HMV 904 Replica

  The following is by Victor Barker, who has built this very nice replica of a prewar HMV 904:

    It was 1958 when I was a teenager and living with my parents near London that I was given an HMV 904 T.V. to experiment with. I managed to get it working by repairing the vertical timebase and replacing a couple of valves, the mixer and the sync limiter. Unfortunately my father was conned into selling it to a associate of his, much to my displeasure. In those days it was the parents who were in charge. I was at the time apprenticed to an electronic instrument manufacturer making R.F. test gear for the BBC amongst others for T.V. transmitting stations and receiver manufacturers. I was very interested in the technique of television engineering and the bug is still with me.

   Having moved to Australia during 1968 I had little chance of finding any old British 405 line T.V. receivers to form a collection but as the years went by I was able to find a few early post war sets when visiting my family in the U/K but never an HMV 904.

   About 7 years ago I received an email from a Doctor in Queensland, Hugo Holden who had seen some articles I had posted on the internet describing various aspects of early television broadcasting and engineering, he told me about an HMV 904 that he had totaly restored, that was followed by a visit to his home in Queensland. I was immediately fired up with the desire to either obtain such a set or build one. I am not sure what would have been the best choice but I decided to start gathering parts to build one. The build philosophy was to copy the original circuit as far as possible using components that would not add or detract from the original design performance, this entailed using valves (tubes) of similar performance and preferably made or in common use before the WW2. Rectifiers and audio valves would not present a problem, however R.F. I.F. and video types were a problem.

   It would have been possible to locate a set of original valves with difficulty, perhaps even a set of spares but with little guarantee of their condition. I therefore decided to use valves designed pre war and made during the war by the thousand for military use. I chose the British SP61/VR65 a direct copy of the SP41 the only difference being that the SP61 employed a 6.3V heater rather than a 4 V heater. The valve had a higher gain than the original Marconi MSP4 employed as R.F. amplifier but a lower gain than the three KTZ41s used as I.F. amplifiers, the MSB4 active video detector also has a lower gain than the SP61. The total conductances in practice with this set using SP61s total about the same as the original valve complement and the performance of the finished set is similar to the original 904.

   All the octal valves were as in the original or their direct equivalents whilst the pulse diode a D42 was changed for an EB34 double diode, the two sections being wired in parallel. The X41 with its B7 pin base has been retained. The KT41 sound output valve has been changed to an octal 6V6. The rectifiers are as original. In all cases where required slight changes to cathode bias resistors have been permitted, these have been minor and few in number to accommodate the different slopes of the SP61s in three cases and the vertical output valve required a different bias resistor for another reason as will be explained later in the article.

   The original CRT was a major stumbling block because of the near if not total impossibility of obtaining one, various 5 inch tubes were tried but only two types were in any way acceptable. The tubes were both of American origin being the 5AXP4 a non aluminised electrostatic test tube and a 5FP4 a magnetic focused white screen tube very similar to the very common 5FP7 but designed for TV camera viewfinder use. The latter tube was the type finally used, it performs flawlessly and even at the very low E.H.T. of 2.5 KV employed in the 904 design is not the limiting factor determining the picture resolution. The brightness is adequate under subdued lighting as with the original set using an EMISCOPE 3/1.

   At the time of construction access to any coil winding information was not available and still isn't I used valve data, circuit fixed capacitances and guestimates to determine the total circuit capacitances in order to design the R.F. and I.F. coils. Some degree of empiric determination was used in a few areas. Aluminium tuning cores were used where variable permeability was required.

   The tuning scale mechanism employing an epicyclic drive was engineered to the same spec as the original set, the logging dial is somewhat simpler at present but designed to be modifiable if I can find suitable parts to replicate the original chain drive. The dial was made from a photograph of an original dial that was re- drawn using cad program on a P.C. and finally printed onto photographic paper and used as a rear dial plate rather than printing onto the rear of the glass frontplate.

   All laminated transformers were hand wound or modified types except for the audio output transformer, it was a standard 4500/4 Ohm radio output transformer to suit a 6V6. The focus coil is entirely home wound whilst the present deflection coil assy is a commercially made unit.

   All nut and bolts used on the T.V chassis are zinc plated steel cheese head 2, 4 and 6B.A. types, whilst the screws used in the cabinet construction are counter sunk brass types.

   Most of the wiring unfortunately is carried out using plastic covered wire, whilst many of the short interconnects are wired with 22 s.w.g. t.c.w. and sleeved with mil grade black cotton impregnated sleeving removed from wartime radar equipment of British design. The tagboards are of wartime vintage and also British, they are almost identical to the ones used in the original 904, they appear to be made by the same company. How lucky can one be?
 

   The tuning capacitor was removed from an old Australian radio but was made by the same company that made the 904 unit even the direction of rotation and the fixing holes I have subsequently discovered are the same. Different trimmers were fitted to accommodate local requirements. The multiple electro housed in a rectangular metal box was faithfully reproduced using modern electros of higher voltage rating.

