HMV 905 Project

[beery]

Hi all,

after a very busy period of work, I now have a bit of time for the HMV again.

Brian Cuff has very kindly lent me the scan coils from his set to copy.

Using these and the 907 parts that I have, I decided to have a look at how the tube mounting will be arranged.

The Russian CRT is 1/4" under 7", which is only 1/8" edge to edge. I used some 'E' section brown draught excluder to fill the gap. This is shown in the first picture.
I then fitted the CRT with the 905 coils and mounting bracket.

The second picture shows the coils in the position that they would be in with the original tube. Because of the slightly smaller diameter and larger deflection angle of the CRT we can see that the coils are too far back. The centre line of the mounting bracket is 33cm from the inside front of the cabinet, where a supporting block of wood is located.

The third picture shows the coils snug up to the CRT bowl. There is now ample room for the ion trap magnet, but the CRT mounting bracket now falls in an area where the cabinet top is too thin to support it. The centre line of the mounting bracket is 27cm from the inside front of the cabinet.

The forth picture shows the scan coils from a 905, with the focus coil brackets from a 907. This still leaves room for the ion trap maget (I hope) and places the mounting bracket 31cm from the front of the cabinet. In this position the cabinet is thick enough to accomodate another wooden mounting block, in front of the original one, so this looks to be the way forward.

The fifth picture shows the effect of having the CRT in place. In practice the draught excluder can't be seen unless you really look for it

Cheers
Andy

[mark pirate]

The CRT looks to be a good fit, I don't envy the task of replicating the scan coils though!

Mark

[FERNSEH]

Andy has asked me to trace out the wiring of the heater supply to the five 6.3 volt valves that exist in the HMV 905. Three KTZ63s and two KT63s. All are in the timebases.
I see no point taking photographic pictures and the best plan will be to trace out the wires and post up a diagram in the forum.

The set chosen for the job is my Marconi 707, identical chassis to the HMV 905.

DFWB.

[FERNSEH]

The wiring of the heater supply to the five 6.3volt valves employed in the timebases.
Usual octal valve practise, pins 2 and 7 are the heater connections.
Pin 7 returned to chassis at the point indicated in the drawing.

DFWB.

[beery]

Hi David,
Thanks for that.
It is very interesting that the 6.3V runs are two wire all the way back to the transformer. The 4V heaters all use one wire, with each valve having its own return to chassis, relying on the chassis to complete the circuit to the transformer.
It is also interesting to know that the colours of brown and black are the same as for the 4V supplies, which could get confusing...

Cheers
Andy

[beery]

Hi All,

I've been catching up on this set a bit now.

The first 3 pictures show the construction of a replacement for the missing brightness and contrast control.

The original part was by Morganite and had a 5K pot for brightness, with a 1K pot for contrast. I wanted to adapt a dual pot from a Cossor 339 scope. I've tried to find the correct values, but have yet to find a pot with a 1K resistance in the same style, so for the time being at least I have opted for something a bit more modern.

I used a 10K and 50K pot from a scrap Tektronics scope as a basis. I removed the old 10K track and replaced it with a new 1K track from an 'Alpha' pot.
Getting the wipers to reach the new track turned out to be very tricky, so rather than try to fit a new 5K track in the rear section, I decided to put the 10K track instead. If I put a 10K resistor across the track (to make it look like a 5K pot to the rest of the circuit) then the circuit should be fine since the wiper simply feeds the grid of the CRT which is quite a high impedance.

I then turned up a shaft extension using a lathe at a local engineering firm, it is the first time of using a lathe in years. I attached the extension with super glue, this is shown in the 3rd picture, alongside the original line and frame hold control.

The 4th picture shows me adapting new EHT resistors to look old. After painting them with Humbrol enamel, they were assembled as the EHT bleeder chain as shown in the 5th picture. The original resistors (Morganite) had cast ends and neatly painted bands. There was no way to replicate them, so I made these resistors look like the more common Erie type. However my colour spots are not very neat.

I've now removed the bleeder bracket for the moment to improve access. I aim to get the frame oscillator operating next...

Cheers
Andy

[mark pirate]

I think you have done a great job on the resistors Andy, I think they look very authentic

Mark

[Station X]

Well done. I am enjoying reading this thread. You deserve our admiration for taking on an "impossible" project like this requiring a multitude of different skills.

[beery]

Hi all,
Many thanks to those who have written encouraging comments

This project seems very daunting at times. I just have to approach it in he same way as eating an elephant.

I've replaced the last of the fried components in the video output stage.

The first photo shows what was left after parts had been removed. Again the heat had damaged the aged solder tags which simply fell off, whereas most of the rest of the tags have been re-used wth no problem.

