CT100 Convergence Transformer Solid State Replacement Update
 

Hi All. I had taken a year long hiatus from this project as there had been a lot of things going on. I began revisiting this last month and spent time repairing my old Tektronix FG504 Function generator to provide an experimental test sinal for this project. It generates 30v p-p which matches roughly the signal required to feed the convergence transformer.

Here is the link to the last posts a year ago.
https://videokarma.org/showthread.php?t=276331

I left off a year ago with partial success. The circuit improved convergence but the parabola swing was not of sufficient amplitude. The limiting factor was the 400v B+ supply voltage which I was using. I figured if I could find convenient way to obtain a higher B+ to run this circuit I could make it work.

The Power MOSFET I was using had only a maximum Drain Source voltage of 800vdc. I needed to obtain something which would work at a higher voltage.

This past week I cascaded two HP high voltage B supplies to obtain 800vdc. I also obtained an IGBT (Insulated Gate Bipolar Transistor) which is an odd hybrid between a mosfet and a standard junction transistor. The one I have been experimenting with is a IGW15T120CR. You may easily find the specs with a Google search. It has a Vce maximum voltage of 1200v and a phenomenal transconductance of 10S or 10,000,000 umho! (To me a transconductance that high is creepy, especially when dealing with early vacuum tubes like the UX201A which has a transconductance of about 650umho on a good day).

Anyhow, I have been testing the past two evenings on the bench. I was previously obtaining a linear amplification of about 400v p-p which was just barely insufficient to obtain the vertical convergent at the top of picture. The past two days with the 800vdc supply, I am able to obtain linear amplification to 725vp-p. I am tusting this will leave more than ample headroom to prevent distortion of the parabola convergence waveform.

The last item was where to obtain the 800 volts. I had previously considered a separate power supply or tapping off the boost B+. Neither solution appealed to me, Then I turned to the CT100 power supply itself. The Voltage doubler used in the CT-100 can with the addition of two diodes and two capacitors can and to it a voltage quadrupler circuit. I tested it and it works fine with output topping 800 volts. The IGBT collector current is only about 2mA so the transistor only gets mildly warm without need of a heatsink.

Please find my attachments of the circuits. Also the scope output in two screenshots because the lowest sensitivity of the scope was 50 volts per div for a full vertical screen of an only 400v span. The 727vp-p is covered in the two screen shots of the upper and lower half of the waveform.

Glad you were able to pick this up again, staying tuned for further reports!

This evening I have spent time consolidating the plan. The CT100 convergence transformer replacement I plan to fit in place within the box which contained the old transformer. It's purpose is to couple and step up the vertical convergence parabola waveform to feed the 15GP22 CRT electrostatic convergence plates. The convergence plates run at 3000volts dc with the parabola waveform to converge the beams at the top and bottom of the screen. The amplitude of the parabola must be about 500 volts p-p. The circuit I have designed I hope ill fulfil the vertical convergence in a more cost effective way without having to resort to using a specially constructed transformer.

The previous drawings were mainly for concept and included a few errors. The latest iteration I have attached here has the previous error fixed. This has only been tested on the bench and not yet connected to the CT100.

The 800 volts dc to provide a convergence waveform of sufficient voltage amplitude is incorporated into this drawing. Everything is self contained and it will not require any modification to the CT100. All the connections on the box go to the original connection except for one: an extra lead identified as "To Main Rectifier Junction" is to the connection where the two main rectifiers are connected. This connection facilitate the boosting to 800 volts to operate the IGBT, This will require a wire to be run from this convergence circuit to the main rectifiers.

Connected up the breadboard circuit to the CT100 for a test today. The wider voltage swing was sufficient to bring in convergence. I can get a parabola amplitude of 750 volts p-p.