   I have employed different sockets for the inputs at the rear of the chassis to match the banana plugs that were available at the time of construction, I refer to the aerial panel. Cotton covered mains cord is employed to enhance the appearance but does employ an earth lead to conform to local safety regulations, the original set did not, a separate earth was recommended.
 

   Circuit differences are minor and few in number, they are essentially: the use of a vertical output transformer because I have not been able to find suitable wire to make the very high impedance vertical coils needed in the deflection yoke when resistive/capacitive coupling to the output valve is employed as in the original design. When a suitable yoke becomes available the transformer will be removed and the circuit retuned to the original arrangement. The circuit is otherwise unaltered except for a change in the output valves bias resistor. Use of silicon diodes is employed in a voltage doubler at present for the E.H.T. supply because the only available
transformer to hand had been wound with a 900V secondary. however the original E.H.T. rectifier has been installed on the chassis so that in the event of making or obtaining the correct deflection coils I will revert to the original arrangement. The H.T.supply is wired to operate with either a valve rectifier or silicon rectifiers incorporating series resistors. The latter is to avoid the danger of having a very hot rectifier at the rear of the set whilst the set is on display and operating. Non of the above changes in any way affect the performance of the set.

   The only real departure from the original set is the use of high quality capacitors in place of the original paper type, their effect being only their vastly longer longevity.

   The design of the cabinet presented a number of problems being my first attempt at cabinet making, a direct copy was ruled out, instead it was my opinion that the appearance should show clearly that no attempt was made to make a copy but rather to make a similar cabinet with no doubt as to the origins of the design. The top of the cabinet showing the most prominent difference. The dimension are the same as the original and the timber is solid maple, no veneer is used.

   The chassis layout is a direct copy of the original and the dimensions are nearly the same, the exception being that the chassis is almost half an inch higher to accommodate one of the power transformers.

   Obtaining suitable control knobs was solved by having the volume and tuning knobs made from aluminium and machined on a lathe whilst the other knobs were standard radio knobs suitably drilled and modified where required for the concentric arrangement for the contrast and brightness and the vertical and horizontal hold controls. All knobs were dipped in enamel paint to give them the original bakelite appearance. The escutcheon around the radio dial is made from thin timber as found in
cigar boxes and hand painted with many coats of gloss enamel paint.

   The underside of the cabinet has the same system of plinths as the original also the CRT mask is a copy of the original and made from timber.

   A few comments about the original design may well be in order. Firstly I have every admiration for the designers of the time. The development of the circuitry during 1937 shows how well they mastered the myriad of problems they were faced with. The sync separator for one is quite a clever arrangement whereby a diode suitably biased increases the effective video detector anode load during the sync pulses thereby more than doubling the sync pulse amplitude at grid of the sync limiter. This is achieved without reducing the video bandwidth and the method employed enables the sync circuits to be isolated from the video feed to the C.R.T. cathode, again helping preserve bandwidth. The video detector is an anode bend detector and quite progressive in the television technique for its period, it re appeared after the war in a number of Philips T.V. receivers.

 
   The sound system for T.V. is surprising , one would have not expected to have produced sufficient A.M. sound signal from the anode of the second common I.F. amplifier to drive the T.V. sound detector directly without running into problematic intermodulation between sound and vision signals at 4.5 and 8 MHz. respectively. The secret to achieving a satisfactory result was found to be in the design of the sound I.F. transformer feeding the detector, it has a significant step up ratio feeding a much larger signal to the diode detector than appears at the anode of the I.F valve . A well designed trap in the cathode of the final video I.F. amplifier ensures total elimination
of the sound signal at the video detector. Only when the alignment procedure is followed exactly (.this is no easy task) is adequate volume available, in contrast to this the radio performance is outstanding. The designers saved a valve by eliminating the need for a separate sound I.F. amplifier.

   Neither of the timebases call for special mention in my opinion, the circuitry being employed by British firms in one form or another right up to the introduction of transistors in the late 1960s,critical damping of the line flyback is used rather than the use of a diode, this was quite common before and just after the war. Damper diodes were known and sometimes used however as early as 1937 perhaps even 1936. The method of feeding sync to the line timebase is a little unusual in that sync of negative polarity is amplified in the line output valve and fed back to the grid of the line oscillator in the correct phase, somewhat unusual and possibly not as good as some methods. It does however work and shows quite a measure of ingenuity. Remember all this only ten or so years after the introduction of sound radio.

   The switching system employed in the 904 was nothing less than a masterpiece of engineering and design to enable normal radio techniques and the new very high frequency and wide bandwidth circuits to share common valves. The switching is obvious but very complex for it time, it performs faultlessly, it would be difficult to improve upon even today employing valves, the valves themselves could of course be improved upon but that does not detract from the brilliant circuitry. The 904 being described is almost a direct copy of the original. The change is minimal and was caused by an error occaisioned during construction involving the placement of a shield
between switch sections. The performance is not compromised. At the time of finishing this article a replacement 5FP4 has been fitted with excellent results. The magnetic focusing seems better than the electrostatic method employed with the 5AXP4 at the low E.H.T. employed
 

VB July 25. 2009