The second photo shows the process of fitting a replacement tag. To do this the insulator behind the tag board has to be removed. The tags were kindly supplied by Brian C of course and the round spacer being clamped up is one I had made for re-winding 3000 type GPO relays. Initially the shaft of a Philips screwdriver is placed behind the tag to make its retaing wings bend the right way, ie outwards, and then the screwdriver is removed for the final tightening.

The third photo is a partial view of the insulator that goes behind the timebase tag board. In these sets the line timebase tagboard clashes with the unused chassis mounting lugs, I'm not sure why these extra lugs exist.
To solve this problem the tag board was spaced off by using spacers at the back of the insulator (4th photo). These spacers are only held in place by the mouting screws, so re-assembly after servicing can involve much swearing! Well of course this tag board and its insulator are new on my set. The spacers came from GPO telephone dials (behind the off normal cam) and to make my life easier I superglued them to the insulator

The fifth photo shows some of the components fitted behind the width and height controls. The control knobs are ex TV22.

Cheers
Andy

[beery]

Hi All,

Another update...

Using Fernseh's diagram I was able to re-wire the 6.3V heater circuit in the timebase section (First photo). I was then able to make a start on wiring up the line timebase.

The second photo shows the large inductor that I found in an odd location under the chassis. When I was working out what parts were original, but not in their correct location, I found that this inductor had at one time been mounted in place of CK1 which feeds HT to the vision and timebase sections. As found though had much too high a resistance at around 400 ohms instead of the 160 required. CK1 must have been a bit unreliable as I know at least one other set where it has been changed for something bigger.

How to rewind the inductor? Well I had no winding details for CK1, so I first looked at similar items. The 100mA inductor from the Maplin 'Newton' amplifier is physically the same size as the original CK1 and measured 150 ohms with inductance of about 10 henries.
So just a bit under 100mA would be about the right current rating.

The inductor to be re-wound is bigger than the original, so I picked 32 SWG as being suitably overrated to exploit the space available and make the inductor more reliable. I measured the circumference of the inside of the coil former and the outer circumference. Taking the half way point between the two measurements gave a distance of 0.111 metres.

32 SWG has a resistance of 0.278 ohms/m. If we divide the resitance value of CK1, 160 ohms by 0.278 we get a wire length of 575.5 meters.

575.5 divided by the average core circumference of 0.111m = 5148 turns. I rounded this down to 5000 turns and re-wound the inductor (third photo), which ended up with a reistance of 163 ohms - near enough. The inductance must be more than the original, which is fine in this application. It is worth noting however if CK2 (to the left of CK1) needed replacing then inductance would be important as CK2 acts as a reactive dropper in radio mode.

The forth photo shows the inductor in its original position in the set. Note that I doubt that the original chassis bolted C38 would have been in place when CK1 was originally replaced. The fifth photo shows the original arrangement.

Well I'm glad that the idea of estimating the turns for CK1 worked out, because I plan to use the same trick to work out the number of turns for the frame coils...

Cheers
Andy

[PaulR]

I continue to sit back in awe and admiration over this project!

Paul

[peter_scott]

Me too!

Peter

[beery]

Hi all,

I'm still working out the solution to the missing IF coil. In the mean time the focus is on the timebases.

This weekend I managed to get the line oscillator running - sort of.

Both line and frame oscillator coils are from a scrap 1804 and their resistances are quite different from the originals. This should make little difference to the line oscillator as it is in the tetrode form with the saw tooth charging circuit (R47, C68) separate from the blocking part. With the frame circuit however, the triode blocking oscillator circuit is used, as opposed to the 1804 which uses a pentode/tetrode circuit, so linearity might be a problem - I won't know until I try it.

The first picture shows the line oscillator coil croc clipped, ready for testing. I don't yet have a can in place on the line oscillator coil. The sync amp is in place, only to make sure that the oscillator has the right voltages fed to it (more on this later).
At first the oscilltor was running at 40KHz! I realized that with both the lopty and the feedback capacitor (C85) missing things were not going to go too well.
The frequency is set by VR6, R46, C67 and C85. I put C85 in parallel with C67 (C85 is now actually 220pF)to give 450pF and things started to look better. The frequency was still to high, so I checked R46 which was reading a little low. The 1804 circuit uses a single 500pF timing cap with a 50K hold control which has a 47K resistor in series, so I changed R46 to 47K. Now I could almost get the right frequency. I may need to raise the value of R46 again, but I'll wait until I get the lopty in place as I'm not sure what effect the feedback will have on the frequency.
The third picture shows the results, taken from the grid feed to the line output valve (not fitted), the width control was in the middle position.

The fourth picture shows the arrangement of using a PSU for the 6.3V suppplies. This got very annoying as the bench is small, so I decided to wind a cheeky 6.3V and 4V heater transformer just for testing. The final picture shows this transformer in place in the top left-hand corner. The line oscillator coil now has proper leads on it, though they are deliberately long for now as I may need to remove it again when I fit the screening can.

