Piet's Home-built Television

Television development at Philips 1946-1956

One of the childhood memories I have is on the home-built television receiver of my father Piet Hooijmans. A large (and heavy!) grey metal box with a very small picture screen. I must have been 5 years old, when on a Saturday afternoon around 1963 my father switched on the television set. I vividly remember images of a horse race, probably at nearby Duindigt in Wassenaar. It was the last time I've seen the set working, but it has always been carefully preserved. Later when I studied electronics myself I opened it once in a while, but never had the time nor the means to dive into it. It has always fascinated me. Now I have the time. The first thing that triggered me was the exotic EQ40 valve of which my father always talked about as the "enneode". This has resulted in my page on the history of the EQ40/EQ80.
Next I wanted to trace back the history of the home-built TV itself, although I was hampered by a severe lack of information about it. When starting to dive into the Internet I quickly discovered that the home-built set was tightly linked to the development of the first televisions in general, and that the activity of the Radiobuizen Lab where Piet Hooijmans worked had an important influence on the direction of these developments. The intended story on his home-built TV thus quickly became the integral story of the development of the first televisions at Philips. Or at least an effort to reconstruct it, since it is a much more complex story than it may seem. It will involve multiple departments, multiple standards, multiple models and valve choices, official prototypes, in-official do-it-yourself models, and the first commercial products. It is a reconstruction based on available data, which at least on the level of actual television sets and valves is very extensive thanks to forums like Radiomuseum.org, the Nederlandse Forum voor Oude Radios, UK Vintage Radio and others. And although my friend Ronald Dekker has already done a lot of pre-work on the internal Philips organisation at the time for his EF50 story, there are still many gaps related to the internal Philips activities. My analysis, which roughly covers the period 1946-1956, is thus also based on my interpretation of facts and developments, and consequently a "living story" since never complete nor 100% correct.

Chapter navigation

Pre-War Philips Television activities

One of the two Philips Television Demonstration lorries that toured Europe between 1937 and 1939. Here seen during one of its first appearances at Eindhoven Welschap airport. The base of the 10m antenna is also visible. [Philips Technical Review 1948 and Eindhoven in Beeld]

One of the many demonstrations with the television system, this one in 1938. The receiver is clearly based on the back-projection principle, since at this time there were no picture tubes of the size shown. [DBNL]

Between 1935 and 1939 Philps Gloeilampenfabrieken, as it was officially called, was already active in promoting and demonstrating television.
These activities on television were still rather limited. In the Philips Research Natuurkundig Laboratorium there were J. van der Mark working on the transmitter and Herre Rinia on the receiver, while in the components groups M.C. Teves worked on inonoscope registration tubes and M. Wolf on display picture tubes. It seems that the main receiver set activities were in the UK at the Mullard Lab, with C. Richards driving that. A very first public demonstration prototype was built for the 1937 "Radiolympia" show in London, which will be discussed in one of the next sections. It was a system based on rear projection, which was at the time the dominant display technology endorsed by the management. It must have used the new UK 405-line electronic system then introduced by EMI and used for the transmission from the Alexandra Palace transmitter station.

The company built a number of demonstration cameras, low power transmitters and receivers, which were built into two dedicated lorries from the Eindhoven car and truck manufacturer DAF. It was a 180-line system also based on rear projection, which was at the time the dominant display technology endorsed by the management. After initial demonstrations in the Netherlands and Belgium, the two vans started tours throughout Europe, mainly visiting the capitals of the European countries. These were often seen as astonishing breakthroughs, attracting large crows and much publicity. For example in Romania demonstrations were given to King Carol II and his family, where the arrival of the King was broadcast live. When the 2nd World War broke out in September 1939 both demo vans were in Eastern Europe, one in Zagreb (Croatia) and one in Warschaw, Poland. The first one could be re-patriated by sea, but the one in Poland had to be abandoned due to the fierce fighting in and around the Polish capital, the Philips crew was barely able to escape alive and return to Eindhoven. The remaining vans were parked on the Eindhoven-Strijp complex and were destroyed in the December 6, 1942 RAF bombing raids on the Philips factories. This and the subsequent occupation of the Netherlands put an official stop to Dutch and Philips television developments.

Television at the Natuurkundig Laboratorium

From the beginning, the general enthusiasm for television within the Philips management was lukewarm at best. In this period television was only a research activity, and thus driven by the Natuurkundig Laboratorium (NatLab) at the Strijp complex in Eindhoven. Especially its emblematic director Prof.Dr. Gilles Holst (1886-1968) didn't see the need for or potential of television, where his main arguments against it were a lack of programs and the size of a television set with an acceptable screen diameter. On the latter point it is fair to say that cathode ray tubes at the time were still very narrow, a 6cm screen being the typical size. The problem was especially the then limited deflection angle, which would give on extrapolation enormous picture tubes for larger screen sizes. It was therefore decided that the Philips television receivers were to be based on rear projection systems. And despite the reservations, activities continued.
The work on projection TV concepts was thus concentrated in the NatLab research organization. The pre-war activities were done in two groups: transmitter work in the Advanced Radio group by the famous Dr. Balthasar van der Pol (1889-1959), led by J. van der Mark. Van der Pol had obtained fame with the first short wave transmission link to the Netherlands East Indies in 1927, as well as his work on non-linear relaxation behaviour of circuits, which had led to the famous "Van der Pol"-equations. He also discovered the "White Noise" phenomena. At the time van der Pol himself was working on ionospheric propagation. He and van der Mark already worked together since the mid twenties, when they built an electrical model of the heart, that could be used for heart rhythm analysis.

The receiver was developed in the group of Dr.Ir. Herre Rinia (1905-1985). Rinia was one of the first NatLab employees, joining in 1922 and working for van der Pol on amongst others the first pentode valves. He was very much involved in the definition and development of the projection TV concept, such as the optics. Probably after the war he took over the research group on television receivers from Oosterhuis. One of the engineers in the group of Rinia was Dr. Johan Haantjes (1908-1978), who joined the NatLab in 1937 and worked on magnetic deflection circuits and broadband amplifiers. Another interesting person was Freek Kerkhof, a radio amateur who had built a Nipkov-disk based television at home as early as 1927, and who started experimental television transmissions from his Eindhoven bed room in 1936. Kerkhof joined Philips in 1932, working closely with Haantjes on the receiver while continuing to be a very active amateur in his free time. It is possible that Kerkhof moved to the Apparaten-Lab in the Hoofd Industrie Groep (HIG, in modern terminology a Product Division) to lead product development of televisions. Van der Mark was the author of the system article on the 180-line TV system published in the very first volume of the Philips Technical Review in 1936.

In 1938 the then head of the Apparaten Lab, ir. P.R. Dijksterhuis warned the Philips management that in his view the television activities where too low level and unstructured, and would not lead to a strong Philips position in this domain. He proposed that the majority of the NatLab research activities should be devoted to television, including a strong focus on radio valves optimized for the application. The first was blocked by Holst, but it was agreed that the NatLab would step up the television activities in order to build a strong patent position, and valve activities were also increased.

During the war all product development and system experiments on television were stopped, but theoretical work continued, albeit on a lower level. One of the topics was theoretical analysis of beam deflection in display tubes, an investment that would pay off in the following years.

After the war work on television was taken to a higher level again at the Natuurkundig Laboratorium. In 1946 its founder Holst retired, and was replaced by three directors each covering a fundamental group of activities: Hendrik Casimir for physics/devices, Evert Verwey for chemistry/materials and Herre Rinia for systems/electronics. This NatLab structure would exist for fifty years till well into the 1990s! One of the group leaders was Prof.Dr.Ir. Johan Jonker (1901-1963), who had between 1931 and 1936 had created the Radiobuizen Lab, which will be discussed later. After his return to the NatLab his group continued work on advanced valves, such as high frequency and high power transmission tubes including magnetrons, klystrons and travelling wave tubes, cathode ray tubes and camera tubes. His team was also responsible for the invention of the EQ40 enneode phi-detector as well as the E1T counting valve. In the sector of Rinia Haantjes took over the receiver group, while van der Pol continued to lead the transmitter activities until his retirement in 1949. That group was then taken over by van der Mark. Some time later all television research activities would be integrated under Haantjes, who led them till he became vice president in 1956.

Gilles Holst, first director and founder of the NatLab.

Balthasar van der Pol

Herre Rinia

Hendrik Casimir (left) and Johan Jonker

Johan Haantjes, for a long time leader of the television receiver activities at the NatLab. [Philips Koerier 1962]

Freek Kerkhof with one of his earliest Nipkov-disk based television systems. [dBNL]

Some of the Philips NatLab key players in early television. From left Erik de Vries, producer of the experimental programs demonstrating a camera, prof. Holst, director of the NatLab, van der Pol, group leader of the TV transmission group and van der Mark, group leader of the TV receiver group, around 1938. [NatLab site Hagenbeuk]

In 1946-47 the NatLab groups picked up where they had left the television development in 1940, and built an upgraded system, both a transmitter and a receiver. The core of this was the new high intensity MW6-2 projection tube, which was integrated into a Schmidt optical projection system. The transmitter power was increased from the roughly 500W of the 1937 system to 9kW using new water-cooled PAW12/15 transmit pentodes. Initially (1942) this transmitter was still using full dual side bands (2x2,5MHz) of the signal, only later vestigial side band modulation would be introduced.

