Museum Hours:

Saturday 10-6

Sunday 12-5

Early Electronic Television

Brian Cuff's GEC BT-8161 Restoration

Brian Cuff recently purchased an empty GEC BT-8161 on Ebay. Here is the story of his reconstruction of the set, from posts on the V.R.A.T discussion group. Text in italics are from Brian, standard text is our comments:

September - October 2012

As many of you know, I purchased the empty GEC 8161 TV on ebay. There are no known examples of this set and so this cabinet is very important. I am going to attempt to re-create the innards as closely as possible to the originals but unfortunately, there seems to be a dearth of info on this set or on its smaller cousin, the 9122. I have been trawling through my RX data for a likely candidate for the radio chassis. GEC seemed to have gone mainly for rotary pointers and dials and the (poor) pictures of the 8161 we have indicate a linear dial. Looking at the knob positions and scale size, I think that the radio could be a modified "GEC Fidelity SW5". Does anyone have one of these and even better, does anyone have one of these which they are willing to part with as a possible donor? As soon as I get the cabinet, I will clean it up a little and post some pictures. The vendor lives in Sudbury but I find that there are two Sudburys! One near Ipswitch and the other near London . I am waiting to hear which - I bet it's the Suffolk one!!!

This is going to be a loooooonnnnng drawn out thread I imagine and any advice is welcome.

(Brochure courtesy of oldticktock)

When I get the cabinet, I'll see if I can make some sense of the fittings that are left. Hopefully, the chassis mounts will be either threaded bushes or some sort of counter-bored holes which took rubber washers rather like the HMV 800 chassis mounts. I will also look for witness marks where the brackets were. This cabinet was rescued from an Attic. The builders rescued the cabinet because they recognizable it as a piece of furniture. I wonder if they skipped the chassis because they were not recognizable as anything.

Till Eulenspiegel provided these pictures of his still not restored chassis of his BT9121. Note the compact timebase unit.

He also provided circuit diagrams of the GEC BT8161 and the GEC BT912. Although the two chassis are similar there a few noteworthy differences. Both are superhets operating with low IFs. In the TV only receiver 9121 the RF amplifiers and frequency changer are on the vision chassis section. In the more higher specced up 8161 it would appear that the television early stages are actually on the "A" chassis, the radio receiver section. more research will have to be done to confirm that. The vision demodulator in the 8161 is of the full wave type, which is desirable because of the very low intermediate frequency and the large 15" CRT. The much cheaper BT9121 has a simple half wave vision demodulator. So it would seem that the main differences are in the signal circuits. The timebases and vision power supply arrangements are similar between the two chassis.



Till commented further: More thoughts about the GEC BT8161 TV and radio receiver. The vision IF in the TV only BT9121 is 6Mc/s, the sound is 2.5Mc/s. However I am becoming of the opinion the the vision IF in the BT8161 is even lower, as low as 4Mc/s. That's very low indeed but is worth bearing in mind that early Murphy TVs used a very low vision IF, in the V58A it was 4.25Mc/s. The sound was 0.75Mc/s.
I have no information for the front end of the radio, but it is most likely that the IF is about 500Kc/s, a bit higher than than the standard frequency of 465Kc/s. Operating 405 television on a low carrier frequency was used by Rediffusion on the TDUK2 two channel wired system. One channel was 4.95Mc/s and the other 8.45Mc/s. The system frequency receiver used a full wave vision detector to reduce the posibillity of carrier dot pattern being apparant when the lower frequency was selected. The GEC BT8161 has a full wave vision detector.

Till also supplied a schematic of the radio chassis:

Well, I have the cabinet at home at last. It's not in bad condition - no woodworm or large chunks of veneer missing. Badly scratched and the top especially will have to be re-finished some time.

It hasn't been too badly messed about with although there are some witness marks where blocks have been removed. There are 2 mounting blocks fixed to the outside of the two chassis on each side of the tube but no corresponding pairs on the inside of the apertures. The lid is counterbalanced by two pulley systems, attached to the stays, which lift counterweights as the lid is raised. However, the weights are missing.

One mystery is solved - the knob in front of the CRT screen is the magic eye for the radio (Y63 introduced in 1937). There are small threaded angle brackets which support the wooden escutcheons covering the two chassis so they will be easy to make once the layout of the knobs has been determined.

I have been thinking about the variable bandwidth mechanism and, assuming that the radio chassis is mounted vertically, there will have to be some sort of cam system operating either a system of levers or two cams acting directly on the IF transformers - how exciting!!!

There is a GEC Super 6 on ebay so I have set a bid for it. At least the magic eye mounting looks the same.
I have also sourced a 12" electrostatic CRT (VCR58) as a stopgap until a 15" comes available - probably never, but at least I can hope.

I will post a wanted list in the appropriate section for the valves required and I will require some bases, almost all B7, B5 and B4 - I have some but not enough. Till is supplying me with info and I have cleaned up the circuit scans and will work on the others. Transformers will be a problem but hopefully, this is where Mike B and Ed D can help. The large mains transformer has 7 heater windings plus HT!

I will also cut and veneer a couple of boards for the control panels - the main cabinet is figured walnut, the top being a bit damaged but nothing that a good scrape will not sort out. It will be nice to see it with the two control "escutcheons" in place.

Till provided these pictures of the other GEC BT9121. The timebase module is placed to the right of the CRT rear support. The GT1B thyratrons are inside the screening cans.

A bit of progress in the last couple of days! I have, at last, located the owner of the "other" GEC set. It turns out that it is a BT9122 which is the smaller (12") brother/sister of the BT8161 (15"). He has agreed for me to visit him to see the set and take pictures, measurements and the like - that's a real step forward.

I have spoken with the 9122 owner who is very pleased to help the project along! One very important thing - he has a full manual which covers both the 9122 and the 8161. Evidently, they are identical except for the central CRT mounting (12" or 16"). The manual includes full chassis layout drawings and component values etc, which were missing from the data to hand at the moment. Hopefully, when I visit him, I will be able to scan the docs - he is not willing to trust the vagaries of the postal system, having had some bad experiences.

Meanwhile, I decided to give the cabinet a bit of a polish and take some pictures. Here are a few of them:

The speaker cloth is modern(ish) Tygan so not original - I will see what the original looked like when I visit the BT9122! There is also a hole for a 12" speaker cut in the baffle board (as well as the original 8" hole!) so I guess it was converted into a hi-fi cabinet, fortunately with not too much hacking around!

Perhaps it was not very clear, sorry, but I said in my previous post that there was a 12" cut in the baffle board - I meant cut by person or persons unknown. The hole has been either enlarged or has been cut out for the conversion job.

I can guarantee that the baffle is not a replacement. It has had an extra hole cut in it. The smaller hole to the bottom left hand corner is also original. The large hole placed more centrally has been crudely cut and not that many years ago. The edges are still very clean (but rough).

November - December 2012

A bit more progress. I have sourced an 11.5" electrostatic CRT, a CV1085. Gerry at Crowethorne tubes had one he was willing to part with together with some of the valves required. The CRT is a really bulbous jobbie and really looks the part. I can only imagine what the 16" version looked like!

I have also got, in my spares stock, a 12" GEC mask which fits after a fashion as can be seen from the pictures. I am still waiting for my visit to the 9122 but unfortunately, the owner has had some bad family news and I don't want to trouble him at this difficult time for him.
Here are some pictures of the CRT:

It is a radar tube so the phosphor is not quite right. It should be P4 but is P7. A P7 is a P4 type phosphor and a yellow, long afterglow phosphor which is necessary for radar. However, hopefully, a blue filter will minimise the lag caused by the long afterglow.

It's going to be a long haul, Rich but at least I have now started but need the service info and the visit to go much further. I also need to commission the mains transformer for the TB chassis but need to measure the 9122's example to get the lam stack as close as possible. I have worked out the heater windings but can't do the same for the HT windings until I get the component values (I have the brightness pot's traveler range of 80 - 200V but there are resistors in the chain).