The using the voltage quadrupler extra two diodes and electrolytic capacitors to augment the ones in the set to boost the B+ to work my circuit worked very well. Too well in fact because on power up the voltage rises to 950vdc before it settles back to around 850vdc. I measured the peak voltage across the two 10uF electrolytics in may circuit and it rises to 485 volts on power up before settling back to about 425vdc each. I will have to study the surge rating of the capacitors. Might end up stacking two pair just to be safe.

I found I had too much gain. In the drawing I posted a few days ago I forgot to include the series 22k resistor to the gate. I raised the 22k to 56k. I also reduced the input coupling capacitor to 0.047uF.

I will run it for a few days in breadboad form before building the circuit to mount in the original transformer box.

I am happier with this approach as the components are cheap, it requires no modification to the set apart from running one additional wire to the set's voltage doubler, it only uses 2mA current and I don't have to deal with that fragile convergence transformer.

The picture of the crosshatch was at initial power up with only a preliminary touch of the controls. The subsequent off air photos were after further convergence touch ups.

I am finding the convergence perhaps better before the transformer failure which suggests it was destined to fail.

Here is the latest circuit diagram. The diagram is missing an 820k bleeder resistor to bleed off the 850v supply more rapidly at switch off.

Convergence Circuit
 

Lookin good brother! I have one with an open section where the focus voltage is fed through- I jumped it with an equivalent DC resistance as the open winding and it works but has poor edge convergence.

Thanks,
Kirk

Yeah I hear you. That happened to me and led to a sudden breakdown. Fortunately I caught it before burning up the focus pot.

I am running the set right now and it is looking very good. I am finding the focus is more consistent and I think the focus drift was due to the failing transformer.

Cost of parts is about $20 do this.

Only real concern I have at the moment are the 450v electrolytic. When I switch on the set about a half hour ago, the voltage across them starts at 475v abd settles back to 450v where it is now. Tried looking for 500v electrolytics at Digikey and Newark and they are not available. Digikey had a 26 week lead time and you had to buy 1000 units! I have decided to put the capacitors in series as I want to put bleeder resistors in anyway

I am happy with result and will proceed with the proper construction of it to fit in the original can. It will in effect be a direct drop in replacement and feel ultimately will be more reliable that the the original transformer.

Looking good.

Focus drift with scene brightness is a universal problem with the stock CT-100 even if the transformer is good (based on the look of mine I think the modern replacement transformer John Yurkon? made was installed by the last owner). Another member (Tom Albrecht IIRC) posted a modification to replace the fixed resistor between the focus wiper and ground with 15 200V 1W Zeners in series. I modified my CT-100 and it stopped the focus drifting with screen brightness...15 ended up being too many for me as peak focus was past the end of the control. I ended up reducing it to 12 Zeners so peak was somewhat centered on the pot.
I wonder if this solid state transformer replacement will work on the Westinghouse H840CK15?... I've got one that works but has questionable older repairs (haven't looked at its transformer yet) that I would like to get to and finish up in the next year or so.
This project is a definite asset to the community.

I looked at the Westinghouse schematic. The convergence transformer is similar in concept and the solid state solution would replace T702 and T701.

The Westinghouse uses a 5U4 Full wave rectifier. In order to get the higher voltage to provide the amplitude of the parabola, the extra wire would connect to one of the 5U4 plates. The circuit I have devised will double the voltage to obtain the 800v to 900v to run the circuit. So it could be made to work and would likely just drop in.

Below are some snapshots off the scope of the parabola. The scope minimum sensitivity is 50 volts per division or 400v vertical full scale. The amplitude is adjustable with the CT100 vertical amplitude and shape controls. I have adjusted for towards a perfect parabola. I found optimum convergence is with the parabola adjusted to be a little distorted. The peak amplitude is adjustable linearly without distortion up to about 700vp-p. In the Parabola 3 snapshot, I reduced the amplitude amplitude via the control briefly to show the parabola fitting the scope display.

Note the wide trace is due to the added horizontal modulation so this signal riding on DC roulghly 4.5kV is going to the convergence deplection plates.