Philips UK Television activities 1936-1946

Although my story in principle starts in 1946, it is essential for understanding the first post-war steps to have a brief look at especially the pre-war television developments in the UK. In pre-war Europe the UK was leading by far with respect to the roll-out of television. As early as 1935 the BBC started experimental television transmissions from the Alexandra Palace transmitter, initially still transmitting using two standards: an old mechanical one next to the new 405-line "electronic" standard as invented by EMI and Marconi. The initial number of TV's sold was of course very low, but around London the number of watchers steadily grew. Especially the sales department of Philips, and its local subsidiary Mullard, was pushing hard for an active role of the company in this new market, which was supported by Philips board member Loupard. Although Holst, as the main scientific advisor, was still not enthusiastic about television, the NatLab activities were at the point of being able to deliver a prototype system. It was therefore decided that Philips would present a prototype receiver at the 1937 annual Radiolympia electronic show.

Block diagram of the Philips Tel6 rear projection television demonstrated at the 1937 Radiolympia show in London, showing the valves used for each function. The set also contained a full radio chassis which took care of sound amplification.

This set, which in an overview of the Radiolympia sets was called Tel6, was a rear projection receiver in a large cabinet, also containing a Philips 785AX-model full band radio receiver chassis, bringing the total valves to 29. A few interesting features were:

Video and sound only had the first RF amplifier in common, but then had separate down-converters and IF chains. IF frequencies were very low; 11,1MHz video and 7,6MHz sound.
After AM sound detection the audio signal was connected to the record player input of the radio which thus provided the audio output amplifier.
Both the frame and line time base generators used gassed filled triodes in the oscillators
The high voltage generator only used two high voltage rectifiers in a voltage doubling set-up, and not yet the fly-back concept with efficiency diodes.
The projection picture tube was a 4" (11,4cm) high intensity MS11/1, the output of which was projected via a 45degree mirror onto the back of a glass sandwich projection screen. Because there was not yet corrective optics between the picture tube and the display screen, the front of the tube was concave (bent inwards) in order to provide an undistorted picture.

Given the exclusive use of Mullard valves, in combination with the UK being the only television market at the time, is it highly likely that the set was developed by the Mullard Labs, the UK subsidiary and local brand name of Philips. Clearly with support from the NatLab Research, but almost certainly not built in Eindhoven. The receiver concept was reported in the February 1937 Philips Technical Review by C. Richards (almost certainly from Mullard Lab), while the rear projection concept of the set was published by M. Wolf in the Philips Technical Review of August 1937.
The device initially got a lot of positive attention and reviews, especially for the bright and large screen. Despite that the set was removed from the show after only three days, nevertheless a number of sets seems to have been sold, produced and delivered to customers. Unfortunately most of these had to be re-called, with Philips even offering to reimburse the cost. The reason for all of this was the extremely short life time of the projection tube. This was partly due to the high currents and voltages at which it was operating, and partly due to its sensitivity to breakdown. As soon as something would be wrong in the frame of line deflection drivers the result would be a horizontal or vertical line that would burn into the picture tube phosphors and destroy the tube. Having learned from this, during the 1938 Radiolympia show Philips presented an improved set, the Tel61. Essentially it was the same platform but with three gas filled triodes as electronic relays that switched off the deflection and anode voltages to the picture tube in case of problems. Another improvement was that the IF moved up to 13,2MHz for video and 9,7MHz for sound, values that would remain so for the next decade.

Three pictures of the 1937 and 1938 Radiolympia Philips UK television sets. On the left the 1937 model Tel6, showing the large 20x16" (50x40cm) screen. In the centre the interior view of the same TV, showing at the top the radio receiver and below small signal (left) and deflection/HV modules (right) with between them the picture tube module. On the right the 1938 Tel61 model, slightly more compact with a 18x14,5" (45x39cm) screen. Note that the picture screen dimension where 5:4 for the 405 line standard, only later they switched to 4:3. [Images from review articles Television and Short Wave World Magazine October 1937 and September 1938, via TheValvePage.com]. In the centre below the picture tube used by these two sets, the MS11/1, with the clearly visible concave face. [PTR August 1937]

For 1939 Philips had bigger plans, preparing four or five different TV sets for the UK market: the 2405 (a 9" table model), 2407 (a 9" console model), 2412 (a 9" radio-TV console model), the 2409 (name not confirmed, a 12" console) and 2415 (idem with radio). They were all shown at the 1939 Radiolympia show and production was planned for the fall of 1939, when due to the outbreak of war on September 1st, 1939, the BBC stopped its television broadcast and Philips cancelled production. Pictures show these models to have direct viewing picture tubes, which must have been the MW22-1 or -2. In an author-less article "compiled by C. Heller" in the Philips Technical Review of December 1939 the latest picture tube receiver was described, although without mentioning a single valve type.

One of the most remarkable features of the pre-war models is the RF mixer-oscillator concept, where video and sound have their separate mixer heptodes, while the two oscillator triodes use a common tank circuit. [PTR 12-1939]

Philips 9" (22cm) direct view TV sets for the 405-lines UK market that were to be launched end of 1939. From left to right the 2412 radio-TV console, 2407 TV console and the 2405 table top models. [PTR 12-1939]

The electronics of the 1939 sets: from left to right the MW22 picture tube, the HV and deflection module, behind the small signal module and in front of that the deflection coils. [PTR 12-1939]

During the war years most companies in the UK, including Philips/Mullard, switched to military applications and had no resources to work on consumer applications. When the BBC resumed television broadcasting in July 1946, most manufacturers - with the exception of Pye, which had secretly developed a new platform during the war - had to start where they had stopped developments in 1939. The same was true for Philips, and the first family that was developed was thus heavily based on the 1939 Tel61 platform, the last one that had made it to production. In 1947 four models were launched, the 463A, 563A and 663A, as well as 383A. Where it seems the 383A re-used the 2405 cabinet from 1939, and the 663A looked very much like the Tel61 albeit without the projection screen. All used the new Philips MW22-7 9" picture tube, but otherwise the block diagram and concept were identical to the Tel61 from 1938. The main modernization was the replacement of all pre-war Mullard valves by standard Philips valves from the E30 series, the EF50 - the Mullard work horse since its introduction in 1940, the famous valve "that saved Britain" - and the first miniature valve EB91. [The EB91 in this platform is a bit of an enigma. According most official documentation this valve was launched in 1949 as part of the miniature Noval series. Since the four sets were shown at the Radiolympia 1947 AND the official service documentation clearly lists this valve, something doesn't fit. As we will see further down it happened regularly that valves were used in internal development and reference designs well before public launch. I therefore tend towards the scenario that the EB91 was available in 1948 for internal set development and that there is no conflict in timing.]

Block diagram of the very first commercial Philips TV set after the war, the TG460 platform. It was only produced for the British market (405 lines), mostly using Mullard-produced valves like the EF50.

The Philips 663A from 1948, a high end set with TV and radio in a console. [Philips Service Manual via TheValvePage.com]

Television Standards 1946-1948

During the first years after the war, 1945-1946, there were - in Western Europe - essentially only two countries with regular television transmission: Britain and France. In the background there was of course the US, where television transmission had started during the second half of the war years. It won't be a surprise that all three countries used different standards, summarized below:

The three relevant television broadcast standards in 1945-46.

The British 405 line standard and the French 441 line standard were both based on pre-war systems, from a time when achieving bandwidth was still a challenge. But by 1946 especially the 405 line system was considered as too low in resolution, and on larger screens the lines became visible. "Spot Wobble" was used as one option to smear out the lines, of course at the expense of vertical resolution. Although the French were transmitting the 441-line signals from the Eiffel Tower in Paris, it was clear that it would not be the targeted system for France. Based on this situation as a starting point, there were two more or less parallel development tracks in Europe.

The first was a British-French alignment. On the one hand the British were leading in Europe, with actual television transmissions starting in 1946 from the Alexandra Palace transmitter in London, with more transmitters being installed throughout the country in the years after. Britain openly had the intention to spread its standard towards the continent, to facilitate exchange of programs and to support British industry. However, in France the trend was clearly to higher resolution standards, initially a 1020 lines system was proposed, but by November 1948 French authorities decided for the competing 819 lines system. Compared to all other standards of that time this could be considered as high definition television, which was seen as part of the policy to push French technology excellence, of course at the price of high channel bandwidth and thus fewer transmitters per band. (There was place for only two channels in VHF-I, and most were consequently in VHF-III). Intensive discussions took place between the French and British television authorities, converging on the following plan: the French were to adopt the British 405-line system for the 45-80MHz VHF-I band, thus replacing the Eiffel Tower 441 line transmissions. In return the UK would adopt the 819 lines HDTV system for transmission in the VHF-III (160-210MHz) band once it would move there. Jointly this would then be the proposal into the CCIR (Comité Consultatif International du Radio, headquartered in Genève, Switzerland) standardization committee of the ITU.

The other track was building on the US NTSC system, in practice the best system then operating. Especially RCA was promoting the re-use of it standard in Europe, adapted from the 60 to 50Hz power supply frequency. And this is where Philips comes in! Although it is not possible to re-trace the history in all details, it is likely that Philips decided to build upon the US system due to its own presence in the US, potentially benefiting from a technical synergy between the two systems. As a consequence Philips developed a 567-line 50Hz (25 frames) standard, giving a line frequency of 567*25= 14.174Hz, close to the 15.750Hz of NTSC. In theory 567 lines in combination with 4:3 screen size would require a video bandwidth of 5,5MHz and a channel width of 7MHz. However, for maximum compatibility with NTSC a lot of that standards parameters were taken over: 6MHz channel, 4,2MHz video bandwidth, negative modulation and FM sound. The lower video bandwidth resulted in a lower horizontal resolution, but delivered a very economical system and high synergy with the US system. The only real difference was the transmit polarization of the Philips system, which was vertical as opposed to the horizontal NTSC polarization.