I now have all the valves for the TB (B) chassis except the 2 KT66s and an MS4B. I Haven't checked my stock yet for the Radio (A) Chassis but I know that I have some of them.

Just to put the record straight, the "other" GEC mirror-lid set is not a BT9121 as previously thought. It is a BT8121 which makes it identical to the BT8161 except for the CRT (12" not 16") and the cabinet. The brochure that Till posted shows both the BT8121 and the BT8161 - brother and sister!A bit more information on the set. The following are pictures of the 8121 taken some while ago. Note the simple CRT mounting. That will be easy to reproduce!

The set is in great condition. It came from a gentlemans' club in London so was never in a dirty environment (well, not in the dusty sense).

January - February 2013

I've received the info from John and have been able to do a bit more research and preliminary planning. You may have read about the sound and vision IF frequency discussions in the thread about the Cossor 54 pre-war TV. Well, it turns out that the IFs for the GEC 8161 are also very low at 4MHz and 465(500)kHz for vision and sound respectively. That means that the local oscillator will run at 41MHz giving the differences of 500kHz and 4MHz. It seems strange that the local oscillator frequency is so close to the sound carrier but I suppose it must have worked!

The info from John included layout drawings of both chassis so I have been able to draw them properly using Autocad. I had no absolute measurement as a reference so I have drawn everything referenced to a length I estimated one side to be. When I get to see the existing 9122 in the flesh, I can measure the reference sides and then scale up (or down) all the other measurements to arrive at full size drawings.
I have also been able to detail the requirements of the three mains transformers from the list of measured voltages and currents and the listed valves and am slowly acquiring them.

A long, long way to go but I'm on the way. Till's work on his 12" mask has inspired me to get on with some woodwork, hopefully this month. I now know that the radio or "A" chassis is on the left of the CRT and the TV or "B" chassis on the right and I will start on making and veneering the two plywood decks which cover the chassis.

Hmmm - not a lot to do.

I bought this GEC radio on Ebay.! So I now have the first main hunk of the "A" Chassis (front end and radio). I now have to decide whether I should try to modify its chassis by cutting the holes for the cams, extra coils and valve or to just use the components off it. I am tending towards the latter as the "B" Chassis (timebase and vision IF etc.) must be made from scratch so will be fabricated from steel. Maybe it's better to have them both the same!

The chassis layout looks identical to the A chassis for the 8161 except that it has no RF stage. the 5 valves are in their appropriate positions and there is a gap where the RF coils and the 6th valve would go! I would also imagine that the IF cans do not have variable coupling but the tuning cap is underneath the chassis unlike the BC3760 so a combination of the two would be perfect.

March - April 2013

I have at last got the GEC radio chassis for the 8161. It's turned out to be a BT3750 which is a 5 valve superhet but the chassis is identical to the BT8161 radio section except for the holes for the RF amp. bits. I was worried about 2 things with this chassis, the tuning cap, - must be 3 gang - no problems, the radio has a band-pass aerial section so the tunicg cap is 3 gang. The other was the variable coupling in the IF cans - also no problem. The facility is provided in this chassis too! In fact, the BC3750 is really a better chassis that the BC3760 because of its mechanical arrangement.

Soooohhhhh. What to do. Do I mess around with the set adding an RF stage by drilling holes and mounting the extra valve and coil can (their positions are free) or say bu**er it and just modify the short wave range to cover the correct frequency. Of course, the sensitivity will be down but I'm not short of signal strength as the Aurora pumps out loads!

Just a quick update on the BC3750 radio chassis: I am really pleased that I was able to buy this set. The chassis, I know now, is mechanically the same as the TV version except for the missing RF stage valve base and coil can. The valves have screening cans riveted onto the chassis over the valve holders so there are 4 fixing holes for the RF and low signal stages. In the position where the 6th valve is on the TV radio chassis, there are 4 holes with the correct positions in the radio chassis. This proves that it has been manufactured from the same basic drawing .

However, the band switching is somewhat different and, the TV radio having an RF stage, means that the switching uses more wafers therfore if I want to rebuild to the TV radio spec, I will need to add wafers. This means I will need a second radio chassis as a donor - should I wish to go that route. At the moment though, I will modify the chassis I have to receive 45MHz (the IF frequencies are 455kHz and 4MHz) which means some changes to the RF section and the addition of a 4MHz transformer to tap off the vision signal. As I said earlier, I am not short of signal strength!

I thought it would be of interest if I uploaded some pictures of the variable IF coupling mechanics. It is mechanically a very simple method but I feel that it has been done very well by GEC in this chassis, even down to the little spiral connections to the moving coil.The radio is now working and the short wave is very lively so maybe I can get it up to 45MHz!

I'm glad I didn't need to make this!

Close-up of one cam

Very neat construction. Shown with maximum coupling.

Showing minimum coupling.

I thought that I would do a bit of experimenting for a change! I need to construct some 4MHz IF transformers for the new "B" chassis which carries all the TV only components. I had bought a scrap contemporary GEC chassis a couple of months ago for its large round coil cans. The IF cans are also of similar shape to the "A" chassis cans shown in the previous post and I needed one of these for the 4MHz pick-off transformer which is in the anode CCT of the frequency changer - the vision IF is 4MHz and the sound IF is 445kHz (500kHz).
So onto the web for some coil winding info. I found a site with some useful calculating gizmos and from these made this 1:1 IF transformer. A old tuning cap gave me the value of C for 4MHz with 24T of 28swg wire and so here it is. The coupling should be fairly tight to give me the requisite bandwidth but that needs more investigation. At least it works as a tuned transformer!

The experimental 4MHz IF transformer

As to the pick-off transformer, only the secondary is tuned so I need to make another version. Has anyone any idea of the coupling required? I would have thought that it needs to be fairly tight to give a reasonable signal for the following stages. Should I bifilar wind it of just put two windings close together? Any ideas?

I have been a bit sidetracked so have not looked into Spice etc. (sounds a bit complicated to an old 1V in 75 ohm man like me)!
I visited the owner of the BT8121 last week (about 65 miles) with the aim of taking pictures and measurements. Before I left home, I checked what was on the memory card in my DSLR camera and put the card in my computer reader and deleted all the files I didn't want. I then left for my visit and, unbeknown to me, I had left the memory card in the computer - so no camera although I borrowed a point and shoot one from David!

The visit was very interesting and so useful. I got pictures of how both chassis mount in the cabinet (all the cabinet fittings are in place in my cabinet ) and got a general idea of how it goes together. It confirms that the radio chassis I have (BC3750) is definitely the right chassis to have and it will be possible to fit it properly into the cabinet - even the knobs are the same with the four controls operated from two shafts using linkages, one of which can be seen in the pictures.

The mains power switching is taken care of by two cam-operated toggle switches which are mounted to the front of the chassis with complicated brackets so I thought that I would have a go at reproducing them today. The two pictures show the BT8121 switchery and my attempt at reproducing it. No too bad, I think. The toggle is for switching on the B chassis when TV is selected. I will use the SW position for TV as it will be nigh on impossible to add the 4th switch position. If I get another BC3750 I might have a go but I don't want to bu**er up the chassis I have.

The genuine article

The imposter

A bit more mechanical progress in the last couple of days. I'm getting close to having an "A" (radio) chassis that is a good representation of the original. The dial of the BC3750 radio has been lowered and mounted vertically instead of sloping backwards and I have drawn the proper BT8161 dial in Photoshop. I will probably try and find a tame silk-screener to make a dial using the Photoshop files. Due to the mods., the control knobs and dial will now be in the same horizontal plane and I will manufacture the veneered escutcheon panel. The proof of the pudding will be the fitting of the chassis into the cabinet! Pictures show the progress.

This is how the radio chassis looked before modification

The finished control area of the radio chassis

A print of the new dial

One thing that this work has proved to me is that I have repeated exactly what the designers of the BT8161 TV did when they modified the BC3750 chassis for use in the TV!