I have been running the CT100 the last couple of days with the my circuit transformer replacement. The signal amplitude appears more than ample to achieve full convergence.

The snapshots shows the progress made. Without the transformer removed and Verticak Convergence circuitry bypassed, convergence at the bottom and top of the screen could not be achieved. The first snapshot from 2024 showing the results of my first iteration, depicts bottom convergence made but top still out of convergence. I attributed this to insufficient waveform amplitude as I relied only upon the 400v dc supply. The last photo shows the top and bottom convergence made using the 850v supply. I think I can get the convergence better with a bit more tweaking and after I repair the intermittent 5Mohm Vertical Convergence Amplitude pot.

I have tried the circuit with a small variety of IGBTs and MOSFETs with tranconductances from 2mhos to 10 mhos and the results are similar.

I made minor variations to improve performance and reliability. The surge voltage on the 10uF 450v caps was about 490v, with is typically the limit for short term voltage peaks. I opted to but 500v electrolytics which I will use in the final circuit. I raised the 3.9 Megohm bias resistor to 4.7 Mohm to move the quiescent point up to prevent the MOSFET from saturating at full parabola amplitude. And lastly I included a pair of back to back 20v zener diodes to protect against gate source over voltage. This is an option and I thought about adding the zeners after I blew up an IBJT and MOSFET due accidentally miswiring the 47k gate to ground resistor and then powering up. (I made the same mistake twice!) I don't see this as a problem because ince correctly wired the gate should be pretty well protected.

Further update. Aiming to get the overall convergence as best as possible, these are the adjustments I made.

1. Measuring at the MOSFET drain set Vertical Conv. Amplitude to around 400v p-p
2. Adjust Vertical Ampl. Shape to make best parabola.
3. Adjust Vert Conv. Amplitude to 800vp-p to the point clipping is about to occur. (Minor clipping in the center or minimum amplitude is not detrimental to the convergence).
4. Touch up Vertical Amplitude Shape to bring in convergence at top and bottom of screen.

I found that by increasing the output coupling capacitor to 0.03uF provided enough to bring into best convergence.

The power supply even at 850v dc is only just enough. I considered including the quintupler to yield about 1200v but I am happy with how it currently performs. Focus and convergence is really good.

Judge for yourself I fiddled with the convergence for a bit and could perhaps make it better. But it is good enough for me.

I will begin packaging it up into the old transformer metal box. One I have this in place I can address the AGC and full white blooming resolution. These items I will address in separate threads.

See the attached off screen photos and I welcome any criticism. I am curious how good the convergence is on other CT100s.

Just Radios in Canada stocks 500v electrolytics,

Thanks! I will look into that.

I ended up going to DigiKey. Shipped from Florida and with some other items. Shipping was only C$8 via FedEx. Ordered Friday afternoon and items arrived Monday morning to my door. Not bad!

Been busy on CT100 this past week. Successfully installed my circuit permanently in the CT100. Convergence is very good if not better than the original transformer.

(Was delayed unfortunately by not one but three totally unrelated issues that required a bit of sleuthing! All part of the fun and more about that and related issues later.)

For the construction and installation of my solid state replacement unit, I shaped three perforated boards to fit in the transformer can. They are stacked and held in place with stanchions that would neatly fill the box. The picture describes my construction.

The tapped secondary of the original transformer connects to three points: to the convergence electrode, the focus electrode and the focus control 3000v source. I replaced the tapped secondary with a resistive divider. The focus control may be considered as signal zero or return so the capacitive coupled output of my circuit connects to the opposite side of what was originally the transformer secondary with a divided down parabola output to modulate the focus. For increased reliability, I saw no need to bring these three leads carrying 3000v into the box and so mounted the resistive divider consisting of 2w 560kohm and 330kohm resistors across the three points previously connected to the secondary.