The 567-line standard was defined by a group of people in the research groups of van der Pol and Haantjes. On December 7, 1945 an internal NatLab report was written by J. van der Mark explaining the choices for the bandwidth and line rate. Philips was quite bullish about its standard, and seriously tried to promote it as the European standard to the surrounding countries. Johan Haantjes was the main promoter of television broadcast within Philips, and also the primary external contact to Dutch authorities, the PTT and potential broadcast corporations in Hilversum. He was likely also the representative proposing and promoting the Philips standard in the CCIR.

The four standards fighting for standardization in Western Europe 1947-48, plus the US NTSC reference.

Philips rear projection TV, 1947

As said earlier, the Philips television strategy was primarily built around projection TV. This was the concept that had been used for the 1937-39 demonstrations, and on which the research activities had been continued during the war years. When television started to gain momentum again after the war, Philips development activities continued based on this concept. After the definition of the 567-line standard, the next step was to build a prototype system around it. Many details of this prototype system are known through a number of publications in the research Philips Technical Review journal, with articles written by some of the key players introduced earlier:

PTR Vol7 1942, No5, pp.129-137, A nine kilowatt experimental television transmitter, M. van de Beek
PTR Vol10 1948, No3, pp.69-78, Projection Television Receiver I. The optical system for the projection, P.M. van Alphen and H. Rinia
PTR Vol10 1948, No4, pp.97-104, Projection Television Receiver II. The Cathode-Ray Tube, J. de Gier
PTR Vol10 1948, No5, pp.125-134, Projection Television Receiver III. The 25kV Anode Voltage Supply, G.J. Siezen and F. Kerkhof
PTR Vol10 1949, No10, pp.307-317, Projection Television Receiver IV. The circuits for deflecting the electron beam, J. Haantjes and F. Kerkhof
PTR Vol10 1949, No.12, pp.364-370, Projection Television Receiver V. The synchronization, J. Haantjes and F. Kerkhof.

The core of the receiver was the new high power picture tube, the MW6-2, a development of the pre-war MW6-1. It had a very high light output concentrated on a small frontal fluorescent screen with a diameter of just 6cm. This high light output was achieved with a high accelerating voltage of 25kV, which was quite exceptional for the time. Where exactly the MW6-2 was developed is not entirely clear. Given that a substantial article was published in PTR supports an origin in the NatLab, then most likely the group of Jonkers. However, the formal product code indicates it was developed as a real product in the valve division, possibly by Mullard in the UK.

The optical unit was clearly the most complex of the television receiver. The purpose was to project the image of the MW6-2 picture tube on much larger display screen with as little deformation and loss of light intensity as possible. For this a Schmidt optical system had been selected, the same as used in most astronomy telescopes. [In this context it is interesting to know that Philips also internally produced the 30cm telescope for the Philips Astronomical Observatory in Eindhoven (today the Dr A.F. Philips Sterrenwacht). This telescope was designed by Herre Rinia, who was also amateur astronomer!]. The picture was magnified roughly 6 times and rear-projected on a 32x40cm frosted glass projection screen. This was considered large enough to have multiple people in the same room watch the television program. The optical module was fairly compact in a roughly 30x30x30cm enclosed box, with only the MW 6-2 foot sticking out for electrical connections. The light exited the box almost vertically through a glass plate window, which made cleaning easy and kept the optics dust-free.

The electronics of the first TV set were very basic, and clearly based on the system designed during the previous years. It was thus almost entirely based on the E20/30 Miniwatt series, with valves such as the EBC33, ECH21, ECL21, EL34 and EL38. Only a few modern valves that had just been released like the EA40 efficiency diode and EY51 high voltage rectifier were used. Special attention was required for the 25kV high voltage supply for the MW 6-2 anode, which was at the time a very high voltage for a consumer device. The third paper in the PTR series describes the blocking oscillator concept, followed by a cascaded rectifier stack.

Now it becomes clear why we had to look into the UK television platforms earlier, because around this new Schmitt-optics display unit a television set was designed based on the pre-war (Tel61) and 1946 (2405-2407-2412) UK platforms. The first set designed using the new concept was the SG860, where G indicates a product destined for Britain, the 6 the year of design (1946) and the 0 that it was the first product. [It is unclear what the S stands for, because this letter has not been used any more in product coding. It might be that a separate product family code was intended for projection TV, as opposed to direct view television (with the first letter T)]. This was a large console including a radio just like the Tel61, with which it shared many external characteristics. Like all 1946 designs it had replaced many pre-war valves by the EF50 and E30 octal valves. And it introduced of course the MW6-2. Overall it must have had 31 valves, although the exact valve allocation is not known yet. Despite appearing prominently in the PTR publications, the SG860 seems to have never been produced nor sold in the UK under this name.

Visit to the NatLab TV group by board member Frits Philips (right) in 1947. He's probably looking to the rear projection optics assembly of the prototype SG860/SX861. [Site NatLab Hagenbeuk].

The SG 860 A. All pictures in this section from Philips Technical Review Vol.10.

Philips TX861A, the first 567-line television, 1947

Immediately after the 405-line SG860A developments were started of a 567-line version, which would become the SX861A. Conceptually it copied the SG860A, although the receiver required adaptation to the new 567-line standard. It also introduced a number of new functional concepts:

The RF input (pre-amplifier and mixer-oscillator) was now common for video and sound, and only after the 2nd IF amplifier the sound IF was split off into its separate chain.
To accommodate the 1MHz higher channel bandwidth the video IF was pushed up 0,8MHz to 14,0MHz and the sound IF down by 0,2MHz to 9,5MHz.
The set introduced the line fly-back concept with an EA40 efficiency diode.
It used a separate 5-valve high voltage generator for the 25kV of the MW6-2, using three new EY51 high voltage rectifiers in a voltage tripler arrangement. A single EBC3 was used as protection circuit, switching off the MW6-2 electron gun in case of any malfunction in the deflection circuits.

From the outside the SX861A looked identical to the SG860A. Internally the valve count added up to 35; 23 for the television core, 6 for the 25kV high voltage module and 6 for the BX660 radio. The set was of course not produced in series, given the lack of 567-line transmissions. But some 20 were built by the Apparaten-fabriek as pre-series, and were used in the experimental television transmissions that started in 1948. Most amazing is that one of these rare sets has survived the seventy years since then and is restored and working by Jac Janssen.

The unique restored and working SX861A of Jac Janssen. [Jac Janssen private collection]

The SG860/SX861 was also used for publicity at e.g. shows, such as the June 1948 Feria in Barcelona, Spain.

Block diagram of the Philips SX861A rear projection television plus radio. Only a small series of some 75 pieces was produced and used for field tests during the Philips Experimental Television transmissions, starting March 1948.

Philips Experimental Television (PET), 1948

In order to bring television promotion and publicity to the next level, Philips decided it was time to start actual broadcast transmissions of television programs. Because there was no organization able to organize and execute this, it ended up as a 100% Philips activity. To start a transmitter was installed on the roof of one of the tall buildings of the massive Strijp-S factory complex in Eindhoven. On the southernmost distribution centre building (the so-called "Veemgebouw") of the Apparaten factories a lattice tower was erected, on top of which two antennas were mounted, one for the picture transmitter and one for the sound transmitter. Note that at this time the video and audio baseband signals were not yet combined into one composite television signal, with the sound carrier at 4,5MHz as was done later. In that case the composite baseband signal is in one step modulated on the picture carrier, with a guaranteed frequency distance between picture and sound carrier. By the time of the experimental transmissions picture and sound were transmitted by two separate transmitters, that had to be precisely controlled in terms of frequency separation and power level. In most systems the sound transmission power was 4 times (6dB) lower than the picture carrier. The video transmitter was called PAB3, the sound transmitter PAG3 (my guess is the abbreviations stand for Philips Apparaten Beeld and Philips Apparaten Geluid). The video transmitter had an effective transmit power of 20kW, giving a normative reception range of 30-40km. The picture transmission frequency was 63,25MHz, sound was at 67,75MHz, as mentioned earlier both with vertical polarization.

Only the next requirement were programs to be broadcast. To this end a very small one room 5x11m studio was created on the ground floor of the NatLab, opposite the street of the transmitter building. The studio room was so small that apparently it had to be entered through the window, the door being blocked by all equipment. Especially the cameras were still massive. Programs were organized and filmed by Eric de Vries, who had already been involved with the television demonstrations before the war. A variety of programs was transmitted, often a mix of flower arranging, child care instruction, music, comedy and other forms of risk-free information and entertainment. Much later, in 1950, the first life soccer match registration of the local derby PSV-Eindhoven would be transmitted, which generated much more enthusiasm. It should be remembered that the television experiments were not just to test the electronics of the system, but also to explore what television could mean to the public, which programs could be relevant and what the viewing experience of the viewers was. After all it was an entirely new medium! Philips had requested a radio licence for television transmission from the Dutch PTT, which they received March 18, 1948. The same day the first experimental transmission was made of what was now called Philips Experimentele Televisie (Philips Experimental Television, PET). The first transmission was announced by Bep Schäfer, a Philips secretary, as "Goedenavond, dames en heren. Hier is Philips Experimentele Televisie met de eerste uitzending. Beeld en geluid over de zenders PAB3 en PAG3...". Regular transmissions started April 1st, 1948, three times a week for 2-3 hours.