May - June 2013

I'm not going to bother adding the RF stage, Till, as I might bu**er up the chassis. If I can get hold of another BC1750 chassis, then I may have a go after I have got the "B" chassis done - that's the real big job. I am a trifle worried about the scan output transformers. As you say, they have to have 6kV insulation primary-secondaries and I suppose even more secondary-secondary! If I am going to have them made, I will need to know the turns ratios so when your doing the checks to see if that secondary is faulty, you'll need to bung a sine wave through it so perhaps you could find out for me .

I've made a bit more mechanical progress today and have proved that the "A" chassis fits as it should. I made up a couple of alloy runners using dimensions I took from the BT8121. I used alloy because I had the material to hand and the originals were pressed steel with strengthening bends which I couldn't possibly have produced. As you can see from the pictures, it fits well and I have made the template so that the escutcheon board can be made. Fortunately, the veneering is not a fancy book or quartered type so that will be fairly easy! I don't think that the large knob is correct (I forgot to take a picture of the BT8121 with its knobs on!).

The dial on the BT8121 was as depply positioned as this one - it did look a bit strange. I might re-make the two brackets holding it onto the chassis to move it a bit closer to the panel.

The can on the extreme right is the 4MHz vision takeoff transformer with the phono output.

A full frontal with perspective distortion!

The template in position

Just a thought: It might be thought that using a phono (or RCA) connector for the vision IF connection was a little strange! The BT8161 used a modified 2 pin 5amp mains plug (the pins were turned down a bit) and I didn't want to do that so the best solution for me was a phono socket. Evidently, the phono plug was used a lot for RF in its early days (the ones with the extra long centre pin) so I feel it's OK to use. The pre-war Pye 815 TV uses a phono plug as the aerial input!

A little more progress, this time "theoretical"! I have drawn the B (TV) chassis using Autocad so that I can get a local engineering company to make it. I can't really handle 1.6mm steel plate of this size. The drawing includes the holes for the valves and RF transformers and 10mm locating holes for the capacitor boxes which I have to make. These are where the connection wires will exit the blocks. I have asked a couple of companies to quote and have heard back from one already - a bit expensive, so I hope that the other will be a bit cheaper!

I have also placed a post in the wanted section for the 4 knobs I require to complete the set (9 in total). I was very fortunate to get hold of an extra lever control for the B chassis, the two for the A chassis comming from the BC1750 radio.

I have noticed that there is another BC1750 on ebay at the moment and I have, of course, placed a bid so that if I get it, I can try the full mods to convert it to a real A chassis (adding an RF stage and further switching).

As you can see from the drawing, I will drill all the small assembly screw holes, 6BA, 4BA etc.

Typing this reminds me of something that really surprised me - the GEC radio chassis used metric screws, not BA. However, I won't be using metric screws because I only have Posidrive versions and they would look terrible!

I am still waiting for prices.

However, I decided to start on the capacitor boxes with the hardest, the EHT smoother, first. This cap has rounded corners so I knew it would be a bit of a bu**er but I've started. The pics show progress so far together with a special joggle tool made to enable the two halves to be joined - the 130mm sides could have been formed out of one piece but at over 300mm long it would have been too wide for my guillotine. I have soldered one of the side pieces to the top but it was a real struggle and I just hope it won't pop when I solder the other side on! More later.

The 4 basic parts

The joggle produced by the tool

The joggle tool (150mm wide)

It's easy to see how the joggle tool works from the picture, the tool being squeezed together in the vice which deforms the brass sheet accordingly. I must buy some tinplate. It will be easier to work and doesn't "work harden" as brass does.

Well, the EHT cap is finished - hooray. It was quite a job and I shouldn't "waste" my time on such detail but I took it as a challenge. As can be seen on the original (in the BT8121), the capacitor can is swaged together like a bean tin which, of course, I couldn't do so I soldered some 16swg tinned copper round the rim to represent the seam. It works quite well. It should be 2 x 0.15mfd but I put in 2 x 0.1mfd. That's quite enough at 6.5kV .

There are two other bits of progress on this project: one is that I have managed to purchase yet another radio chassis (a BC3750) on ebay. I am going to wait to see how I get on with the simple electronic conversion of the set which I have already mechanically moodified before doing anything with it. What I really want now is a BC3760 or BC3765 which would give me the 4 position wavechange switches required for the "proper mods). The other bit of progress is the ordering of the B chassis. I went to two companies, one which punches using a Pierce All punch and the other laser cutting. The former was under half the price of the laser man so that's where I placed the order. It has welded corners and will be zinc plated and passivated. Ready in 2 weeks - fantastic. I now need to think about mains transformers!

The cap as it is in the BT8121

My Imitation

As can be seen from the picture, the chassis has been made and delivered (well, collected actually). The owner of the company was very obliging and made a good job of the chassis (although the drain marks on the plating are not too good!). He new most of the guys by and for whom I had mechanical work done when I was in business including some companies that no longer exist such as Braybury Electronics, IPK Broadcast Systems and Ampex Systems.

Anyway, I digress! I have placed the cap and xfmr boxes so far made onto the chassis in the relevant places and it's looking good! I have to make straps to hold most of the caps down so I have bought some 1/4" x 1/16" mild steel strip which I will bend into shape and drill.
There is a complication with the control panel as I think that I may have misinterpreted the picture. It seems that the top end of the chassis is open and there should be a separate panel which carries some of the controls on brackets but I'm not sure whether I can do it that way now! There are three pots which are associated with the EHT bleed chain (focus & shifts) and these are mounted on a thick Ebonite sub-panel because of the voltages involved and I can't make out how the sub-panel is mounted onto the chassis so I will have to improvise there as well.

I have ordered a set of pots from Allied Electronics and they are reasonably priced (all large format 2W components) but the shipping, at $45 is dreadful! Still, what can we do - they seem to have a much better range of components available ex-stock over the pond. Why is that? Is it just that our suppliers are too crippled by taxes to have large inventories or are we even more in thrall to accountants now?

The new chassis with cans placed in posotion (for show!!)

Hi guys. I'm a bit peed off at the moment as I had composed a long post showing the last few days' progress on the BT8161 and, instead of hitting the minimise button, I hit the red cross and lost everything. Surely, it's a great failing in the software not to give the normal window "are you sure" reminder when there has been something added to the page which is open!
Anyway, I will recount some of the stuff but it won't be as detailed as the first version was.

Good mechanical progress which culminated in the attachment of the control panel to the B chassis and the addition of the insulating sub-panel on which the pots in the EHT chain are mounted (they are up at about 5.5kV). To obtain the information required to get to this stage, I had to scale things from pictures I have of the BT8121, the 12" version of my set. From the dimensions calculated, I was able to draw the assembly in ACAD (only 2D, I never got the hang of 3D) and the resultant section is shown below:

The drawing is pretty self explanitary so I won't describe it in detail except to point out that the the pots, in green, are at 6kV (shifts and focus) so are mounted on the 6mm Tufnel sub-panel which in turn is mounted on Tufnel pillars, the isolation being completed by the Acetal couplers. A detail not shown in the drawing are the three 1/4" bushes mounted in the control panel where the pot shafts pass through. I got some threaded 6mm bushes from Farnell and reamed them out to 1/4" so the shafts are well supported

Here are a couple of pictures showing the actual progress made:

For those of you not familiar with metalworking, I thought I might describe how I mark out and centre punch hole positions. A hole position is normally defined by the intersection of two scribed lines (the blotches of black in the pictures are marker ink through which the lines are scribed - this makes them easily seen). I then use what is called an Optical Centre Punch. This comprises three components: a conical body with a non-slip base and a vertical hole through the centre, a special lens with cross-wires which fits into the hole and an accurately ground centre punch which also fits into the hole. In use, the body, complete with lens, is placed over the intersecting lines and positioned, using the cross-wires, exactly over the intersection. The lens is them carefully removed and the punch inserted and whacked! The dimple produced is always exactly on the intersection of the lines which means that, if careful when drilling, the resulting hole is in the correct position.

This picture shows the chassis in the BT8121.

Today, I finished smithying the mounting bars and finally tried the B chassis assembly in the cabinet - and it fits . The pictures show it in position.