The chassis is a bit ungainly now as I've fitted the frame timebase tag strips. These actually protrude beneath the bottom of the chassis sides! EMI got round this by fitting spikey feet on rubber spacers, however, if you place the chassis down across a narrow bench - crunch... I've made a couple of wooden batterns to solve this problem (not shown in these photos).

After this point I had run out of yellow wire. I've loads, but not at the same location as the telly!! So to test the frame stage more croc clips will appear.

I mentioned earlier the voltages from the sync amp. Well the width and height controls, the brightness control and the sync amp are part of a complicated protential divider circuit. I found out that the voltage across the brightness control was 45 volts, but it should be 40. This was actually due to the sync amp valve being dead, which I found out when I tried it in the oscillator position. Now for the KTZ63 confusion...

When I started to amass parts for the set I purchased NOS valves to replace the missing ones. I purchased valves listed at KTZ63. What I recieved was the Brimar equivalent, 6J7G. The plain white valve box was marked 6J7G/KTZ63/CV1074. This is when I found a discrepency... Pre-war KTZ63 valves are tetrodes with beam forming plates connected internally to the cathode. During WW2 the KTZ63 was altered to become equivalent to the 6J7G pentode with the beam forming pins connected to pin 5. This can be seen in the post war HMV 1804 circuit where pin 5 is connected externally to the cathode. In the 905 of course there is no connection on pin 5, but I think I will add it as genuine pre-war KTZ63 valves are not easy to spot (even if it is marked KTZ63 it is no guarantee).
I tried one of the 6J7G valves in the oscillator position and I got a much faster flyback, so clearly my good KTZ63 is a bit tired. The valve even worked with G3 (pin 5) floating, though it was not very happy

I think that's enough for now. And I even found time to have a day out enjoying the sun with my wife

Cheers
Andy

[FERNSEH]

The attachment shows the wiring of the 4volt valve valves employed in the vision circuits. D42, MS4B, and the two KTZ41 IF amplifier beam RF tetrodes.

DFWB.

[beery]

Hi all,

Well the reconstruction has taken a slight back seat whilst I sort through a whole bunch of spares I have now obtained.

A scrap 907 chassis turned up and was stripped for parts. I will say at this point that the chassis was not worth saving in its self, it was incomplete and rusty on underneath the chassis as well as on the top. Most of the recovered parts are faulty in some way.

I drilled out all of the valve sockets and examined them under a magnifier to see which ones were not too corroded.

I aimed to remove all the pots, but the wirewound ones were open circuit anyway. The combined brightness and contrast control needed to be saved, but the outer shaft collar was rusted on. I had to cut it off with a dremmel; the 5K section is in need of repair though.

A substitute for the missing IF coil was saved (it needs winding to the correct value), which is funny as I was half way through trying to make a new coil former from scratch (I will post pictures when I pick it up).

The missing RF choke from the video drive/detector was obtained.

The line oscillator transformer and can are perfect, protected by tar.

The frame oscillator transformer was saved, but has an open circuit winding.

As well as picking up this chassis at the NVCF, I also collected the new line output and EHT transformers that Mike Barker had made for me from scratch. They look amazing. As requested, Mike added a 6.3V tap to the CRT heater supply to enable it to work with the russian CRT.

Now I have to sort through the bits I have and undo some of my existing work to put more orginal parts in.

Cheers
Andy

[draenog]

That was lucky getting a scrap 907 chassis (which I know well from restoring my HMV 907). A nice set of parts You even have something I don't have - an original audio transformer The one in my 907 is a non-original Radiospares transformer, I hope to get Mike Barker to make me an original looking one (judging from the ones you've had made it will be exact).

[beery]

Hi all,

Just a quick update.

Regarding the sound output transformer, well my set also had an RS universal one fitted. Spare EMI audio output transformers I do have, but it is the terminal arrangment - only found in their TV sets, that is the problem. With one to look at, I might be able to copy the terminal board arrangement. I'm going to need a big soldering iron to solder the brackets to the U clamp (the transformer from the 709 has one bracket that has come adrift anyway) and access to a set of number stamps.

From the scrap 709 chassis I was able to recover just enough usable valve socket contacts to replace the four British bases that were missing, I had already replaced them with 'Clix' sockets bodged in with large washers, but it is great to have the correct EMI ones.