The Philips Experimental Television PET transmit tower on the "Veemgebouw", the central distribution center and the southernmost building on the Strijp-S complex in Eindhoven. [Eindhoven in Beeld]

Aerial picture of the southern part of the Eindhoven Strijp factory complex of Philips, looking north. The white encircled low-rise section in the foreground is the Natuurkundig Laboratorium (NatLab), with in one of the left-most buildings the PET studio. The long high-rise buildings running along the central axis of the complex belong to the Apparaten organization, and contain both the radio and television development as well as factories. On its western-most building indicated by the arrow the PET transmitters were located. [PTR Vol.13 1951]

The transmission monitoring and control equipment in the PET studio. [Eindhoven in Beeld]

PET studio during a television program transmission. Behind the camera left Eric de Vries, on the right Bep Schaefer doing the announcements. [DBNL]

The third element of the PET system were of course the receivers. To this end some tens of the SX861A experimental receivers were placed in the homes of mostly higher Philips management and local dignitaries. The actual number produced differs, however, some sources saying 20, others 75. In any case, according all traceable reports, all receivers were of the projection type, which makes it plausible they were all of the same design. People with a receiver set were asked to fill in questionnaires about both the system, perceived performance as well as feedback on the programs.
Next to the officially distributed Philips receivers quickly a serious number of radio amateurs started building experimental receivers, and the PET receptions were received up to 150km from Eindhoven. During a big electronics show in Groningen in November 1948 the PET transmissions were received too, a distance of 220km. Very quickly circuit diagrams and instructions for home-built receivers were published in e.g. the "Electron" publication of Veron, the Vereniging for Radio Amateurs in Nederland.

Philips Protelgram

But Philips didn't invest in all these developments just for fun or for the love of science; it wanted to do business. The first opportunity seemed to be the US. There television was booming since its introduction in 1943. Per month some 10.000 sets were produced, which were for a large part based on projection concepts similar to the one developed in the NatLab for the Philips 567-line system. But one of the main advantages of the 567-line standard was the synergy with the 525-line NTSC system, from which it was derived. The optical engine of the SG860A/SX861A receivers presented above could therefore be used with only minor adaptations in an NTSC television receiver. (The MW6-2 picture tube was registered in the US as 3NP4). Already in 1947 it was thus decided that the optical engine would be produced in the US under the name "Protelgram" and sold by the local Philips organization Norelco. A production facility at Dobbs Ferry (New York) took up production, and in 1948 Norelco started promoting its projection television module. Because at the time Philips didn't have its own television manufacturing organization yet in the US, only the optical engine plus its 25kV supply was sold in different versions as the Protelgram unit; Model 161A for 13x10"displays with rounded corners, Model 160A for 16x12" square displays, and Model 162A for movie-type projection on large screens. It needed be combined with a regular television chassis for signal reception and the deflection signal generation. The unit was used in amongst others the Scott 6T11A and Emerson television sets. From the Protelgram Service Manual as well as the 1948 Norelco advertisement it can be seen that the box is almost completely identical to the one presented in the Philips Technical Review in the same year. This means that the Research module was taken almost directly into production, which has proven over again to be a very risky approach, because the product had not been properly engineered. The same was true with the Protelgram module, and the production in Dobbs Ferry experienced many serious manufacturing issues, limiting the output of the fab to only 200 units per week. As a consequence customers massively cancelled orders, creating a serious business crisis.

The main reason that killed the Protelgram, however, was not the complexity of the design but the development of larger direct view picture screens. Especially RCA, as one of the larger US players, had invested in projection tubes with wider deflection angles. In 1946 it had introduced the 10BP4 10" round picture tube with a 50 degrees deflection angle. In 1948 the slightly larger 12" 12LP4 was introduced and in 1949 the 16" 16GP4. In 1946 most television models were still rear projection (with systems similar to the Philips Protelgram), in 1948 the mix rear projection-direct view was already 50:50, but by 1950 almost all RCA sets were based on the 16" direct view picture tubes.

This didn't go unnoticed to the Philips management. On March 5, 1948 ir. J.A.J. Bouman, head of the HIG Apparaten, sent the following alarming telegram from the US to his colleagues in Eindhoven:
"Try to stop development work broadcast-receivers and concentrate all efforts on television stop Television is our biggest chance stop Protelgram is all-right but we have to work on follow up like hell! stop Write on all doors and walls and blackboards TELEVISION stop Make everyone television crazy stop We have enough people to do the job but most of them work on the wrong items stop There really is only one item: TELEVISION stop The only actual television front we have at the moment is right here in U.S.A. stop
We are able to force it if we are ready to fight AND TO KEEP FIGHTING! stop Mobilise Eindhoven please stop No time to lose TELEVISION IS MARCHING ON HERE AND FROM HERE OVER THE WHOLE WORLD stop The only question is: WHO MARCHES ON WITH TELEVISION, PHILIPS OR THE OTHERS? stop THE OTHERS ARE ALLREADY MARCHING! PHILIPS EINDHOVEN, TAKE THE LEAD ! fullstop''

For all clarity, in his telegram Bouman, when referring to "Television", meant direct viewing picture tube-based receiver sets. Although some involved later disputed the decision was only due to Boumans' telegram, fact is that a rapid switch-over started to steer development activities toward direct view television. This is where the HIG Elektronenbuizen comes in, which fortunately had started work on picture tubes already.
The Protelgram situation didn't improve much, while the television-lobby also started to work against its concept. After 2,5MioHfl investments, a considerable sum at the time, the Philips Board took its losses and decided to stop Protelgram in September 1949. However, concepts can be persistent and we'll see rear projection television coming back later.

The Radiobuizen-Lab

The development of television was tightly linked to the development of new valves. All post-war Radio Valve activities within Philips were concentrated in the HIG Elektronenbuizen (Division Electron Valves) initially led by Dr. Theodoor Tromp (1903-1984), until he became member of the management board of Philips. Also in the Philips Board re reamined the "patron" of the radio valve division. This included all manufacturing plants but also the valve development in Eindhoven. From 1946 onwards this was done in the Radiobuizen Lab, located on the 3rd floor of the "Witte Dame" (White Lady) building along the Emmalaan in Eindhoven. (The building still stands in almost perfect condition, although no longer owned by Philips. It now houses the Eindhoven City Library, the well-known Design Academy and apartments). Originally the activities had been split into two groups, the "Proefafdeling" (Sample Department) which was able to prototype all new valves, and the "Elektronisch Laboratorium" (Electrical Laboratory), responsible for specifying the valves, measuring them and making application reference designs. Since 1946, to improve the up till then notoriously bad co-operation, these two groups were combined into the single Radiobuizen-Lab organization under ir Gerrit Alma (1902-1997). Alma worked with Philips on the development of radio valves since 1929, and was one of the leading and most experienced managers in this domain. He seems to have been a very pleasant person, nice to work for, totally non-political and good for his departments. Under Alma were two dedicated labs: Radiobuizen Lab 1 with a focus on valve definition and their application, and Radiobuizen Lab 2 under ir. G. van Beusekom responsible for the valve design and engineering.

Under Alma the Radiobuizen-Lab continued to grow, hiring engineers that were to play leading roles in the coming developments and the further growth of the valves business. One of them was ir Klaas Rodenhuis, a young Delft graduate who joined Philips in 1941 and would later lead the Professional Valves section of the Radiobuizen-Lab 1, before moving to Hamburg in 1959 to become head of Philips-Valvo. A second was Dr Bert Dammers (1910-1968), who started at the NatLab Research in 1937, almost certainly working in the group of Jonkers. Around 1946 he was asked by Tromp to join the Radiobuizen-Lab 1 to lead the Consumer Valves section in Alma's organization. Dammers was to become the expert in valve-based application circuit design, and at the same time also the person defining combination valves, to optimize the application with the minimum number of valves. Between 1948 and the mid 1950's Dammers published large numbers of application notes, publications and book chapters related to valve applications, especially in the emerging television receiver domain. He also had very good relations with the valve engineering teams in the factories, as well as with the set development organizations of the HIG Apparaten in the many different countries where Philips was active. It was often Dammers who made the road trips to the different countries to explain the latest television valves and application trends to both the internal organization as well as set making customers of the valve sales organization, often talking to large audiences.

[Most of the data on the Radiobuizen-Lab organization in above section is based on the research of my friend Ronald Dekker, also a valve aficionado and creator of the by now famous DOS4ever valve tester.]

Dammers in turn hired young assistants for much of the practical work involved with valve and application measurements. And one of them was my father, Piet Hooijmans (1918-2006). While working in Amsterdam as telegraph operator for the Dutch PTT on the long-distance connections, he had finished a professional training in electronics. Since Philips was expanding and could not find sufficient engineers locally in Noord-Brabant, there was quite some advertising, to which Piet reacted. He was invited to Eindhoven for interviews in February 1947, a very harsh winter and it took quite some effort and time to reach Eindhoven. But apparently the interview was successful, because on May 1st, 1947 he started in the Radiobuizen Lab 1, working for Mr Dammers.

It looks that the Consumer Valves group worked essentially on three applications: radio receivers (mainly AM but with FM reception in development), amplifiers (for microphones, record players and later tape recorders) and television. From the available reports it seems that the application activities were exclusively using the E40 Rimlock series of valves up till end of 1948 or the beginning of 1949, at which moment a new series of Noval E80 and P80 valves was released. The Radiobuizen Lab application notes then also switch to the Noval series. Regularly articles were published in the Electronic Application Bulletin of the HIG Elektronenbuizen, introducing new valves in combination with recommended circuit applications. The consumer group and its valves will play an important story in the coming story on the development of television. In parallel the professional group of Rodenhuis worked on such valves as thyratrons, X-ray, RF transmitters, gas filled switching valves etcetera.