Bracketry required to support the chassis

Sorry about the screwdriver but it's holding the lid up - need to order some counterbalance springs!

We are progressing and soon, it will be time to start the electronics, the mains transformers are on order - the EHT jobbie is already finished.

I have today cut a plywood escutcheon panel for the B Chassis (TV) side. I was amazed how out of square the cabinet is. I thought that I was the only one to - what's the polite word "compromise" on the meaning of accuracy .

Anyway, I've cut it to size and now have to place the control holes in the right place and to cut the aperture for the 6 pre-set controls. At least, if I can accurately place one control, probably the dual concentric one which is in the middle, I can use my acad drawing to position the other 8 holes. Once the board has been completed, I will endeavour to veneer it in a suitable veneer. Unfortunately, the veneer I have in stock, the leftovers from a Defiant TV job, are all too small and I don't want to make the job too difficult!

July - August 2013

I've just finished a tricky mechanical bit! The two hold controls are in a dual-concentric form but they drive separate pots so I had to put my thinking cap on. I saw from the BT8121 that the "inner" pot was fitted to the front of the chassis with a Z type bracket so assumed that the inner shaft passed though the "outer" pot. So I dismantled the outer pot and drilled a 3.4mm hole right through the shaft and a corresponding hole in the rear cover. This allowed me to pass a 3/32" shaft through the pot. An adapter bush joined this shaft to the 1/4" pot spindle. Then I could fit a control knob to the inner pot shaft after I had soft-soldered a 1/4" bush with a flat milled into it on the top end. The outer control uses a GEC lever-type "knob" which sits under the normal inner knob. I modified a bush onto which the lever knob fitted from the sacrificial radio so that it could be clamped (grub screw) to the 1/4" shaft of the outer pot - the one with the 3.2mm hole drilled through it. Thus, when the outer pot is mounted on the control panel and the inner pot on the chassis front, the 3/32" shaft passes through the outer pot and connects to the inner one. The pictures help to decipher my description.

A bit more progress today (too hot to be outside!). I have started the trial mechanical fit for the cap and transformer boxes I have made (and stuffed in the case of the capacitors). I came across one snag where the flanges from one of the capacitors fouled those of the line-scan output transformer so I had to cut away part of the transformer flange. No real hardship! I guess that the transformers are mounted in another fassion in the real thing but that's a minor detail which does not effect the overall appearance. Here are a few piccies:

One thing I had to do before this assembly was to make a couple of tag strips for the scan transformers. The secondaries of the Xfmrs are up at 6kV so the insulation needs to be very good and they have heavy Ebonite connection blocks. Fortunately, Mike the dismantler had some very useful blocks of Ebonite which I was able to use. I also had some eyelet tags which I bought for the resoration of a Pye 8151 when the mains transformer had to be re-made as a poor substitute had been made. The second two pictures show the tag strip both outside and inside the can - note that the Ebonite passes through the brass can to help to prevent arcing.

A bit of cabinet progress today. I have made the basic escutcheon for the TV side which carries all the TV controls, 6 through a rectangular window and the others via shafts through it

Unfortunately, I have been unable to source the other three + one large knob I need. The ones shown are the only two I have so far. They are used on some GEG radios (several models BT37** and BT38**). Please note the extra 6 "ventilation" holes. These are a result of thinking three times and cutting twice! Normally, I think twice and cut once - I got mixed up this time. The holes will be plugged before the veneer goes on! Then, I have to make a little rectangular lid which covers the pre-set control aperture. The original on the BT1812 shown below gives an idea of what I am trying to achieve.

I need to get hold of some Sapele Mahogany veneer and a small piece of solid mahogany. It will also be fun trying to veneer the finger-chamfer above and below the pre-set aperture! Slowly but reasonably surely.

One thing Till's scan transformer pictures have shown me is that the line scan Xfmr box is smaller than the frame scan unit (thanks, Till). This is, I assume, because the frame scan transformer (secondary DC resistance 14K) has a much higher inductance with many more turns than the line scan (secondary DC resistance 800R)! They both have to have >6kV inter-winding insulation.
I will have to remake the can for the frame scan - I have scaled the size from the manual and pictures.

My next report will be on the electronics/mechanical side where a little has been achieved.

Alas, there is still little progress on the mechanical/electronics side of things - in fact I have gone backwards! When I came to start fitting the valve bases to the chassis, I found that I had got many of the large hole sizes wrong! Some were too large and some too small (this is what happens when things are rushed. I should have waited until I got the actual components). Although the small hole could be enlarged, those that were too big cannot be made smaller (tidily). So I bit the bullet and modified the drawing and sent it in for re-manufacture! Expensive but why spoil the ship for a ha'pth of tar!

I have now primed and painted all the cap cans in battleship grey and they are now ready for mounting on the new chassis it should be here during next week. I have also received the two mains transformers from Ed Dinning and they will be also be mounted in due course. Never having seen under a chassis, I have no idea how the transformers were terminated. Were the connections just wires brought out from the windings or were there tags on them? I think that I will attempt to make some heavy duty tag strips which will be mounted on the chassis via stand-offs so that I can keep the wiring reasonably tidy.

Because of the chassis problem I decided to carry on with the cabinet and today, have started the veneering process for the escutcheons. I bought some mahogany veneer off the web and salvaged some nice solid mahogany from an old test equipment box kindly given to me by Gerry Horrox (many thanks Gerry). The solid wood is to face the rectangular holes for the tuning scale (45° chamfer) and the TV pre-sets. I have ordered a Japanese mitre saw to cut the facings as I saw a guy using one the other day and it gave such a superb cut.

The pictures show the progress so far. Initially, my homemade glue-pot was switched on as soon as I got into the workshop as it takes quite a time to melt the glue (I made the pot out of a very cheap mini deep fat fryer as the traditional glue pots are very expensive). The method I use for veneering is the “hammer” method. The hammer is the tool on the right of the picture and it is used like a very hard squeegee to squeeze out excess glue starting from the centre out. It seems strange but the glue is applied to the base wood and both sides of the veneer as the hammer needs to be lubricated and the glue does this. It also penetrates the veneer and makes it airtight. This helps the atmospheric pressure to press down the veneer onto the base. You can see the plugged holes in the panel. These were the ones drilled in error due to my carelessness mentioned in an earlier post!

A bit more progress in the last couple of weeks. The cabinet work is progressing. I have filled the large speaker hole that the murderer cut in the baffle for his loudspeaker by cutting the hole square and inserting a piece of MDF. This repair was painted black and the new speaker cloth applied using spray adhesive. It looks so much better than the dark Tygan which had been fitted for the "conversion".

The pictures show how it is looking now that I have stained and lacquered the escutcheons (I might have a go at making a couple more later as I'm not totally satisfied with the finish. As an aside, we had a wooden worktop in our kitchen repaired a couple of weeks ago and I asked the very skilled joiner how he would cut the escutcheons to fit the apertures. Easy, he said - just place a board on the top of the cabinet and from underneath, draw round the aperture with a pencil. Providing care is taken, it is accurate and is much easier than making a template (which I did for the radio side).

The new speaker cloth looks much better

The match isn't too bad from a distance

The knobs are on the home made "dual concentric" pots

I will be picking a GEC radio up in a couple of days - it has three correct knobs! Hooray.

I will post a bit more progress in a couple of days, this time on the new chassis. All the valve holders are fitted and the mechanical work done on the first chassis has been transferred to the replacement. Pictures soon.

As mentioned, albeit a bit late, I have taken some pictures of the progress on the B chassis. Apart from "transferring" the work already completed on the first chassis, I have added all the valve holders, the sheet metal screen for the N43 high-level IF stages and, under the chassis, the EHT mains transformer and the three smoothing chokes which are mounted on a vertical plate.

