Deunan wrote on 2024-12-26, 11:09:

x86_guy wrote on 2024-12-24, 23:23:

I replaced the capacitor - it didn't help 😒

Pity, I was rather hoping it would do the job. Because there isn't much else in there than can cause that,

I agree. I'm surprised that changing the cap didn't help. Did it at least change the picture in some way? if not, the vincinity of that cap should be checked, especially the diodes 6515 and 6514. These diodes are meant to create a DC voltage on the capacitor 2515 (which has been changed), and that DC voltage will likely affect horizontal picture position.

We still have two possiblities why the picture is folded at the right margin: Either the HOT is turned off too early, ending the sweep. In that case, the root cause is likely located around the TDA2579 chip, or the HOT is still on, but the circuit is unable to drive the beam more to the right due to issues in the horizontal resonant circuit. In that case, the issue is likely right of the HOT (7512). We should be able to narrow down the issue by scoping TP11 (the flyback voltage feedback) and TP4, TP5 or TP6 at the same time on two channels. The waveforms at TP4/TP5/TP6 are the RGB components. The expected waveforms shown in the service manual are for a specific test pattern with color bars. Assuming you still use the "please insert the AmigaOS floppy" image as test screen, we expect a high level during most of the image time (except for the floppy image in the center), and a low level during blanking. Furthermore, we expect that the spike at TP11 occurs shortly after blanking begun.

If the spike at TP11 starts while the video signal is still high, the horizontal output transistor is driven off too early, and this causes the folding of the image at the right edge. If the spike at TP11 is some microseconds into the blanking, the horizontal output stage works correctly, but the horizontal deflection circuit is unable to provide enough deflection current through the yoke.

This troubleshooting hint does not supersede what Deunan suggested, but is an alternate approach to finding the root issue. I recommend to do both.

Invest in ESR meter.

This way, you will find lot of bad electrolytic capacitors at this age of this monitor. Not just specific. You will need to find all and replace that is out of specs.

Cheers,

nocash wrote on 2024-12-25, 07:14:

What I still don't understand is: Are you really using the white boot screen with the floppy symbol for testing? That looks completely unsuitable to me. Don't you have a workbench disc, or game disc, something with a real picture? Or are you an expert on Amiga hardware and know exactly how the white boot screen should look like on that particular monitor?

I am not too familar with the amiga appearance, I assume that the boot screen consists of a bitmap with white pixels, surrounded by a black or white (?) screen border, so I couldn't tell which of the pixels on you photo are part of the bitmap or border. And I assume the floppy and/or hand is supposed to be centered in the middle of the screen, with the same amount of white pixels on each side, but I am just guessing there. And of course, all pixels should have the same width, but it's impossible to tell if that's the case with all-white background on your photo.

Doing an image search for Commodore 1084S-P1, it seems the games do usually have a screen border about the size of the black region on your photo. I have also found a screenshot of the v1.3 bootscreen, which seems to have the whole screen including border filled with white pixels, unlike as on your photo... but that screenshot has a gray slider on the floppy symbol, why is yours having a red slider???

The floppy slider supposed to be gray instead of red, you're right. It's an old screenshot, I fixed it pushing "SK4" button to "analog" position instead of "digital". However, it didn't fix the image horizontal fading issue.

As a result of your above reply, I tested the monitor on another Amiga screen, and also connected it to a CGA machine, just to see how it displays a CGA signal.

The Amiga test on another screen shows clearly that the image is actually folding (at each side, depends on horizontal positioning) and the borders are not in place.
On the other hand, on the CGA test the display was good and the folding issue disappeared.
On both tests, the vertical size seems to be a bit shrunk, tuning it doesn't help. This is probably another issue.
I attached photos of both tests.

Deunan wrote on 2024-12-26, 11:09:

Pity, I was rather hoping it would do the job. Because there isn't much else in there than can cause that, except the whole size control circuit perhaps. This is where pincushion correction is injected, which also looks bad on your shots.

What do you mean by "pincushion correction"?

Deunan wrote on 2024-12-26, 11:09:

First, a question: In your opening post you've wrote that all the potentiometers on the back were replaced but you also gave the values: "2x10k, 1x220k". Which one is the 220k? Because AFAIK all three of them should be 10k.

3550 is the 220k POT. This is what mentioned in my schematic, do you have any other schematic that states differently?

Deunan wrote on 2024-12-26, 11:09:

Second, can you measure the voltage on the 1uF capacitor 2526, it's connected to that /!\ marked resistor 3526 on the schematic, right next to the flyback transformer. There shouldn't be any nasty voltages on this side of the circuit, the schematic shows 13V4 being the expected voltage. I am interested in what voltage do you get at both opposite ends turning the H-size control pot. I would expect this to be between about 1V at minimum and 25V or so at maximum.

2526 capacitor voltages:
19.76V when screen positioned all the way to the left,
20.2V when screen positioned all the way to the right.

mkarcher wrote on 2024-12-26, 18:40:

I agree. I'm surprised that changing the cap didn't help. Did it at least change the picture in some way? if not, the vincinity of that cap should be checked, especially the diodes 6515 and 6514. These diodes are meant to create a DC voltage on the capacitor 2515 (which has been changed), and that DC voltage will likely affect horizontal picture position.