Dr Theodoor Tromp, in 1946-50 head of all Philips Radio Valve activities.

ir Gerrit Alma, head of the Radiobuizen Laboratory.

ir Klaas Rodenhuis, leading the Professional Valves group in the Radiobuizen-Lab.

Dr Bert Dammers, leading the Consumer Valve group of the Radiobuizen-Lab. All above pictures courtesy Ronald Dekker.

Piet Hooijmans, young assistent engineer in the Consumer Valves group of the Radiobuizen-Lab.

The "Witte Dame" (White Lady) building along the Emmasingel in Eindhoven, where the Radiobuizen Laboratory was housed on the 3rd floor. With the exception of the wall around the complex and the square brown building on the right picture, not much has changed on the look of the building, and today it is still as impressive as it was in the 1940's. [Ronald Dekker -DOS4Ever]

New Rimlock valves for television

Since 1946 the HIG Elektronenbuizen had been working hard on the roll-out of a new generation of radio valves: the Rimlock 40-series. Their main advantage was a much smaller size, pre-dominantly due to the new low-temperature all glass assembly process. Two major elements allowed this shrinking: 1. a method was found to have metal leads protrude through a glass footing without loss of vacuum inside the valve. This allowed for much smaller valve foot dimensions, which were standardized in the Rimlock base with eight pins equally spaced on a circle. 2. low temperature glass welding. This allowed the internal electron assembly to be mounted on the inner parts of the pins, where-after the glass encapsulation could be welded on the glass base plate at relatively low temperature without damaging the sensitive electron assembly. Philips used three standardized base dimensions: diameters of 22mm (Series A), 32mm (Series B) and 36mm (Series C). Only the 22mm Series A valves were formally called Rimlock. The first products brought to the market were the the Ux40 valves, using 0,1A heater current. These were pre-dominantly used in the modern small form factor radio sets that could be used in countries with 110V, 220V or even DC supply. The Rimlock series was announced in a Philips Technical Review article in October 1946, by Alma and Prakke. First products were released in 1947.

The three all-glass tube series of Philips introduced in 1947: left Series C (EFF50), Series B (ECH21) and Series A (UCH41). [PTR 10-1946]

The first series of Rimlock valves that were released specifically for radio sets. From left: triode-heptode UCH41, RF/MF pentode UF41, diode-pentode UAF41, power supply rectifier UY41 and and 9W power pentode UL41. [PTR 10-1946]

Where the 1946 PTR publication was a typical research story, focussing on the technical issues of the all-glass valve, the Rimlock series was introduced by the Radiobuizen organisation in publications of their Electronic Application Bulletin (EAB) in October 1946, followed by an updated family overview in December 1948 "Survey of the "MINIWATT" Rimlock E-valves". By this time the family had extended to some 14 E-valves (plus another 10 U-valves), which grouped in a number of application sub-families:
a) RF valves, mainly for radio
     EAF41, ECH41 and EF41
b) other radio valves necessary to make a set
     EL41, a 9W 10mA/V power pentode
     AZ41 (directly heated full wave rectifier, the only Rimlock valve without filament)
It is expressly stated that these E-valves are equivalent to the already existing U-series equivalents.
c) new valves
     ECH42 variable mu oscillator
     EAF42 very similar to the EAF41 but with g3 accessible with a separate pin
     EBC41 diode detector plus 70x triode amplifier
     EB41 for frequency discriminators
     EZ40 90mA full wave rectifier with high filament-cathode isolation
d) new valves for amplifiers, television and measurement equipment
     ECC40 for push-pull amplifiers, 2,7mA/V, 30x amplification
     EF40 1,85mA/V LF amplifier with sharp cut-off
     EF42 high slope 9,5mA/V especially for television receivers
e) battery supply power pentode
     EL42 6W pentode with only 200mW filament current

All these valves (either as E- or U-valve) with the exception of the EL42 will appear in the television designs of the period. Two non-Rimlock valves that were used additionally were the ECH21 triode-heptode (which because of its 9 electrodes did not fit in the Rimlock series A base, and used the series B base) and the EY51 high voltage rectifier diode. The EY51 was used inside the high voltage supply for the picture tube anode and didn't have a classical base. At the bottom two wires (cathode and heater) and at the top one wire (the anode) protruded from the glass bulb, which could be soldered with very short contacts.

The evlution of the pentode. From left to right the AF2 (1934), AF3 (1935), EF5 (1936) and EF41 (1947). [Radiomuseum]

The EA40 high voltage efficiency diode, one of the very first Rimlock valves.

The EY51 high voltage rectifier, without a classical base but only metal wires protruding from the glass bulb.

The MW22 picture tube

The last, but very important valve, was the picture tube. So although the majority of the Philips television developments was using the MW6-2 high intensity projection tube, the Elektronenbuizen division had been working on a direct view larger picture tube. Surprisingly this development had started already well before the war! As it seems Mullard, the London-based UK subsidiary of Philips which was acquired in 1927, developed a first 9"/22cm picture tube as early as 1938. This tube, the MW22-1, was used in the 405-line UK television system, both the Philips models 2405/07/12 from 1939 as well as e.g. the Pye televisions models 815 and 838. Mullard was sailing a quite independent course from the Eindhoven headquarters, and in contrast to the formal doctrine on projection television believed much more in the direct view concept. This concept was industrialized by RCA together with the glass-making company Corning. They had invented the concept where a metal cone could be welded to the curved glass front section. This required two technology steps: glass moulding and the glass-metal connection. Mullard probably licensed the technology and bought the sub-components (the moulded glass front plate and the metal cone) for assembly in its own valve factory.

Essentially there are two generations of MW22. The first one was the pre-war version, with a "hexode" electron gun assembly and a 5kV anode voltage. This assembly required a fairly high grid voltage of -60V (maximum light intensity) to -80V (black). The grid capacitance was still fairly high at 13pF, limiting the bandwidth of the video signal.
The second generation, developed after the war, used the same glass balloon shape and dimensions, but had a re-designed internal constructions, of which apparently 2 existed. The MW22-3 had a "triode" electron gun like the MW6-2 projection picture tube, and although this tube was used in at least one UK television set in 1950, this doesn't seem to have become the standard solution. That was the concept as in the MW22-7. It had an improved electron gun assembly with less than half the grid capacitance (6pF), a much lower gate driving voltage (-11 to -35V) and higher brightness due to a higher anode voltage (7kV nominally, 9kV max). It used the same parallel 6,3V heater supply as the other Ex40 Rimlock valves.

The Philips Miniwatt MW22-3 as it was shown in the official specification leaflet.

The MW22-7 became available in 1947, and for the coming 2 years would become the standard picture tube in all Philips direct view television development sets. In parallel it was of course sold in the UK under the Mullard brand and used in multiple television sets of Ferguson (model 841T), Bush (model T91) and Pye (B16T and D16T), mostly sets introduced in 1948-49. Where the MW22-7 had the parallel 6,3V heater supply of the Ex40 series, the MW22-14 was the same tube with 0,3A serial heater supply equivalent to the Ux40 series.

In parallel to the MW22 series also the MW31 was developed, with a 64 degrees deflection angle and a screen width of 31cm (12"). The story on its development, sub versions and availability is almost identical. Because the 31cm tube was certainly more difficult to produce, its actual application lags the MW22 by roughly a year.

With this picture tube Philips now had all valves available to design direct view television receivers.

TX380, television sets based on the new Rimlock valves, 1948

At this time, we write end of 1947 to early 1948, matters become more complex. Philips is rolling out its limited scale NatLab-based 567-line experimental system, Protelgram is being introduced in the US, the UK television roll-out is picking up speed and needs Philips solutions, and in France Philips also wants to have sets that can receive the Eiffel Tower 819-line transmissions. After the initial work by the NatLab research, the centre of gravity needs to move to the HIG's. Two organizations therefore scale up television set development:

HIG Apparaten, which focuses on actual television receiver development. Central television development is in the Apparaten Lab (from here on AL, where "Apparaten" can best be translated by Devices) on the Strijp complex in Eindhoven, but in the big countries local development groups are active too. Especially in the UK, where the Philips subsidiary Mullard had a television development and production site in Micham. The official policy was that the platforms should come from Eindhoven, for local industrialization in the satellites. Not surprisingly the local "satellites" didn't like that concept.
HIG Elektronenbuizen, which focuses on reference designs of the new valves that are being developed and released. These developments, both of the valves and the reference designs, take place in the Radiobuizen Laboratorium (RBL) on the Emmasingel in Eindhoven. This organisation doesn't suffer from the multi-site structure of the HIG Apparaten, but in turn doesn't produce sets themselves, so always needs to "sell" their concepts to HIG Apparaten or third party customers. Or phrased more bluntly: the reference designs of the Radiobuizen Lab are not necessarily taken into production.

In these two organizations essentially parallel tracks of television chassis developments took place, with a lot of exchange for system optimization. It is known that Bert Dammers, the head of the Consumer Valves group of the Radiobuizen Lab (the boss of my father) had a very good relation with the Apparaten Lab groups, and this resulted in regular introductions of circuit optimizations based on new valves. In the following I've tried to re-trace the main television platform developments up to the first volume sets. This is based on the use of valves and their functional circuits, on the naming of the products and on the date of publication of circuit diagrams or service documentation. Especially the Philips product naming convention is very helpful, since it indicates the year of design (see below explanation of the product codes).