One of the more interesting bits was making custom tag boards to match the pictures of the BT1821 I have. After scaling off the pictures I did some drawings using Autocad and ordered some suitable material - I used Tufnel whale Brand which is fabric based and less likely to split and crack. Fortunately, I had bought some eyelet tags for an earlier project and was able to use them for this project and they look the part. The job was made much easier by using the digital readout on the vertical mill. The Tufnel blank is clamped in a machine vice, one corner lined up exactly under the quill (I use a centering microscope which is installed instead if a chuck) and the digital readout zeroed. Once this is done, it is a simple matter of positioning the table using the coordinates on the readout display to correspond to the dimensions on the drawing. Thus it is very easy to drill a series of equally spaced holes such as those in a tag board.

Unfortunately, this picture shows bad linishing before the chassis was plated.

Just for fun, I decided to complete as much of the chassis as possible, plugging in the valves I have so far and placing the coil cans in position. The picture shows the result!

I hadn't made the tag block when I took this picture

Not too much more to do now before I begin the wiring, starting with the heaters. Unfortunately, the main power transformer had be returned for modification but should soon be back and I can start it then.

I know that this is supposed to be a replica but maybe it shouldn't be called that. I am very near to starting the wiring and, in the absence of the main power Xfmr, I have decided to start with the timebases which are fairly simple and use standard components (apart from the O/P transformers). Not having an original under-chassis picture, I have decided to build as much as possible on tagboards as being the best and neatest way to progress. I will document the layout for an extra page(s) in the manual and to get as logical a tagboard layout as possible. The pictures show some of the tagboards that I will use for the timebase circuitry together with the proposed layout - with the vertical tagboards laid out flat for clarity. What do you think? Should I take this route or will it be too drastic a change from the original (possibly)? Your comments will be appreciated.

Layout for the first 5 tagboards

Partially assembled

One very odd thing about the set is the range of resistor values. Typically there are 125k, 99k, 55k, 990R, 440k, 400k and several others, some of which I have never met before!

I have started the wiring in earnest now by wiring the heaters. There are 7 heater windings on the mains transformer, 4 for the timebases & signal valves, 1 for the sync separator, 1 for the CRT and, of course, 1 for the HT rectifier. The EHT transformer has just the high voltage winding and the heater winding for the EHT rectifier. Once the heaters are finished, I will start with the timebases and perhaps power them with bench PSUs to start with.

Just a quick note to say that I've been looking at Till's layout sketches of his BT1921 which is a console direct-view TV of the same vintage. I seem to have made the right decision about using small tag/group boards as the sketches show that this is what GEC used in the BT9121. One of the really useful drawings is the layout of the EHT bleeder and CRT resistor chain boards! As the circuit is almost identical, by checking with the photos I have, I have been able to tweak my layout so that it must be a reasonable facsimile of the original BT8161.

Progress is steady on the 8161 however and the pictures show how the wiring is going. The long tagstrip will act as terminations for all the heater windings except the HT rectifier (I will make sure that the tags will not carry the full current for the two 6A circuits!) - otherwise, the heater wiring would be rather unwieldy.

The heater wiring is temporarily tacked onto the appropriate tags

The two timebases - Line above field Field.

I have been doing some mechanical work on the IF transformers for the B chassis. I have mentioned the electrical problems with this part of the circuit earlier in the thread but I would like to raise the question again. There are 7 tuned circuits (tuned by self and stray capacity only), and damped with reasonably low value resistors to form the 4MHz IF strip. The table below shows the winding resistances and the damping values. There are no capacitors except the one at the head of the list. This sits at the end of a long coaxial cable between the two chassis.

In order to achieve the bandwidth required, I assume that the transformers must be stagger tuned and be reasonably tightly coupled. The center dips caused by over-coupling perhaps overlap one another resulting in a fairly flat response! Is this possible? What do the experts think? All suggestions welcome.

A bit more mechanical progress this last week. I have fitted the transformers and powered up the HT side (no EHT yet!). The two timebases seem to work, both producing sawtooth waveforms although into a resistive load as I have no scan transformers yet.

Mechanical progress on the IF transformers is pretty good and I have determined the way to mount the screening cans, tags and formers and the pictures show how I have achieved it - the triangular Tufnol panels are copies of some used in a contemporary GEC radio.

The compound picture shows the method and tool I used to make the 6 panels and fix the eyelet tags into the panels. The bottom RH picture shows six layers of Tufnol sandwiched between two steel templates. This allows the 9 holes in each to be drilled exactly in the same place.

The various stages plus the swaging tool.

The former mounted on its termination panel

The IF screening cans are mounted the same way as the valve screening cans

The chassis underside so far (sorry about the mess!)

In order to run the timebases properly, I needed to have the hold and amplitude controls operational so I decided to fit the control panel and wire it permanently to the main chassis as the panel carries. The pictures of the BT8121 chassis shows a cable form passing down the chassis from the control panel to the rear where the timebase circuitry is so I connected long wires to the controls and have started lacing the form together. Before fitting the panel, I wired the EHT cabling from the panel to the EHT tagboards and from the mains transformer to the rectifier etc.

On the original, the cable used for EHT is about 3/8" thick and very unweildy so I looked around for some more suitable modern equivalent. RS have some test lead cable rated at 10kV and only about 3/16" in diameter so I decided to use that. It has turned out ideal and will be great for the wiring to the CRT base, (most of which is at EHT potentials).

The pictures show the progress:

The panel cable form taking shape!

Close-up of the EHT sub-panel wiring

When the panel wiring is complete, I will be able to check both timbase oscillators properly as I have three GT1B thyratrons now (thanks again, Rich).

All this means that I am slowly working towards a display on the CRT, albeit a blank raster but nearly half the battle will have been won!

Today, I completed lacing the cableform from the control panel to the rear of the chassis where the timebase circuitry is, drilling holes for grommets where the cables pass through the chassis.
However, when I came to check where the individual wires went after passing through the rearmost grommet, I realised that I had confused two colours - one of the ones that passes through the chassis at the front and one that goes to the back. So now, I have to unpick the whole form and re-lace it with the cables in the correct places! A little job for tomorrow.

For interest's sake, I use a stepped cone cutter in a cordless drill to cut the largish grommet holes in the steel chassis as it cuts much more gently than a normal twist drill of the equivalent diameter. The resulting hole normally needs a lot of de-burring but it's worth it as it is much kinder to the drill and the hole is much "rounder"!! Also I'm very careful to remove any swarf produced by the machining. It is very easy for bits of steel to get trapped under components and wiring.

When I have re-made the cableform, as I said earlier I will be able to test the scan generators up to the output stages including the operational controls.

I did get the urge to lace properly from my days in the BBC. I remember CAR at TVC when it first opened. The box lacing was something to wonder at and, throughout my days as a TV systems engineer, I was very conscious of the way the wiring in the racks looked - woe betide any wireman that didn't do it properly

September - October 2013

A fair amount achieved today:

The the surface silvered mirror has been ordered from a company called Vacuum Coatings in Walthamstow. They undertake to cut and coat or to strip and re-coat surface-silvered mirrors so I'm having my 703 mirror rec-coated as someone tried to clean it and damaged the coating quite badly!

In addition, the molding to mount the new mirror in the BT8161 lid has been ordered. Nothing available of the shelf so I'm having some custom profile made (ouch!). That's the "office work" done for today.

After cutting away the lacing which I had done, I sorted out the routing of the control panel cables and laced the cableform from the front to the back and then under the chassis. It seems to have worked quite well but I am sure to have forgotten something . The pictures show how it is looking now.

Are there enough components for 2 timebases?

The completed control panel


Now came the time to see if the scan generators worked so I plugged in the HT rectifier, the two thyratrons and the one KT66 I have and switched on, monitoring the HT line. Up came the HT with no sign of smoke or smell so on went the scope. The picture shows the two waveforms - I haven't checked their frequencies, but one is fast and the other slow. The phase difference is because the line scan is after the output valve and the frame at the grid of the output valveholder.

In a previous existance, I used to manufacture (from raw castings and stuffed PCBs) Link Electronics Vidicon cameras (types 108/109. IIRC). These consisted of four PCBs with a loom connecting them together. The looms were laced on a nail board so that the individual wires came out in exactly the right places to meet their appropriate pins on the PCBs. This cut wiring errors down to almost zero and save a huge amount of test time. As you say, Rich, the looms were absolutely stable and would keep their shape whilst being cajoled into position in the equipment.