I was able to recover the combined brightness and contrast control, the 1K section was ok, but the 5K section was not well.
The 5K section was measuring 7.5K in total and more disturbingly it rattled when shaken. When one of these 'Morganite Stackpole' pots rattles, it means that the wiper disc has fallen off from the wiper arm - which results in the track getting scratched.
Upon opening up the pot this was indeed what was found. The first photo shows the deep scratch which increased the overall resistance from 5K to 7.5K.
The second photo shows what it should look like.
I briefly tried a new track, but it was a log type, so in the end I decided to put the scratched one back as I figured that the increased resistance would not effect things too much.
Having swapped the tracks about meant that I had to dismantle the 1K pot at the front in order to reach the two track retaining screws. Putting the front pot back together meant balancing the wiper disc on the wiper arm (third photo) and then carefully tightening the retaining nut. A shake test confirmed that the wiper disc was still in place.
Upon testing the set with the 7.5K pot in circuit, no real change was noticed, with the pot still having about 45 volts across it.

By studying the circuit and looking at the other 8MHz IF coil, I worked out that 26 turns of wire were needed for the missing IF coil. I took one of the coils recovered from the 709 and removed turns until I reached the correct number. I then re-created the whole coil and trimming capactior assembly, as shown in the fourth picture.

The fifth picture shows the timebase and vision output area of the top of the chassis as it is now, with the correct valve sockets, the new lopty fitted (the very large round can on the bottom left) and the correct line oscillator transformer (in the small round can on the right of the picture). It all looks good, but there is still lots of wiring to do on the underside. I'm considering retro fitting genuine EMI tag strips, but I'm still working out how feasable this is.

In theory there is not a great deal to do before I can test the set with the borrowed scan coils, but this will have to wait a while as I will be spending a week or so back at my parents' place. However this will give me access to a pillar drill and so enable me to work on the winding bobbins for the scan coils.

That's all for now.

Cheers
Andy

[beery]

Hi all,

Having recovered valve sockets and tag strips from the scrap 907 chassis, I then set about fitting them to the 905.

The reason behind using original tag strips is because the new ones I had were quite a bit lighter in colour than the originals and as I now had some genuine ones, I felt it best to fit them.
The recovered solder tags were too corroded to use and anyway, I had already put together the tag strips with the new tags from Brianc. What I was able to do was to fit the new contacts into the original tag strips without removing the components that I had already fitted.

I mentioned earlier that I had started to make the replacement coil former assembly before I got hold of some original ones. I've included a picture of how far I got, with the original tuning slug on the right of the picture and the new parts on the left. I enjoyed doing it and it was about the most work I have done on a lathe (apart from CNC repairs!), so not a complete waste of time.

The chassis is now just about complete. I've added before and after photos.
I plan to re-replace a couple of waxies in the audio section with older ones and I need to change the sound output transformer to one with a terminal panel on the top.

I decided to carry out two modifications from the original design.
The first was to join pins 5 and 8 of the KTZ63 valve sockets to enable the original tetrode and the later pentode types to be used. This was done on the post war HMV 1804.
The second mod was to make the HT feed to the line output stage go via the scan coil plug. This is to prevent the lopty from breaking down if the chassis is powered with the coils disconnected. This is not an original idea, but I had read of someone else doing it on their set (was it on this forum?). I felt this was wise as I still don't have my own set of scan coils for the 905.

Continued...

Cheers
Andy

[beery]

Hi all,

Now the fun part. Alignment and first light.

I aligned the set as best I could, but the IF coils don't really have a sweet spot. This is partly due to the circuit design as they are heavily damped. A worse thing though was that the sound rejector was not working properly, so I need to investigate why the trimmer circuit I had replaced was not having the desired effect. I did however forget to check the local oscillator setting (I had set it ages ago, but more parts have gone in since then).
For safety reasons all the alignment was done with the EHT rectifier's top cap anode connection unsoldered from the EHT transformer.

The scan coils from Brianc were combined with the focus coil that I have. The tube was slotted in and a Mullard ion trap magnet was fitted.

On switch on, after fumbling about with the tap magnet I got a very blurred test card C. It turned out that this CRT needs a lot less focus current than the EMI one. They are of course both triode tubes, so why the difference??

With more resistors added to the focus circuit I was finally able to get a picture. Over all the set is fairly deaf, needing the full output of the Aurora. There is pattering on the picture and of course sound on vision due to the fault I mentioned earlier.

The brightness is not all that, but then we are talking about only 2.5KV, so it is probably like new! The CRT is really good actually and there is pots of width and height available, which is important as my home made coils may not be as efficient as the originals.

I settled back to watch an old British 'B' movie on it. Can you guess what is was from the picture?

After soak testing the set started to become unstable. Both the local oscillator and the video amp are very old NOS, after a bit of use it seems that the soldering on the valve bases is starting to fail. Things improved after re-soldering the video amp base, but I have yet to give the local oscillator the same treatment.

I'm quite happy with the results thus far though...

Cheers
Andy

[Jac]

Very impressive work Andy!

I can imagine how many (sleepless) hours have gone into this.

Jac