It must have been around 1947 that real television product development started in earnest. As we will see later, it is highly likely that the Radiobuizen Lab and the Apparaten Lab really combined all their knowledge to define a state-of-the-art television platform based as much as possible on the latest and most compact valves available. Here the first subtleties came in.

The HIG Apparaten logically wanted to have product platforms that could be used in as many countries as possible. This involved flexibility in not only the television standards, but also the power supply, which still varied across countries between 110 and 220V, 25 and 50Hz and AC or DC. To create this flexibility the use of power supply transformers was avoided whenever possible, which meant that the 6,3V heater supply for E-series valves was not available either. As a consequence all AL-designed sets used U-series valves for serial heater connection at 0,1A current. So the Apparaten Lab set itself to designing a TV platform based on the valves they were using already for radios: the UF42, UAF42, UB41, UL41, UCH42 and UY41. The only dedicated television valves were the ubiquitous EA40 efficiency diode, the UL44 line/frame time base output amplifier and the EY51 high voltage rectifier.

The RBL in contrast focused on the application development, and although they made reference designs of complete sets they didn't focus on the power supply. In fact in the RBL they used standard power supply bench units supplying 250V and 6,3V, and the prototype units did not contain the power supply. See e.g. the reference design AM/FM receiver on the EQ40-page. As a consequence RBL reference designs always were based on E-series (6,3V) parallel heater supply valves. These designs were essentially the latest television design but with new valves inserted for performance optimization or valve count reduction. Examples are the EQ40, ECC40, EZ40 and EL43. It will be interesting to see which of the new valves of the RBL actually ended up in production televisions.

The naming convention of Philips Television receivers 1945-1950

Starting from the second half of the 1940's the television receivers of Philips were coded as follows: TX123A/00
1st letter: always T for Televisie (Television)
2nd letter: country code of the production (!) facility

X=Netherlands & Belgium
D= Duitsland (Germany)
F= France
G = Great Britain
I= Italy
W= USA

1st number: luxury level

This was roughly
1 /2= low end, 3/4= standard, 5/6=luxury, 7/8= high end, often multi-function, 9 = top models

2nd number: year of development

the last digit of the year of development (not commercial introduction!); so e.g. 1947 gave 7

3rd number: product identifier

first product was 0, if successive products were developed in the same year they obtained the next number

3rd letter: power supply

multiple options, including batteries and accumulators, but for early televisions the two relevant ones are:
A = mains AC supply only
U = mains AC and DC possible

Suffix (two digits)

many different combinations. On the early sets these are seen:
00 = the basic first release
01, 02, ... = set upgrades/retrofits
29 = French version, 68= Argentinian version etc.

This coding system survived for most products till around 1966 (!) with only few modifications/additions to accommodate the growing product diversity
1. When codes threatened to repeat (esp. due to the design year number after one decade) the 2nd letter and 1st digit were swapped. This happened mostly with Radios.
2. Not surprisingly the UK did not comply with the system. Originally the letters TG were omitted (e.g. 383A being the first UK TV set), later Mullard used its own coding MTSxxx.

Although started in 1947, the first television designs were ready in 1948, and thus the first generation televisions was coded as TX380 (with the 8 referring to product release in 1948). It is interesting to note that according to the product naming system the first digit 3 indicates it was positioned as a standard product, although given the price (85.000 old French Francs at the time) it could be considered a quite expensive product. All the more so because the product leaflet suggests the set also contained a full three-band radio receiver. The most logical explanation is that the screen size of 9" was already considered on the low side. For actual market introduction the most urgent version was for Great Britain (405-line system), the next France (still the 441-line system transmitting in Paris) and next the support of the fledgling Dutch (567-line system). In this period the number of design iterations must have been high, because the sets TX380, TX381 and TX382 are unknown, if they ever existed. The first product released was the 383A for the UK, followed by the TF384A for France. Of these the 383A has already been discussed since it was not based on an RBL reference design but on the pre-war 1938 Tel61. The TF384 seems to have been the first and the closest to the reference design, so we'll have a look at it first. Where the reference is a 567-line set design of RBL dated 23-8-1948. The RBL design will be discussed in more detail later, but is useful here for the modules it has in common with the TF384U.

Block diagram of the Philips TF384A television set for the French 441-line standard. Designed 1948, on the market 1949.

The valve specification of the TF384A comes from a leaflet on Radiomuseum.org, with the remark that almost certainly this only refers to replaceable valves, and the two EY51's inside the high voltage unit are thus not mentioned. Twelve of the twenty valves are 1-to-1 compatible to the RBL reference design (except for the U vs. E-series difference), making it highly likely this is the design used. In the RBL design the video output amplifier is replaced by the EL43, a first design of television-specific power pentode. This valve would not make it to volume production, however, in turn being replaced by the EL/UL44. Other new valves included the ECC40 dual triode for both the frame and line time base generators, of course the EQ40 phi-detector for the FM detection. It also introduced the EZ40 dual phase power supply rectifier.
The RF-IF-video chain is thus common between the two (apart from the EL43 for the UL41), and shown below. At least for the first generations one clearly tried to avoid having to handle (especially amplify) the RF signal. The RF input is thus directly fed into the hexode section of the ECH42, which acts as a mixer, the triode section acting as oscillator. Tuning is limited and the set is designed for only a single RF channel reception! For the TF384A this was the 42-46MHz channel as used for the Eiffel Tower transmissions. Because on this transmission the sound carrier was below the picture carrier the local oscillator had to be below the RF too, in order to guarantee a sound IF below the picture IF, a quite unusual concept.

The Philips 383A set for the UK (405-lines) as sold in 1949. A 9" screen and a basic table-top cabinet. [Philips Service documentation]

Drawing of the TF384A, showing the radio receiver dial plate at the bottom. [Philips leaflet via Radiomuseum.org]

From right to left the RF, IF, video detector, video amplifier and sync separator chain of the early 1948 Philips television platform, as used in the TF384A and as the basis for RBL1 reference design. At the output of the 2nd EF42 the sound IF is tapped off towards the sound IF section. [Legacy Piet Hooijmans]

The UK situation was a more complex one for multiple reasons. To start, the British television transmissions were well under way since 1946, with multiple vendors selling television set, e.g. Bush, Invicta, Philco and especially Pye. The latter company had aggressive expansion plans towards the continent, and Philips clearly had some catching up to do. At the same time the Philips UK subsidiary Mullard had been active throughout the war, having both production facilities and development labs fully staffed and running. There was a clear animosity between the Mullard Lab in London and the Apparaten Lab in Eindhoven, each considering themselves better suited for UK television set design. Mullard because they worked locally and had all insight into and availability of the British 405-line standard. The AL because they wanted all sets to be based on their reference platform, for maximum synergy across the countries. Ultimately the "official" approach with the central role of the Apparaten Lab gained. So where the 463-563-663-383 platform from 1947-8 was still largely designed by Mullard Labs based on their favourite valves and the concepts of the UK televisions from before the war, from now on the UK sets would comply with the formal platforms, albeit with the local adaptation to the 405-line AM standard.

TX390 upgraded platform, 1949

The TX380 platform, being the first, undoubtedly suffered from some infant diseases. One of them was almost certainly receiver sensitivity, with the antenna signal - after some lossy bandpass filtering - was directly fed into the mixer with its conversion loss. This is known to give bad sensitivity due to a large noise contribution of the mixer input. It is then no surprise that the TX390 used a new RF input concept, with the UCH41 mixer-oscillator being replaced by a first UF42 for RF amplification and a second UF42 acting as a self-oscillating mixer.

The new RF-IF line-up of the TX390, with B9 an UF42 pre-amplifier and B10 an UF42 self-oscillating mixer-oscillator. At the output of th mixer the sound-IF is immediately separated through S14-C69, thus reducing the bandwidth requirements for all following stages. This architecture would be the bases for the next two years.

Block diagram of the TF390 441-line set for France. Designed 1948, commercially launched 1949. First implementation of the new RF input stage.

Other optimizations concerned the two time base oscillators, where the larger UCH21 was replaced by the new smaller UCH42. The second UL44 in the line output stage was deleted, while the power supply rectifier was reduced to a single AZ50. Overall a saving of 2 valves. Furthermore the sound chain - for AM detection - was reduced to a very basic design, with just three valves of which one diode was not even used.

The audio chain of the TX/TF390 for standards with, AM sound. B5 and B6 are UAF42's, B8 an UL41 power pentode. Note that the diode of B5 is not used. The diode of B6 takes care of AM peak detection. The detected signal is, after volume control with R24, fed to the pentode section of B6.

The French model TF390A with its 9" screen. This set was first produced in Eindhoven as the TX390/29 and exported to France, but then transferred to Suresnes for local production as TF390A. [Musée du Radio RTF]

TX594, the first 567-line set, 1948

The next challenge was to develop from the same chassis a Dutch (and of course potentially European) version for the 567-line standard, with its higher line rate, higher bandwidths and FM sound standard. The official documentation has a first release date 24-7-1948, so design must have started end of 1947 and the design clearly based on the TX380 platform, picking up some improvements of the TX390. However, although we know there was a design in July 1948, we don't have the details on it (yet), only those of a much later 1949 modification that will be discussed further down. The block diagram can thus only be speculated, but would most probably look as follows:

The block diagram of the first release of the TX594U for reception of the 567-line experimental PET transmissions. This block diagram is not 100% confirmed, but based on comparative analysis with TX380 and TX390 platforms adapted for FM detection.