I don't know if anyone has tried "Spirap" (or some such name). It was (is?) a coiled plastic extrusion which was wrapped around a cableform, with the "arms" passing between the turns, a bit like finger trunking. I found it hopeless as the cableform twisted under the stress of the Spirap and ended up with the arms coming out at all sorts of angles - hopeless. In my opinion, decent lacing cannot be beaten.

I've started on the RF section now as I'm waiting for the first iteration of the line output transformer without which I can go no further with the scanning and display side.

The first thing was to work out how I should approach it. My first thought was to have a small tag-strip for each stage as there are quite a few components round each valve but in the end, I decided that a long strip on the chassis side would make things much simpler by removing the HT distribution and also carrying the gain control bus which controls the gain of the first three stages. The wiring diagram shows the result of my deliberations:

The resulting Acad drawing

And here it is in real life:

A bit like the drawing! The tinned copper wires are the HT feeds to the IF transformers.

The IF transformers (4MHz) present a problem. I have no idea what they were like in real life so it really has to be a bit of calculation and then empirical trial and error. I am thinking that the best way to proceed is to install B4 valve bases in the transformer positions and make some plug-in coil formers using old valve bases and the Acetyl rod which I intend to use in the finished article. It will certainly make life easier and would be a lot kinder to the wiring that is already done.

I have all the input and output capacitances for the IF valves and will add a bit for strays and start from there. As to the coupling between the windings, which will be very important, I will make a special coil former consisting of two bobbins, one of which can slide up and down a central rod. This will make it very easy to "adjust on test" whilst looking at the response curve.

I am off to the Isle of Wight for a few days soon so will be taking a rest from this most absorbing project.

Here's a picture of the first variable IF transformer attempt. The two pictures show the upper bobbin in two positions. I will fit 4 eyelet tags to the Tufnel plate (inside the base diameter), which will be wired to the valvebase pins. The "works" are held in position in the base by a plug of the appropriate diameter which will be different for each valvebase, I would imagine!

Do I need to make 6 of these or say three and work backwards - or maybe forwards. What do you guys/gals think?

I have, at last, found a donor GEC radio (thanks Mike) which just about completes the mechanical wanted list (apart from the large tuning knob) for the TV chassis. I will be collecting a GEC BC3550 at Harpenden. It will donate the last three IF screening cans and, more importantly, the three circular flanges by which the cans fit to the chassis. In the original, the cans were swaged into chassis but copying the way the valve cans are mounted, I have fitted the valve-can mounting flanges over the holes in the chassis and the IF cans are just fitted over the flanges - the valve cans are fortunately exactly the same diameter! .
I have also decided to revise the way the IF transformers are wired into the chassis. I previously made 6 triangular tag-boards on which the transformer formers were mounted but I am going to scotch that idea. Instead, I will make 6 round Tufnol panels, each with positions for 6 tags which will be fitted using the same 2 screws as the flanges. This will ease assembly and enable an easy passage from the temporary 4 pin valve base development assembly to the permanent one.
The other good news is that, when I get back to Reading, I should have the scan output transformers. This means that I can fully test the timebases although I still need one KT66. I will have to connect up the CRT and get the EHT PSU going (I hope not!). I have got some very flexible 10kV test lead cable from RS and that is ideal for the job for the reasonably long runs from the chassis to the CRT base. Unfortunately, I still have the Ekco TSC1113 chassis on my second bench so I will have to make more room on the bench that I am working on!

There has been a bit of a pause in the project while Ed Dinning and I work out what is required of the scan output transformers. These are different to most other scan transformers for two reasons: they drive into what is effectively an open circuit which is the deflection plates and they have to have very high insulation as the plates are at about 5.5kV. We seem to have got somewhere and, after the second iteration, I have a 300V p-p linear line sawtooth which looks as though it could be OK. The frame transformer is being rewound as a result of very reasonable tests on the first iteration example and I should get the second version by the end of next week. Hopefully then, I should be able to display a raster with reasonable geometry!

That brings me to what I consider to be the most difficult bit of all the IF strip. A centre frequency of 4MHz with a double sideband bandwidth of at least 2 MHz. I have no clue as to the construction of the originals except that they are in 2” diameter cans which are about 4” high and their primary and secondary DC resistances. I have absolutely no RF design experience and I must admit, I got a bit lost reading some of the treatises I found on the internet. I decided that what I needed was a spread-sheet which enabled me to vary a number of coil parameters such as diameter and wire gauge to see what effect they have on winding length, resistance, inductance, capacitance and turns ratio. I have uploaded a jpeg of an example of the spread sheet, the yellow boxes being the input cells. Embedded formulae automatically calculate the unknowns which are displayed in the appropriate cells. So far so good! Now I have to decide what to do with the information thus produced to define what I require for the IFTs. Both primaries and secondaries of all transformers are damped by resistors in the low kΩs which will obviously help to widen the bandwidth. However, I am not at all confident in my ability in this area and would appreciate any help offered.

November - December 2013

A bit of progress on the cabinet today - I sent the lid to be professionally finished as there is quite a large flat area to be polished and it's so easy to get that wrong without the proper gear - I know, I've done it! The guy delivered it today and did a little touch-up work on the cabinet and it now looks "the business".

Before fitting the lid back onto the cabinet (16 screws on one leaf of the piano hinge) I fitted the new surface-silvered mirror I had had made by Vacuum Coatings Ltd in East London (I also got them to re-coat the mirror in my 703 Mastergram as it had been cleaned incorrectly at some time and there was a large area of damaged silvering. It looks so much better now!).

In order to fit the mirror, I had to make some molding out of hardwood and stain it to suit the cabinet finish. I even managed to find some 3/4" x 4 brass raised head wood screws to fit the molding! The cabinet outcome is very pleasing and it has put even more pressure on me to get the electronics sorted but there is still a long way to go. However, I do feel that I have also come a long way forward from the original decision to remake a set of electronics so maybe half way there.

A Cossor 15" CRT hiding behind the GEC

Cheating a bit - no electronics behind the knobs!!

The beautiful book veneering on the lid.

At last I've started to "design" the IF transformers with a little help from FJ Camm's Practical Wireless Encyclopedia (12th Ed.) - thanks for the tip, Cathy. There is a design for a TV receiver with a sound IF of 3Mc/s which is pretty close to the 4Mc/s of the BT vision IF amp which is pretty close. Their design uses 3/8" Aladin coil formers placed side by side inside a can, each winding having 50T of 34SWG wire. I just happen to have 12 suitable coil formers so I will use them. However, I need to be able to adjust the tightness of the coupling between the windings and the side by side construction would not be suitable so I will arrange for 2 vertical 6BA studs on which two formers, each carrying one winding, will be mounted using nuts to position them. In this way, I will be able to move them independently. The assemblies will be mounted on the chassis and the screening cans mounted separately. I have already done a test with two formers and it does show promise. The next step is to put the theoretical plans into practice and make a complete transformer - I have enough 6BA studding for about three!
Incidently, the TV design in the PW Encyclopedia looks very interesting. One day ............perhaps

The sound on the BT8161 comes from the "A" or radio chassis and as you know, I will be using a modified BC3750 for this chassis which also produces the 4Mc/s vision IF. If I do get sound first (450kc/s), I will automatically get an IF signal which I can use for the vision IF development!!! So I may just get on with that once I have finished the mechanical design of the vision IFs. It is just a re-design of the frequency changer to take it up to 45Mc/s. The TV band will be in place of the SW as the BC3750 has 3 bands, not 4 and will have bandspread tuning to overcome the inevitable oscillator drift.