The leads from the box are yellow the capacitively coupled parabola output, red to +400v, blue to the 12AU7 vertical convergence amplifier, black to ground and white to the diode junction of the main rectifier doubler. The white wire is to feed the internal doubler circuit to effectively create a voltage quadrupler to provide the +850vdc for the mosfet parabola amplifier. The white wire I discretely ran with the set wiring along the rear chassis underside and up to and after the main 400 volt rectifier fuse.

After installing I was concerned about the heat dissipation of the load resistor and the resistive bleeder to discharge the 10ufd capacitors. The resistors get only slightly warm but after an hour the box was reaching about 40 degrees Celsius. There was ample room to mount the resistors on the rear panel externally and had the added benefit to allow me to scope the 750vp-p parabola and measure the 850volt supply.

Below are some photos of the final install. The diagram I will add to the next post.

Here is my final as built diagram. I left out the 620kohm bleeder resistor which is connected between the +850v supply and ground.

Fantastic! Is the semiconductor in the final version the IGW15T120CR? You did a great job of engineering a circuit that was not possible in 1954. I will build it and let you know how it works on the CT-100 that I am working on. Too bad John Folsom is not on here anymore to see this- he struggled with this problem 20+ years ago and even had some replacement transformers wound. I tried one of his and it was "OK" but not good enough to get a perfect edge convergence- Your circuit solves that and the focus drifting that was inherent with this design- thus the focus and convergence HV controls were made to be customer adjustments - EEK! Thank you for your time and effort on this-

In my final design I used a 1200v MOSFET. I used the IGBT as well and they both perform similarly. The bias might have to be changed slightly. The best way is to just build it and turn the parabola amplitude up and you should reach close to 800v p-p without clipping.

The edge convergence is the problem. Best to start with a symmetrical parabola and only vary the amplitude at first. All the convergence controls interact and I had to patiently circle over and over all the convergence adjustments to get best convergence. But edge convergence pretty good now.

Had a set back tonight as a second metal film replacement cap appears to have failed. Earlier the 0.001ufd cap across the vertical output primary failed. It was a 1600v unit. I replaced it with a 2000v cap and after working for a couple of days I have a compressed distorted scan. Will have to figure that one out tomorrow.

I found the problem with the video sweep. I will mention it here as it may affect others.

While adjusting the convergence, I found the Vertical Conv. Amplitude control would not adjust linearly. So i removed the control to more closely examine. Testing with a VTVM, I saw no problem. I replaced the control and found when powering the set the vertical sweep was non linear and compressed.

Disconnecting the 0.22 uF capacitor connected to the vertical amplitude shape control I measured only 6kohm to ground. Looking at the circuit diagram, it should have measure minimum about 7 megohm. The 0.22 was shunting the output of the vertical oscillator to the vertical output grid. Moving the potentiomenter lead, the resistance jumped to anbout 7 megohm. Replaced the lead powered the chassis and the vertical output signal still compressed. Measured the vertical amp. shape pot lug to ground, the resistance was again 6 kohm!

I carefully washed and brushed with an artist's paintbrush the area around the pot connections. Problem gone.

It looks like 70 years of grime and the the high circuit impedances (around two 5 megohm pots!) caused the intermittencies. The pots looked quite clean and I was not as scrupulous as I should have been around the pot.

It was following the horizontal hold control exhibiting the same leakage problem suddenly last week. The control runs at about 200 vdc above ground and a minor leakage made the the oscillator control section of the 6SN7 to pull the horizontal oscillator off frequency. Odd as both these pot problems happened suddenly without warning.

Convergence Update.

I have run the CT100 about 24 hours and the convergence remains very good. It is actually better than the transformer that was in the set before it failed.

The transformer failing first sign was the drifting focus and convergence on warm up. With the convergence pot fixed, the convergence is shows minimal drift from warm up from cold. Also the focus drift is very much reduced.

I used three 10kV .01 uF capacitors in parallel in my final circuit although I earlier used 0.02uF (two in parallel) worked fine.

I also noticed an error in my earlier as built diagram: the 0.018 uF coupling between 12AU7 and IGBT should have read 0.1uF.