Most remarkable is the fact that the FM sound detector does not use the EQ40 as recommended by the Radiobuizen Lab, but a classical Foster-Seeley detector using an UB41 double diode. This requires an additional UF42 sound amplifier before the power amplifier. Secondly the line time base amplifier seems to use a dual UL44 self-oscillating output stage, which seems a rather expensive concept. With three power rectifiers UY41 the total number of valves came to 23, the highest number of any direct view television set of the first generations. Around 150 pieces were built of the TX594U and distributed throughout Eindhoven, probably replacing most of the SX861A research sets that were called back. A number of sets might have gone to Denmark too, which was the only country supporting the 567-line standard. Philips had also supported local test transmissions in Denmark. For reasons not entirely clear, the set, which was of the same design as the earlier TF384A, received the nickname "Hondje" (Doggie). Nevertheless, even despite the low volume, the TX594U will be in the annals of history as the first (almost) commercial television set in The Netherlands when it was launched end of 1948.

Philips TX594U, receiver for the 567-line Philips standard, as produced end of 1948 for reception of the Philips Experimental Transmission TV. Note the rather unusual reclined position of the picture tube. [Technisch Museum]

Backside of the TX594U. Note the high quality polished wooden housing, standard for TV's of the time. [Beeld en Geluid]

The last sets of the first generation, 1949

During 1948 and 1949 the TX390 platform was used as the basis for a number of last new products: the 385U for the British market, and the TF402 for the French market. They had both 19 valves, and were based on the same platform architecture with only minor valve upgrades. The most remarkable of those was the replacement of the good old EA40 efficiency diode by the UY41, a more powerful valve that had already been used as power supply rectifier. In the British sets the UL41 power pentode was replaced by the UL44 (for sound) and for the frame output the UL46, a short-lived valve that was only used in the UK. Like in the TX594U the UL41 video output amplifier in the TX402A was replaced by a lower power UF42.

One of the landmarks of these sets was the introduction of a new 12"/31cm picture tube: the MW31-14 or -17, giving an almost twice larger image. These tubes had the same characteristics as their MW22 predecessors, and didn't require any electronics changes. The 485U and TX402A both used the 31cm tube, later followed by the 683U.

And this in turn illustrates a next step in the growing maturity of the product: based on the same platform multiple models with different selling prices were introduced, especially in the British market where volumes were highest. In 1949 the 385U (mid end table top 9") and 683U (luxury console 9", also sold as the Mullard MTS684) were introduced, along with the 484U (mid end table top 12"). In 1950 the 492U followed (mid end table top 12"). In relation to the British market it should also be noted that in December 1949 the second transmitter station was activated at Sutton Coldfield near Birmingham, serving the Midlands. While Alexandra Palace in London was transmitting at 45MHz (picture, sound at 41,5MHz), the Birmingham transmissions were at 61,75/58,25MHz. Since the sets still did not contain a tuner, each television needed conversion when receiving the other transmitter, and extensive change-over and alignment procedures are part of the Service Manual.

With these products the TX390 platform came to the end of its (design) life, and no further derivatives were developed. Started in 1947 it had served its purpose, releasing the first sets in 1948 and establishing Philips as an emerging television player in the UK (through its Mullard branch), France and even The Netherlands. All television receivers launched in 1948-50 were thus based on this platform.

The interior of the Philips 385U 405-line set for the British market. On the right side in the back is the oil-filled high voltage unit with the two EY51's horizontally mounted on top of it. [Vintage Radio Forum]

The TF402A, the French version of the chassis and one of the first with a 12"/31cm screen. [Antique-TV Blazianu]

Block diagram of the Philips 385U television set for the UK and the TF402A for France. Design from 1948, introduced 1949 and some models 1950. Almost all valves are identical, the differences for the TF402A are in italics. The sets were still single-channel without a tuner, so they had to be adapted for either London or Birmingham reception.

The variants of the 1949/50 UK television sets based on the 385U platform. On the left the 9"table top 385U, centre the table top 12"492U (identical to the TF402A shown above) and right the 485U 12"console model. [Philips Service Manual]

Activities at the Radiobuizen Lab, 1947-49

While their colleagues at the Apparaten Lab were busy rolling out the TX380 and TX390 platforms and their British and French derivatives, the Radiobuizen Lab worked on the application of new valves. For this they used the latest AL platform as the starting point, based on which they would then change specific circuit blocks using the new valve. As mentioned when introducing the RBL they used - at least initially - exclusively the6,3V E-version of the Rimlock family.
The primary role of the consumer group (Dammers) in the RBL was to promote the application of the new Rimlock valves in consumer applications, which meant first and foremost radio, (record player) amplifiers and then the emerging television. As mentioned when introducing the Rimlock series the initial publication on the technical concept of the Rimlock all-glass base was in the research Philips Technical Review in 1946, but in the Electronic Application Bulletin (EAB) of 1949 the series was introduced on a practical application level. And since then there was a regular stream of specific valve introductions. The first one of these was on the EA40 efficiency diode for the line output amplifier and high voltage supply (EAB Vol.10 No.1 December 1948). The EA40 was in that sense successful, since used without exception in all first generation Philips television designs. The same was true for the EF42 and EY51, both designed specifically for television applications.

Normally it would be extremely difficult to trace back the activities of an organization like the RBL, but luckily I have three sets of schematics of television sets in the legacy of my father. These were the basis for the home-built televisions we will discuss in the next section. I have three of these schematics which I will call RBL1 to 3. They relate to the AL product designs as follows:

RBL1
This is a schematic dated 23-8-1948, signed by P. Huijser and titled "Schema van Televisie-ontvanger" (Schematic of television receiver). The concept is clearly based on the TX380, which was being developed at the same time.
Four different new valves were applied in this design:
- the EL43 power pentode, replacing the UL41. This valve never made it to volume production, and was superseded by the EL/UL44.
- the ECC40 dual triode, used in the frame and line time base oscillators, replacing the large and expensive UCH21.
- the EZ40 dual phase full wave supply rectifier. For this story the same holds as for the ECC40: a long career, but not in television sets. Cheap receivers would use a single EY/UY half wave rectifier, sets with full wave rectification mostly used the older but more powerful AZ50, which could handle as much current as three EZ40's.
- the EQ40 FM phi detector. This was clearly the most successful exercise of the RBL to get its reference designs adopted by the AL. In parallel to the RBL1 design the AL was working on the TX594U, their first FM receiver television set. However, the result - 6 valves - must have been considered disappointing, because the subsequent platform used the EQ40 design from the RBL. All details on the development and application of the EQ40 can be read here.

The most interesting from this list is the ECC40 as used in time base oscillators. In the same EAB edition as on the Rimlock series and the EA40 an article by J. Jager of the RBL discusses this double triode in time base oscillators with linearisation for feedback. One of the possible implementations in this article for a frame time base oscillator uses a transformer for feedback in what is called an intermittent or "squegging" oscillator, and invention by - who else - Bert Dammers, published in EAB of April 1946. This circuit is exactly identical to the one in the RBL1 reference design. Despite the RBL pushing the ECC40 in this role, the Apparaten Lab apparently didn't follow them because they would move to the UCH42 instead in a blocking oscillator arrangement. The ECC40 had much more success in the amplifier and radio application, where at this time my father was working on. See e.g. the amplifier reference design report from Bert Dammers and Piet Hooijmans shown here, where the ECC40 is used as phase inverter in the single-to-balanced converter of a balanced push pull power stage with two EL41's. An article on this was published in EAB of March 1949. The ECC40 had a long and successful career of 10 years, including its professional variant E80CC. Interestingly there never was a UCC40 version.

The frame time base oscillator from the RBL1 circuit diagram, using a squegging oscillator with ECC40, the concept of which is hown in the right picture from the Electronic Application Bulletin. [EAB December 1948]

RBL2
This schematic is originally of exactly the same date and authorization as RBL1, but updated 3-3-1949.
It clearly bears resemblance with the TX390 platform, using the new two-stage RF input without the ECH42. And it also copied the synchronization separator from that design, replacing the ECH21 by the much cheaper EF40 (which is the only use of this valve in any of the schematics). The power supply also copied the powerful AZ50, replacing the three EZ40 power supply rectifiers.
Both the frame and line time base oscillators are still based on the ECC40, although the frame time base oscillator has been replaced by a simpler and more classical blocking oscillator. Although in the EAB article of Jager the "squegging" oscillator was presented as an improvement of the blocking oscillators, practical considerations have apparently led to falling back on the older and simpler design.

Given the date, early 1949, I originally considered the RBL2 as the then optimized design of a 567-line Philips-standard television, with 4,5MHz picture-to-sound carrier bandwidth and FM sound. With 21 valves it was an efficient design compared to the contemporary TX594U of the AL with its record 23 valves. Also compared to the TX390 for 405 and 441lines, the penalty for higher bandwidth and FM sound of only two valves should have been acceptable. It is then not unlogical to assume that this design would have been the basis for a real 567-line product development of the TX594U successor. However, as we will see next, the RBL2 design was almost certainly something else.

