequence.
The first few times an expansion took place CTBD bought assistance from Hilversum, but soon they wanted to do it themselves. I had to be a back-up during the expansion but had no training from the daily work with the exchange. I saw to it that I got a course in Hilversum during March 1989, during which I could train on an exchange in the training centre supervised by one of the trained installers, Pols. It was a very intense week but I was more at home with what might happen. The next expansion after my return to Copenhagen went flawlessly.
Summer 1989: A Short Visit to Stockholm
Besides the commercial contacts in a country the development laboratories were in contact with corresponding laboratories of non-Philips companies. This was in my opinion a bad system, the local management ought to have control with all contacts in their area, but that was the way it was. In this way it had been arranged between Hilversum and the laboratory for ISDN of Televerket that they should have a SOPHO-S 50 for their experiments.
If they had asked me I would have protested with much vigour. SOPHO-S could at that time not serve ISDN sets at the extension side except its own digital system sets (which were by the way far further developed than anything ISDN could provide at that time), and it had no possibility for ISDN links to the public network. In short: it could not at all be a part of a trial of ISDN!
As it was I was only asked to go to Stockholm and bring the exchange into operation. So I went by plane to Arlanda and by taxi through Stockholm to the laboratories in Farsta and there the SOPHO-S 50 exchange was waiting.
SOPHO-S 250 and 1000 should like TBX have their project data loaded from a cassette tape read on the MPU, the special suitcase with in-build processor etc. SOPHO-S 50 and 100 might have their project data loaded in the same way, but it was also possible to load them from an internal ROM-chip (Read Only Memory) with a fixed set of national project data. Then one should only switch on power and after a short time these data were loaded into the volatile memory. After this one should just load the variable data from the operator desk and start operation.
The laboratory in Hilversum had not sent an MPU with the exchange and it was soon clear for me that there was also no ROM in it. The exchange could not at all be set into operation! A fast call to Hilversum to report this and less than an hour after my arrival to Farsta I sat again in a taxi going to Arlanda and home.
A week later an installer from Hilversum came with the required ROM and installed it. I was again in Stockholm although I rather had the impression from Hilversum that they wanted me to be anywhere else. I tried to be very subdued so the installer would have no trouble when he came home.
In the first moment the exchange did still not come into operation. A renewed inspection of the card with the ROM showed it to be mounted in the wrong way, being turned 180º. In one end of a chip there is a mark and on the circuit card it is shown which way the mark shall be turned. The installer had not been careful enough when mounting the chip and it was burning hot.
We were of course afraid that the chip was spoiled by the wrong power connections and the heat. But this should be tested, the chip was turned and the exchange was again powered up. A miracle! It worked perfectly. It was for me quite fantastic that the chip had survived the wrong mounting.
What use the laboratory of Televerket made of the exchange or which discussions they had with the laboratories in Hilversum I do not know. This went quite outside Philips Nordic.
Sales names for telephone sets
Along with SOPHO-S we also got digital system telephone sets, i.e. sets with a digital connection to the exchange but with more facilities than ISDN sets. The exchange used the standardised ISDN interface with 144 kbit/s in both directions on the line simultaneously (this had been planned since the early 1980es), but the signalling protocol was adapted to the possibilities of the exchange and PABXes have always provided far more facilities than public exchanges – it was a simple necessity at this market. Services appearing in public exchanges in the 1990es (automatic ring back, call notification, calling number display) were standard in PABXes already in the 1970es.
With the availability of the new sets came the question what they should be called when they were marketed?
Finn H. Nielsen, the head of Philips Tele in Denmark (since Øberg became head of Philips Tele Nordic) started a competition between the employees (including us in Tele Nordic): Who could come up with the best name? I do not remember what the winning names were for the three system sets or what I proposed (only that it was not the winning names), but all participants remember the proposals from the secretary of Tele Nordic, Betty Bjerre.
Her inspiration was the difference between the three types, a simple one having only a little to offer above a normal analogue set, a larger one with a row of facility keys and the largest type with a large keyboard making it possible to load alphanumerical messages into the memory of the set and have them shown on the display. One could e.g. make one’s own telephone directory in the set, search a name and when it was found call it with pushing a key.
Thus her proposals, giving her a “commendable” were in Danish fedterøv, pralerøv and blærerøv, three very denigrating expressions about people!
September 1989: Test Results for SOPHO-S
Hilversum had modified a SOPHO-S 2500 exchange and sent it to Stockholm for approval by Televerket. It had been tested in the laboratory of Televerket and the result was a list of points where the system fell short of an approval. This was a serious matter and in September 1989 there should be a meeting where we and the laboratory should discuss the results with the participants from Televerket in the standardisation work and with the head of the approval office. Two people had come from Hilversum with strict orders not to be present at the meeting, “as their statements could to a higher degree than mine make obligations for Philips”! Thus it was from Philips just Mats Rasmusson from Philips Sweden and me who came to the meeting. As Mats was commercial I had to do the talking while Kenbeek and de Winter from Hilversum covered up in the hotel.
The meeting was an examination of all the points where the laboratory had found that SOPHO-S did not comply with the requirements in the standard. And I could at all points refer to decisions in the standardisation group, where Televerket had accepted modifications to the standard such that the requirements were complied with after all. The participants from Televerket in this meeting could only confirm this, the standard had only not been made up to date yet. The approval office noted this and shortly after the meeting SOPHO-S got its approval and the sales work could start in earnest. This led to several orders during 1990.
We met the two Dutchmen after the meeting and reported the result. We also teased them for not daring to be present (or rather that the management in Hilversum did not let them be present). I wonder what they reported to Hilversum, in the first instance it must surely have sounded much too good for the management there, but as the exchange as said after a short interval was approved with no further modifications they could see that the report from Mats and me was realistic.
Visits from Sweden
Just like we had helped the British sales organisation to get the sales of EBX 8000 going in the end of the 1970es we could now in 1990 help the Swedish get the sales of SOPHO-S HR going. We had one in our own offices and could also show the operators positions for the Swedish customers. This became my main job for Sweden during 1990 as the type approval was in place and the work with the standard was also finished. We received several visits (no Swede will mind a visit to Copenhagen) and I think we were in this way a help in getting Philips faster on the market in Sweden.