With regards to the device I used I was mistaken. It is a 1200 volt IGBT type GF3NC120HD. They are available at Mouser for $1.62 each. (I would be inclined next time to buy 10 for $1.02 each).

https://www.mouser.com/ProductDetail...8Tvd683-o9yPh_


The 0.01uF 10kv capacitors I bought on eBay:

https://www.ebay.com/itm/32496656125...F+10kv&itmmeta

Lastly is attached my most recent as built diagram.

Doodling with the CT100 schematic, you can see the connection to the main power supply and the circuit inserted within the circuit itself in the attachments below.

I am highly impressed with your work on this tough issue, and your good results and documenting and sharing everything. Thanks!

Thanks Chris.

After earlier toying with a transformer rebuild and pondering the focus and convergence high potential had to traverse over 2 /1/2 miles of AWG #44 wire to reach the CRT with the strict insulation requirements, it seemed appropriate to examine an alternative approach requiring capacitive coupling. My alternative approach is much simpler and cheaper and in my opinion works better.

I have been running tests. I have gotten the convergence best I can on my tube. It matches what I was able to get prior to the transformer failure. It has the benefit of less drift: from switch on to warm up there is very minor drift which is due to the 12AU7 coming up to temperature. The convergence becomes optimized at about 4 minutes and holds well after that. I have been running the set for two to three hours at at a time and the convergence box remains cool.

Here is a snapshot of the crosshatch with the convergence about the best I can get. I could never get the lower corners as good as I would like but as it is a rebuilt tube and I am extremely fortunate the tube is still good and appears to have a lot of life left, I am extremely grateful with what I have.

Great work, I’d say the pattern is excellent. Now show us some screenshots. We know what a CT-100 looks like, so photograph in the dark showing the screen only. You will have maximum contrast that way.

I need to do Color purity and Greyscale again. But here are four random off air shots. Via a Zenith coupon set top box circa 2005.

Excellent. Soft and pastel. Very pleasing. I put your last photo in my calibrated Adobe Lightroom editor and shows your photos are spot on. A small purity issue on the left edge, otherwise very clean.

https://visions4netjournal.com/wp-co.../IMG_3259.tiff

Proceeded with purity greyscale adjustments. And then had to redo convergence. Getting the corners is a real challenge especially with only a few dynamic controls and interactivity between all the controls.

Here are some more snapshot from a few minutes ago. My circuit is holding fine and the convergence and focus drift from off to warm up is now inconsequential.

Fixed the AGC issue, couple of intermittent pots. Blooming isn't bad but to get the best picture need a dimmish room.

Excellent work! I better get to it here- I will let you know in a few weeks how your circuit worked for me- Many Thanks!

Excellent!

Looks pretty darn good. I submit that very few of these sets were that well adjusted "Back In The Day". Attention to detail; looks terrific.

I was not content with the edge convergence so I conducted further testing. This has led to some modifications which resulted in better overall convergence. I think overall it is now better than I ever had with the transformer.

I added a 220kohm resistor in series with the output and changed the external resistors to 1.0Megohm and 620kohm. I also changed the coupling capacitor from 0.1uF to 0.47uF.

Convergence on old color sets is a process of going around around and around making marginal improvement which interact with other adjustments. Creating accurate parabola waveforms with crude circuits like this is often an exercise in futility. I am happy the new circuit has the range and control to fix the top and bottom convergence with a reasonable interaction with the horizontal parabola.

Anyhow this is likely my last post on this subject and I am very happy with the result. This proves to be a viable cost effective and technical solution to the vertical transformer problem.

The CT100 has been running for about four hours straight and apart from minor convergence drift from switch on for 10 minutes, the convergence and focus has been stable.

Please find attached the documented modifications plus a couple of screenshot I took this evening. I have also included photos of the installed transformer. the color of the wires follow the diagram.

:thmbsp:

Excellent work! Your screen shots are very impressive.