RBL2, the ultimate reference design for a 567-line television receiver based on Rimlock E-valves. [Legacy Piet Hooijmans]

RBL3
This an undated and unsigned schematic.
It is a receiver using the MW22-14, dating it to the 2nd half of 1948. It is - with the exception of the EQ40 - entirely based on U-series valves and has indeed most in common with the contemporary 385U and TF402A developments. It seems an effort to design a TV with minimal number of valves - 15 - which makes it a typical RBL activity.
This was achieved by a number of design tricks: one UF42 IF amplifier less than normal; no sound IF amplifiers; the triode section of the UCH21 used as peak detector; no separate line time base oscillator and only one EY51 booster. It all seems a bit too much squeezed out, and I have some doubts whether this set would have functioned properly. However interesting, I have no further traces of its application.

The Radiobuizen Lab RBL3 television receiver reference design with only 15 valves.

Based on their activities the HIG Elektronenbuizen published a continuous stream of application articles in their Electronic Application Bulletin. Most of these were a combination of introducing a new (Rimlock) valve and circuit optimization. The table below gives the overview, which shows that all new valves in the core of the RBL television reference designs were covered with a dedicated article: the EA40, ECC40, EF42 and of course the EQ40. Interestingly articles are not only from the Eindhoven RBL, but also from Mullard (Mr. Coxall) and the new Paris Lab (Mr. Chambes).

Television-related publications by engineers of the Philips Radiobuizen Lab in the Electronic Application Bulletin (EAB) of the HIG Radio Valves, 1949.

Test equipment

It is a bit of an excursion from the main story on television development, but it is nevertheless interesting to have a brief look at a side activity in the Radiobuizen Lab. Today it is hard to imagine with how minimal means the development teams of the day had to work. Only the most basic frequency generators, voltmeters and the oscilloscope were available. In many cases television sets could only be measured integrally using signals received off-air. Therefore it was e.g. very difficult to optimize a 525-line UK set in Eindhoven because there was no reception of British broadcast signals. Only around 1950 video test pattern generators appeared. Also in the Philips Electronic Application Bulletin regularly articles were published about new measurement equipment for the demanding new applications and circuitry. For example in EAB June 1950 was an article on "A simple oscilloscope for unit function response testing of networks", which was used to measure amplifier characteristics, ringing and filter responses. In later EAB articles measurements were referred to this equipment. Similarly in December "An instrument for recording the frequency drift of an oscillator" was presented. But these were only modest instruments compared to what came next; the valve tester and curve tracer.

Up till then valve testers were essentially voltage sources that could be applied to valve contacts to measure the anode current; nothing sophisticated. Operation was manual and analysing a valve took many readings. In order to speed up the analysis of the many new valves that were being developed by the Elektronenbuizen division, Dammers and his RBL team developed what looks like the first and most advanced automated curve tracer valve tester.

An article about this machine was published in the Philips Technical Review of April 1951, "The Electrical Recording of Diagrams with a Calibrated System of Coordinates" by Bert Dammers, van der Knaap and Uitjens, all three from the Radiobuizen Lab. For its time this was an incredible piece of test equipment, showing simultaneously on a cathode ray display multiple traces with fixed Vg steps, a calibrate horizontal and vertical grid and a movable cross-hair pointer to measure the exact voltages on specific points of a trace. It contained a stepped-pyramid voltage generator for the grid, an advanced timer assembly to time-interleave the different signals to the deflection plates and of course the sweepers for the curve tracer. The total device contained a stunning 200+ valves, making it almost a analogue mini computer avant la lettre. It is a clear illustration of how creative Dammers and his Radiobuizen Lab team were.

Front view of the curve tracer and valve tester. The engineer is looking to the cathode ray screen through a viewing cap. The large displays above are indicators for the cross-hair curve analysis. Each module was self-supporting, with its own power supply for easy replacement. [All pictures PTR 4-51 p.283]

Screen shots of the curve tracer while measuring the pentode section (left and center) and triod section (right) of an ECL80. All visible lines are displayed simultaneously, only the text is later added to the pictures.

Home-built television sets

At this point we come to one of the most intriguing elements on this saga of early television development, and the original trigger for me diving into this: the home built television sets. As already explained, from March 1948 the Philips Experimentele Televisie (PET) was transmitting in Eindhoven, albeit for a very limited number of watchers. Originally some 20 SX861A research sets were handed out to dignitaries, towards the end of 1949 replaced or supplemented by some 150 TX594U's. As soon as the PET transmissions started amateurs all over the country started to build experimental home-built receivers, where it is not unlikely that quite some Philips engineers tried the same. Although it is so far not possible to trace back the decision, apparently the Philips management decided to allow some of its employees to build a television receiver at home, based on Philips components. There are strong indications this initiative originated from the Radiobuizen Lab, although it might later have allowed people from other organizations (Apparaten Lab and Research notably) to join. Piet Hooijmans, young RBL employee was one of them! Because the RBL1 schematic is the oldest in his files, I assume that is the one he started building on, which means the original set was built second half of 1948.

The home built activities were organized as follows. For efficiency reasons it was decided that everybody would work from the same reference designs; RBL1 as introduced above. Once an employee was allowed into the program he (there are no traces of she's in this story) would receive all material from Philips to build a TV according the design. This included, in order of material cost, all valves including the picture tube, the transformers and speaker, the passive components and the metal frame. Knowing the Philips culture, it was clearly not the intention that people worked on their TV during working hours, but after-work activities in the office were allowed. Most people - like my father -were unlikely to have all or any measurement equipment at home, and were thus forced to use the lab equipment. At least for Piet this shouldn't have been a problem, because he was still unmarried and - according to him - there was anyway nothing to do in post-war Eindhoven. The most complex part of the construction was the high voltage transformer, which was housed inside a metal can that was filled with oil (for isolation) and then soldered close. This was allowed to be done by the workshops during working hours. One interesting fact is the power supply transformer, number 7 in below picture. This is an extremely heavy component, in fact it creates quite some torsion in the entire set frame when lifted. This transformer was required in view of the 6,3V heater supply but also provided all other voltages. In Piet's notes it says "voedingstransformator (supply transformer) A3 161 21 app. SG860a" which suggests this is the same transformer as used in the SG860A receivers. An unexpected historical link!

No two home-built television receivers are of course identical. This one has a 31cm picture tube, a large HV cage covering the entire deflection unit and also the layout is different. [Private collection Jac Janssen]

Rear side interior view of the home-built television of my father Piet Hooijmans. Note the very high ressemblence with the official picture. [Private collection Pieter Hooijmans]

However, like so many activities started with the best intentions, also this one got out of hand. On March 2, 1949 Mr Alma, the head of the Radiobuizen Lab, issues below memorandum

                                                                               Radiobuizen Lab 1
                                                                               March 2, 1949

Announcement concerning the home building of television devices

Lately, some drawbacks have become clear related to the do-it-yourself construction of television devices.
A lot of working hours are spent on asking for and the giving of technical advice.
To avoid these problems in the future, it has been agreed that from on this technical advice will only be provided at defined time slots by defined experts.
The following gentlemen have declared themselves willing to act as so-called "television doctors":
Dammers
v.d. Knaap
Uitjens                       Tuesday and Friday 17.00- 17.30
Huyser
Cock
Everyone participating in the do-it-yourself project can address any of these doctors at the prescribed times for advice.
Furthermore it seems desirable to limit the period allowed for the construction of a device, e.g. half a year.
Only in exceptional cases it will be allowed to deviate from this, e.g. due to problems in component supply or illness.
Because the NV [Philips] has already supplied a considerable amount of material for the home built televisions, from now on, both for those already in the program but need new components for an extension, as well as for those who will start for the first time, permission will be required from undersigned.
Furthermore we beg the do-it-yourself people for considering not to disturb the workshops for trivialities like bolts and wires.
           In general, from now on no longer materials will be supplied for the construction of antennas, this will be limited to giving advice.

                                                                G. Alma

Draconic measures! The distribution list contains 36 names, including a number of managers. Although the majority has not been identified, they seem mostly from the RBL Fifth name from the bottom is NatLab group leader Haantjes, which might suggest some four researchers had joined the program. Al in all it seems some 25 people were already participating in the do-it-yourself television program. Including Piet Hooijmans on the fourth line (here erroneously written as Hooymans).

But apparently the first round of measures was not sufficient to get a grip on the program, because two months later Alma issued a second memo:

                                                                                      Radiobuizen Lab 1
                                                                                      May 11, 1949
                                                                                      Dr.A/EM

Additional announcement concerning the do-it-yourself construction of television devices.

From now on, as a rule only one new member will be admitted to the program each month.
The finished and under construction devices remain property of the NV [Philips].
Upon departure of a program member from the NV, the device should be handed in.
The program member shall, if necessary, allow the device to be at the disposal of the company for half of the time.
The device can be given on loan to the member for a period of maximum two years after completion; thereafter the member can opt to buy the device at a to-be-defined price.
Servicing of the device will be at the expense of the department.
Also for serious modifications of the device permission is required.

Eindhoven, May 11, 1949
G. Alma

It is beyond the scope of this page to dive into the details of the home built TV. In any case Piet Hooijmans finished its construction and got it all working, most probably towards the end of 1948. But he had it working and was allowed to take it home, where he invited two of his sisters to come over and watch the PET transmissions. When he left the Radiobuizen Lab early 1950 he was apparently allowed to keep it, whether he had to pay money as per point 5. of the last memo I don't know. But 67 years later the set is still in my possession.

What happened further with the program I don't know. But given the memo's it is not hard to imagine that it didn't live too long, and was stopped ultimately the introduction of operational TV reception in 1951.

Piet Hooijmans' home-built television set 1948-1949. [Private collection Pieter Hooijmans]

Continue reading Part.2 covering the years 1950-1955

Piet's Home-built TVorHow Philips developed Televisionpt.1 1946-1949