Till Eulenspiegel wrote:

With a vision IF as low as 4mc/s achieving a bandwidth of 2.7mc/s is not going to be easy. So does the GEC BT8161 operate with double sidebands or with the upper or lower sideband only? My GEC BT9121 has a vision IF of 6Mc/s which is bit better when trying to achieve the required bandwidth but nevertheless even this low IF must tax the skill of the coil designer. Those monster N43 IF amplifier valves have a much lower anode impeadance when compared with the more normal early television RF pentodes such as the TSP4 and MSP4 so I suggest that's the reason for the choice of this kind of valve.' The low ra. Some details of the N43:

It would appear that the N43 was developed for television work. It has a CV number so it must have been used for industrial and military applications as well.

I can find no reference to whether the 8161 uses double, upper or lower sidebands in its IF stages and, as the transformers are described as band-pass transformers and have no variables, either capacitive or inductive, there are no alignment instructions. I guess, therefore, that I can use the USB which makes the centre frequency about 5.5MHz which could ease the design a little. This is quite close to the 6MHz of the 9121! Comparing the dc resistance of the windings between the two sets however, does suggest that the 8161 uses the LSB as the resistance of the windings is much higher suggesting more turns of thinner wire. I wonder if the reason for the differences between the two sets was the better frequency response of the KTZ series over the MSP/VMP series, the N43 being used only in the final two IF stages driving the full wave detector. Perhaps the MSP41 couldn't handle the voltage swing required! . Some investigation is required.

Till Eulenspiegel wrote:

The HMV 903 and Marconi 704 TV receivers of 1937 were superhets. There is plenty service info including alignment data for these models. I know the IFs were low but not as low as the GEC. I kinda remember that the vision IF was aligned to the USB. This not only keeps the bandwidth down, they were only after 2.5mc/s for this nine inch CRT set, but there was also a large gap between sound and vision IFs.
Disadvantage of using the USB is that if the set is supplied with a vestigial sideband signal source there will be only a limited signal available. This was a known problem when the DSB AP transmitter was replaced by the VSB Crystal Palace Tx. Sets aligned to the LSB were OK.

I must pull the chassis out of the BT9121 in order to take some pictures of the IF transformers It is certain that the construction technique will be the same as those in the BT8161

I'll go for the USB as, as you say, it's easier to get the bandwidth higher up! The photos show the first prototype IF transformers assembled for testing! The windings on the third IFT were done before I had the idea of Aladdin formers and 6BA studding and it will be rewound later so that the coils can be closer together. Obviously I have to determine the right number of turns as well as the coupling between them. I feel that this is going to be a long slog but it will be worth it. The primary and secondary windings are terminated on the tags top and bottom so the primary will be on the top for valves which have an anode top cap and the secondary at the top for the grid top caps.

Till Eulenspiegel wrote:

I've measured the RF and IF coil cans in the BT9121. The RF cans are 2 inches diameter and 2 5/8 inches high. The vision IF transformer cans are 2 inches in diameter and 4 3/16 inches high.
The sound IF transformer can has the same dimensions as the RF cans.
The vision and sound chassis can be separated from the other two assemblies and I think this is what I'll do in order to test and align it

Thanks, Till. Those cans are exactly the same as used by GEC in their late 1930s radios for RF coils which is the source of the cans I will be using. They are usually swaged into the chassis but can easily be removed by cutting through the alloy on the corner of the swage.

The valve screening can mounts that can be seen in your second picture fir the RF cans perfectly so I am using those to mount my IF transformers.

I have been doing a few IF transformer tests using the new construction technique and the first tests give me a bit of hope! I have realised that my Philips sweep generator will sweep from 4MHz to 7MHz which is exactly what I want so I have a good source for my experiments. It's all downhill from here.

Back to the IF transformers:

Looking at the DC resistances of the originals, I can only come to one conclusion - they were wound in Eureka or Constantin resistance wire to keep the Q down. There is no way the I can reach those figures with copper wire and keep the inductance low enough to resonate at 4-6Mc/s unless I use a 6" diameter former . This adds yet another variable to an already confused/complicated formula.

My experiments with the "faux" vision receiver I have built don't seem to be getting anywhere as the first stage doesn't seem to be working so I'm going to fit a standard AM IF transformer and see if I can peak that at 465kc/s to remove one variable factor. When that is working, I can concentrate on the 4-6Mc/s transformers.

Another thought is to start at the end - so as to speak - and work from the detector backwards. The last transformer drives a full wave detector so the secondary of the IFT is centre-tapped. The resistance ratio is about 1:3 so I will wind with a 1:3 turns ratio.

My coil winder is now working and I have a nearly full set of change gears (Delrin) so this should help me but it is yet something else to learn about - it's not as easy as one might think!

As a wide IF bandwidth is required using a centre frequency of 4MHz, perhaps parallel Rs for damping would have been too low in value for the valve anodes to drive properly without limiting so adding some series damping by using Eureka was not so daft: as there are parallel Rs in addition, these could have provided the last tweak, the basic Q being decided during the original design program with the external Rs added after the production performance spread had been determined.

I don't have a definite example of resistance wire being used to reduce Q but I have definitely read it somewhere and Michael, the resistance measurements I gave in a much earlier post came from a manual so could not have been due to corrosion.
Unfortunately, my theoretical ability is not good enough to explain why the series resistance of a coil should be distributed along its total length but there must be a reason why this technique was used. Perhaps someone can work out why!

A bit more progress on the IFT front. I have bitten the bullet and "designed", thanks to articles in a couple of books, a set of transformers - six in all. The spreadsheet shows the details of the windings etc. which were arrived at. I then Christened my Douglas No6 coil winder winding them, although, not having a suitable motor yet, I turned it by hand (I think that I will make this a permanent option).

Now comes the moment of truth - I am going to install them in my IF test chassis to see what happens.

The transformer in the front is the full-wave detector driver transformer with the primary between the two halves of the secondary.

Excuse the mess - I'm a real beginner here!

There are 5 IF amp. valves, VMP4G, 2 x MSP41 and 3 x N43. The detector is full wave driven by a centre tapped secondary on the last IFT.

What I did to get the minimum value for the capacitor was to add 10pf to both the g-k and a-k capacitance of each valve type as it is almost impossible to get below that figure - I realised that the inductor needs to be as high a value as possible.

One thing I am beginning to think is that the coupling between the windings needs to be very tight, even perhaps the secondary wound over the primary! I am half way through some experiments and tomorrow, I will wind an IFT using that principle!

I feel that this is going to be a fairly long journey.

The schematic doesn't show any coupling , either top or bottom, between the coils so coupling must be achieved inductively. I can imagine the coils on their formers, being mounted horizontally on a vertical strip of Paxolin, the coupling being varied by varying the distance between them - Velcro springs to mind for the experimentation phase!

The BT8161 sound IF is 456 kHz (not a typo, the normal radio IF of 456kHz is used for TV) so the centre frequency of the vision carrier is 3.956MHz, the manual rounding this up to 4MHz. The VSB system of transmission started in UK in March, 1956 when the AP transmitters were closed down and the new station at Crystal Palace was commissioned.

I received the Frame O/P Transformer back from Ed yesterday and, after installing it in its box, I lashed it into the set to try it out. This is only the second iteration of the completely unknown transformer and as can be seen from the picture, I now have a reasonably sensible free running raster. The linearity is not too hot but I will leave the final tweeks until I have a modulated raster available. It has prover, however, that the TB stages can provide a full scan on the 12" CRT. I am now making the EHT installation safe with acrylic guards over the 6kV tags etc. There are quite a few on this set and I really do want to live a little longer!!

Till Eulenspiegel wrote:

The final vision IF transformer in the GEC BT9121, the transformer that drives the vision detector.

The former is 29mm diameter. Two two larger windings have approximately 63 turns and the inner winding has 23 turns. I removed the RF section this evening. It is in a bit of a mess, but at least everything is there. It should possible to bench test this unit before it is reinstalled into the set.

Thanks very much for the picture, Till. I must admit that that is very much how I imagined the basic structure of the coils. The DC resistances of the windings are, however, much lower than in the BT8161.

Looking at the circuit, the 9121 detector is different to the 8161 detector (see pic). I originally thought that it was a full wave circuit but it doesn't seem to be. What is the point of the tertiary winding?

