CT-100 B1001599 Failed Convergence Transformer

[Penthode]

Originally Posted by Doug

one guy that was a member on arf has
My friend John Folsom says, 'It's all about the spares.'"
-Bob the Antique TV Guy in 2021

so
good luck

Yes, it is all about spares. But as time marches, it it all about creativity and much of the time making spares!

I was looking at the high voltage MOSFET option to replace the transformer. It is a way lot cheaper (parts under $10) and maybe more efficient and less prone to breakdown.

Looking at the FETs these days, they have a very large transconductance so with such a light load, the gain should be achieveable with a single moderate power MOSFET.

I have thrown in some values for a generic MOSFET to achieve the bias voltages and quieesent point of around 225 vdc to allow for a wide swing. I anticipate average vertical convergence level will only be in the region of 50 to 100 volts p-p to the convergence plate. But that is just a guess.

Find updated my circuit with some component values. In the drawing where my circuit is inserted into the CT-100 circuit, I left a gate resistor with no value to provide the option to reduce gain.

[Doug]

Last part of my post didn't come thru
https://antiquetvguy.com/

Your second images doesn't work

Doug

[old_tv_nut]

@Penthode, are you sure the amplifier output will be the right polarity?

[Penthode]

Originally Posted by old_tv_nut

@Penthode, are you sure the amplifier output will be the right polarity?

Haha. I have thought about that. I should find out shortly.
My guess is that perhaps the parabola inverts depending upon the position of the CT100 convergence control? Eg the midpoint there is no correction and one control extreme the parabola is inverted from the other extreme? I have nowhere read of a reported phase error when wiring the primary of the earlier replacement transformers or if any attention was paid to phase. It would appear logical to have a plus/ minus correction assuming when the 15GP22 was constructed, the convergence was close to begin with and that the dynamic correction was to correct for minor physical discrepancies.

If it does need to be inverted I could add an inverting amplifier stage.

Maybe I will be lucky? Unfortunately there are no sample waveforms to reference. It will be interesting to find out and dig into this more deeply.

[old_tv_nut]

The customer control is DC only, so it would be set for proper convergence at the center.

The H and V waveforms will have a specific polarity with variable amplitude because the beams must converge at a sharper angle at picture center than at the edges due to the geometry of scanning a flat surface.

I also note that the transformer wires have different colors on each end of the winding, which would indicate polarity was important.

So yes, it appears you have a 50/50 chance of needing an extra inverting stage.

Following your progress with baited breath!

[Penthode]

Originally Posted by old_tv_nut

The customer control is DC only, so it would be set for proper convergence at the center.

The H and V waveforms will have a specific polarity with variable amplitude because the beams must converge at a sharper angle at picture center than at the edges due to the geometry of scanning a flat surface.

Yes I have thought about that too. I found that the later color faceplates which have a curvature, I believe the deposition of the phosphor was aligned through the faceplate itself. This would mean that the dynamic correction would only need to be minimal. I thought the flat faceplate in the 15GP22 would have been manufactured similarly so as to minimize dynamic correction suggesting only a plus/minus minor correction would be needed.

I am curious now to find out. I plan to pick up a MOSFET tomorrow and externally breadboard it to measure waveforms. I can also try the chassis with a plate load resistor on the 12AU7 to examine the waveform and the dynamic vertical control range waveforms to see what will be input.

Fingers crossed!

[old_tv_nut]

Only the DC convergence of a CRT needs to be plus/minus. Because the radius of curvature of the screen is/was always larger than the yoke to screen distance, there is always a positive polarity dynamic waveform required. In other words, the screen edges are always farther from the yoke deflection point than the screen center is. Much later designs that simplified the convergence were based on deflection yokes with a center of deflection that varied with the direction of the beam.

[trinescope]

Somewhere I have the winding data for the John Folsom replacement parts. The details seem to match closer to the Westinghouse 15" transformer, but the basic operation is the same: 1:10 step up ratio with a secondary having a tap around 1/3. The winding resistances are not critical here because both windings carry very little current. The problem with the original parts is the same as all the vertical output transformers going bad now; the older materials have not held up and are breaking down under voltage stress.

[Penthode]

Here is the results on my first test with the single MOSFET. Yesterday I bought an IRF830-1 500v 1A MOSFET locally. It has a forward transconductance of minimum 2.5 Siemans which is 2,500,000 umho. That is a whopping lot of gain available espectially when you are used to dealing with only a few thousand umhos with vacuum tubes!

With the single FET, there was too much gain. I added a 330k resistor on the input side to brring the gain closer to 11 times. I was able to achieve 350 v p-p on the output with the input only 20 v p-p. (My HP 204 Sine Generator maximum output).

I found non-linearity (no clipping) became noticable at 250 v p-p. You will see the output trace slight distortion in the photo.

I found I could slightly reduce the non linear distortion by increasing the drain current slightly, hence the reduction of the source resistor from 5.6 kohm to 4.7 kohm.

Note the phase shift. Testing the circuit at 600Hz the phase inversion is exactly 180 degrees. At 60Hz as depicted in the photo there is a shift. I tried bridging the coupling capacitors which did not make much difference. I believe the shift may be due to one of my scope probes. I am investigating this.

the circuit was tested as per the diagram including the 27k 12AU7 plate load and the 560k and 330k load resistors.

Next step is to put the CTC2 chassis on the bench and perform some measurements.

[old_tv_nut]

A couple of comments:

Phase shift at low frequencies could occur if your scope is set to AC coupling. If you have been using AC, switch to DC to double-check.

The 100 microfarad source capacitor is a 27 ohm impedance. It's no wonder you have too much gain, and the circuit is non-linear. I'd suggest bridging the 330k, removing the 100 microfarad, and changing the source resistor from 4.7k to about 2.7k. This may require adjusting your gate bias, which I have not calculated.
This will reduce the gain dependence on the FET transconductance and linearize the circuit. Of course, eliminating the 330k will mean more loading on the preceding tube circuit, so the gain there may be reduced, requiring some increase of your FET amplifier gain. I have not gone back to the original circuit to calculate the loading of your amplifier compared to the original circuit loading.

[Penthode]

Originally Posted by old_tv_nut

A couple of comments:

Phase shift at low frequencies could occur if your scope is set to AC coupling. If you have been using AC, switch to DC to double-check.

The 100 microfarad source capacitor is a 27 ohm impedance. It's no wonder you have too much gain, and the circuit is non-linear. I'd suggest bridging the 330k, removing the 100 microfarad, and changing the source resistor from 4.7k to about 2.7k. This may require adjusting your gate bias, which I have not calculated.
This will reduce the gain dependence on the FET transconductance and linearize the circuit. Of course, eliminating the 330k will mean more loading on the preceding tube circuit, so the gain there may be reduced, requiring some increase of your FET amplifier gain. I have not gone back to the original circuit to calculate the loading of your amplifier compared to the original circuit loading.

Thanks. Excellent. I will make your suggested changes and repost.

[Penthode]

I removed the source resistor bypass 100uF. I then chaned the bias resistors to raise the input resistance and keep Id at 5mA.

The generator was in series with the 27k simulating the 12AU7 plate impedance. The 330 kohm resistor was removed.

The gain is now just a hair under 10 times which should be close enough. The phase error was due to the source bypasss capacitor which is now gone and the remaining is the AC scope coupling: switching to DC coupling on the input reduced the phase error by about 1/2. I cannot remove the output AC coupling because the drain is at 200 VDC. So I believe the phase error has been acounted for and is not in the circuit.

I have insufficient input signal (it was only 15 v p-p) so I cannot check linearity at the higher output voltage. Although removing the 27kohm input resistor pushed the output to 200 VDC and the sine was perfect hence the non linearity appears to be no longer there.

[Penthode]

I had made a minor error in my second implentation. The Drain resistor in the diagram states 47k and I had inserted a 33k resistor. As drawn for the second test with the correct value of drain resistor, the output voltage is 200 v p-p with no evidence of any nonlinear distortion. That is a gain of 13 times which should be more than ample. Now on to the chassis. The attached photo shows the input vs the output.

There was a concern that the polarity of the applied parabola after the MOSFET amplifier may be inverted. I went back to the schematic and reasoned that the polarity after the inverting MOSFET amplier should be correct. I have attached a pictorial reasoning for my assertion.

[Penthode]

This evening connected my breadboard circuit to the CT100. The first thing I found was the waveform on the plate of the 12AU7 inverted compared to the waveform printed in the service manual. This means I will likely have to invert it with another stage. Nevertheless I connected it up and was measuring voltages and looking at waveforms. I powered down and up again and found the MOSFET gate to source had shorted. I tried installing clamping diodes on the gate and replaced the MOSFET only to find it shorted. I found that the B+ shot up to just over 500 volts before the tubes warmed and as they were 500v fets that spelled their demise.

Anyhow, without the vertica dynamic convergence correction, with resistor feeding the convergence and focus electrodes only, the focus and convergence did not look bad. The worst was at the top which I feel touching up the vertical linearity will address. The two photos attached is the set running with no vertical convergence transformer. Only the 560k and 330k resistors substituting for the convergence transformer secondary.

I have buttoned the set back together while I think this thing through.

Post thought: Because the vertical convergence transformer has only a minimal effect on the convergence, it is conceivable that a replacement may be installed with a reversed phase primary which may not be noticable. Certainly I saw no impairment in focus and the convergence except for the very top is remarkably good.

[old_tv_nut]

Close!
A transformer with capactive coupling seems like a good thing to try. I would then be on the lookout for any phase distortion that would make a difference between the top and bottom convergence but it may not be a significant problem.