Till Eulenspiegel wrote:

I'm back home now. Tomorrow I'll trace out the connections to the coils. It is certainly an odd arrangement. My interpretation is that IF is injected into the video amplifier along with negative going video. R42 is the load resistor and C32 is the IF filter, nothing unusual about that. Surely the designer was not concerned about the input capacitance of the video amplifier valve?

Tomorrow I'll check out the value of R50, the cathode bias resistor. Because the input video is negative going and the coupling is direct R50 should have a low value.

Here's the GEC description of the vision stages:

Till Eulenspiegel wrote:

The BT9121: Further examination of the receiver chassis reveals more horrors. That old GEC bete noir valve holder troubles has manifested itself. The three B7 holders of the KTZ41 IF amplifiers are falling to pieces. Lots of leaky waxies as well. This resto job is going to take a long time. I'd reckon the cabinet rebuild was the easy part.

Your latest picture is very useful as it shows that the IFTs are wound on large formers - I thought that they would have been.
It looks as if a fair amount of work has been done on the 9121 after the war - those Eire resistors are a bit too modern to be original - or are they!

Till Eulenspiegel wrote:

The BT9121 TV: These B7 valveholders were removed from the Philips 274A radio I'm rebuilding.
They'll be fitted in the GEC TV.

Till Eulenspiegel wrote:

Some info about the frame sync transformer. The transformer laminations appear to be 12.7mm X 47mm and there appears to be 35 of them in the stack. A standard part perhaps? The close up pictures show the method of consruction.

The sync separators of the BT8161 and BT9121 are very similar. The timebases are the same except in the BT9121 there is a series inductor and capacitor between the grid and chassis of the frame oscillator thyratron GT1B


Hi Till. I noticed that the two GEC TVs shared virtually identical scan circuits (except for, strangely, the line output transformer) and sync separator. The sync transformers (which, I think, acts as a low pass filter and integrator) are identical and of fairly high impedance (secondary DC resistance 3,600ohm).

The series inductor in the BT9121 must be a second integrator to aid the separation of the frame sync from line - possibly a problem with the BT8161. I guess I will see what I can do with winding a transformer with my new coil winder using an old output transformer or even the "Is" from a small mains transformer.

I think Till has the right idea with his BT9121 restoration by getting the sync separator going - at the moment I am a bit stuck with the IF stages so perhaps a little break is in order and, as I have a raster, I will now get the sync separator going and even get some modulation onto the CRT!

Till Eulenspiegel wrote:

Revision notes about the sync separator: Note the direct connection from the anode of the video amplifier to the grid of the sync separator. There is no coupling capacitor. The video waveform present at the grid of the sync valve grid is positive with negative going syncs. Critical adjustment of the operating conditions of the valve ensures that the valve is cut-off during the syncs period resulting there is positive sync pulses at the anode. My take on this circuit is that the sync valve is driven in to saturation during the video period, but due to the nature of the design there must be some video present at the anode. We'll see if that is so when I get the RF/IFs unit working. It will be possible to bench test it before returning it to the main assembly which also consists of the timebase, power supply and the CRT.
A more satisfactory sync separator is the more familiar circuit in which positive going syncs clamp the pulse tips to near ground potential. The valve is cut-off during the video period. To employ this circuit a video inverter stage would be required. However, this type of sync separator produces negative going syncs at the output which are no use to trigger the thyratrons. The GT1Bs need a positive sync pulse to discharge the ramp forming capacitor.

I agree with your analysis of the circuit, Till. The reason for the Sync Lock pot is to set up the cathode potential to clip off as much of the video as possible while leaving worthwhile sync pulses at the anode (positive going, as you said). The sync transformer is perhaps to isolate the line and frame timebases -they are supplied from separate HT smoothing circuits with a separate choke for the frame. I have attached the relevant parts of both the BT9121 and the BT8161 to show the differences.

The BT9121 circuit

The transformer secondary feeds the thyratron grid via what I assume is an integrator R112/C87. The inductor inhibits the line sync pulses from charging C87. Also, there is a feed, via C86, direct from the sync separator anode. Is C86 small enough to differentiate the pulses - I have no values? It could be used to sharpen the triggering edge, perhaps?

The BT8161 circuit

As can be seen, the BT8161 is somewhat simpler than the BT9121. The transformer again feeds the thyratron but this time through a simple RC integrating network R80/C62. There is no inductor involved and no direct feed from the sync separator anode. One wonders whether GEC had trouble with the BT8161 and introduced the changes to overcome these difficulties - interlace, perhaps?

Till Eulenspiegel wrote:

I notice that in the BT8161 the anode and screen grid decoupling capacitors of the N43 video amplifier are electrolytics. In the BT9121 that role is performed by a paper capacitor block.

It's certainly true that in many pre-war TVs the designer paid much more attention to isolating the sync drives for the line and field oscillators. It is curious that the cheaper BT9121 has the refinement of the L & C filter in the sync feed to the frame oscillator. On the sync feed isolation subject what comes to mind the extraordinary "split anode" sync separator valve employed in the Cossor 1210 and model 54. Of course the 4TA did not have a split anode, it consisted of two identical pentodes in the same bulb.

What we need to see is the circuits of the low cost GEC vision only receivers to see what economies had been effected to keep the sellng price low.

I have made a set of coils to your table except that the detector transformer is not bifilar wound, a technique which I will have to experiment with once I get used to my coil winder. The picture shows the coils ready for mechanical assembly following Till's pictures of his BT9121 and installing into the test chassis. I also have to wind a sync. transformer which I will do shortly.
Here is a picture of the coils:

I have, at last, got some results from the prototype IF strip, a great way to welcome in the new year!
May I take the opportunity to wish everyone a happy new year for 2014, the year of first modulated light for the BT8161 (say!!).

I built the 6 IF coils I wound into my test IF strip (see picture) and did some tests. The signal is generated by a video test generator and is a frame rate sweep from 0-4MHz. The markers are at 1, 2, 3 and 3.89MHz.

The first three pictures are reasonably satisfactory but after that, there are serious problems.

The test IF chassis with the first four IF coils installed

The input to the first stage with markers at 1,2,3 & 4MHz. Output attenuator at 0dB

g1 of second stage - -20dB output attenuator at -20dB

g1 of third stage - -20dB output attenuator at -20dB

g1 of fourth stage - -20dB output attenuator at -20dB

Anode of fourth stage (resistive load 4k7) -20dB output attenuator at -20dB

I would really appreciate some input here. There seems to be a strange "inversion" mid-band in the third stage which is not due to overloading - what is it?

January - February 2014

The IF amp has been put, for a while, on the "nearly too hard" pile. A friend of mine is looking into the problems I'm having and we'll have a chat about it in the near future so what next?

The sync transformer beckoned so I've had a go at that. Thanks to Till, I have some pictures of the original in his BT9121. The manuals give identical resistance readings so it's safe to assume that there are one and the same part.

I took a smoothing choke off a gash GEC PSU chassis kindly donated by Trevor (Murphyv310) and was delighted to find the "I" stack was of an ideal size. I drilled a couple of fixing holes through it and selected the correct height of lams required. In order to fit it onto my coil winder, I made a couple of holders which can be seen in use in the first picture.

The coil winder in use winding the secondary.

The completed windings and the "chucking" jigs. I chose a 1:2 ratio!

I then made a tag strip using 2mm Tufnel and a 1.5mm steel mounting bracket. The hexagonal pillars and nuts were custom made to make it look really authentic and I thing it has worked although I might add some washers to increase the spacing between the tag strip and the windings.

I should have mounted the tags at right-angles!

The next job is to fit the transformer and try driving the sync separator but as mentioned before, a large positive going video signal is quite difficult to produce without building a special amp so I will have a go at that. To make it really useful, it should produce variable +ve or -ve signals up to about 30-40Vp-p from a standard 1V 70/30 video signal (from the Aurora for example). There is no need for a bi-polar PSU as the output will always be AC coupled (with adjustable DC restoration?). Does anybody know of a design? When completed, I will make this a bit of permanent test gear as it could be very useful for fault elimination in sets undergoing repair or restoration!