I'm not sure, but I think it moved the left horizontal border of the screen to the right. This way - I can see the screen folding even on the left side, when moving the picture to the left.
I tested the diodes, here are the results:
6514 diode - Vf = 610 mV
6515 diode - Vf = 609 mV

For the rest of the components in this area, it will be a bit tricky as many of them are behind EMI shielding, I'll have to take them off to measure them. If it's the last resort, I'll do it also. Anyway, visually they look fine.

mkarcher wrote on 2024-12-26, 18:40:

We still have two possiblities why the picture is folded at the right margin: Either the HOT is turned off too early, ending the sweep. In that case, the root cause is likely located around the TDA2579 chip, or the HOT is still on, but the circuit is unable to drive the beam more to the right due to issues in the horizontal resonant circuit. In that case, the issue is likely right of the HOT (7512). We should be able to narrow down the issue by scoping TP11 (the flyback voltage feedback) and TP4, TP5 or TP6 at the same time on two channels. The waveforms at TP4/TP5/TP6 are the RGB components. The expected waveforms shown in the service manual are for a specific test pattern with color bars. Assuming you still use the "please insert the AmigaOS floppy" image as test screen, we expect a high level during most of the image time (except for the floppy image in the center), and a low level during blanking. Furthermore, we expect that the spike at TP11 occurs shortly after blanking begun.

If the spike at TP11 starts while the video signal is still high, the horizontal output transistor is driven off too early, and this causes the folding of the image at the right edge. If the spike at TP11 is some microseconds into the blanking, the horizontal output stage works correctly, but the horizontal deflection circuit is unable to provide enough deflection current through the yoke.

This troubleshooting hint does not supersede what Deunan suggested, but is an alternate approach to finding the root issue. I recommend to do both.

I'll try to arrange a setup for this measurement.

pentiumspeed wrote on 2024-12-26, 20:51:

Invest in ESR meter.

This way, you will find lot of bad electrolytic capacitors at this age of this monitor. Not just specific. You will need to find all and replace that is out of specs.

Cheers,

I have an ESR meter. I probably will do so, it's a good direction.

Now my main concerns are:
Does 3550 POT should be 10k and not 220k?
And what is the meaning that CGA PC signal is showing fine with no folding on this monitor, while RGB output of the Amiga isn't?
I want to mention that the folding issue is showing also with no signal at all, when brightness set to a point that a white screen shows up on the monitor.

Attached a couple of screenshots I made during the test.

More pics

x86_guy wrote on 2024-12-30, 20:15:

And what is the meaning that CGA PC signal is showing fine with no folding on this monitor, while RGB output of the Amiga isn't?

CGA is known to have a quite narrow picture. Your current problem is no longer a "picture too far to the right" issue anyway, as you can make the folding disappear on either side of the image. You now have a "picture too wide" problem instead, and that problem only affects the (black) overscan area of the CGA image. Compare your image to photos of actual CGA setups, and you will see that the image of Dangerous Dave is not supposed to stretch the full area between the left and right border. Is the "horizontal width" potentiometer 3534 already set to the minimum?

x86_guy wrote on 2024-12-30, 20:15:

What do you mean by "pincushion correction"?

"Pincushion correction" is the effect of pot 3537. You see that the right end of resistor 3533 is connected to the width pot 3534 to set a baseline picture width. The amount of current that goes into the circuit around transistor 7530 sets the picture width. As you see, the baseline picture width is combined with a signal coming from "above", through capacitor 2537. That signal dynamically adjusts the picture width to remove any pincushion distortion.

x86_guy wrote on 2024-12-30, 20:15:

What do you mean by "pincushion correction"?

The 3537 trimmer pot. Basically the left and right sides should be straight. From top to bottom. Any inward or outward curve will make the screen resemble a pillow/pincushion depending on which way it goes, rather than nice rectangle. Top and bottom are affected too but this is not really a sweep issue but more of the magnetic field, so it's often corrected with small magnets around the CRT neck.

x86_guy wrote on 2024-12-30, 20:15:

3550 is the 220k POT. This is what mentioned in my schematic, do you have any other schematic that states differently?

My bad. Got the pots on the rear mixed up, I was thinking of 3534 potentiometer having wrong value. But funny thing, the 3550 is listed as 220k on the part list, but as 220 ohm (not kilo) on the schematic. I think it should be 220k but if you still have the original pots that were replaced it would be nice to double check. I wouldn't worry about that too much, this one controls vertical and that seems to work just fine.

x86_guy wrote on 2024-12-30, 20:15:

2526 capacitor voltages:
19.76V when screen positioned all the way to the left,
20.2V when screen positioned all the way to the right.

Now that is odd. The way I understand this circuit the horizontal coil of the yoke is biased by DC, and the (nominal) B+ voltage that is charging it sits at 128V. Well the waveform there is anything but DC, it's a big nasty high voltage spike but in terms of energy stored it's not that big. So by having this bias voltage you regulate how much energy can be stored in the coil, thus how wide/narrow the deflection will be. IMO there should be way more control here.

So it might be the 7526 transistor going dead open for example. Can you measure the voltage on its base, or the collector of 7530 - same thing, might be easier to access. Same deal as before, check voltage in both opposite positions of 3534 pot. If you again see little difference repeat this test but at the base of 7530 transistor. But that one should be OK, it's protected by resistors and probably the only thing that is actually making the slight changes you see now.

EDIT: Perhaps I misunderstood, but which pot did you turn to measure these values? 3264 horizontal positon or 3534 horizontal size? You need to turn 3534 for this test, the size pot. The issue with the position might be the size is just too big and thus you can't fit it properly on the screen.

mkarcher wrote on 2024-12-30, 21:15:

CGA is known to have a quite narrow picture. Your current problem is no longer a "picture too far to the right" issue anyway, as you can make the folding disappear on either side of the image. You now have a "picture too wide" problem instead, and that problem only affects the (black) overscan area of the CGA image. Compare your image to photos of actual CGA setups, and you will see that the image of Dangerous Dave is not supposed to stretch the full area between the left and right border. Is the "horizontal width" potentiometer 3534 already set to the minimum?

Yes, it was already set to minimum.

mkarcher wrote on 2024-12-30, 21:15:

"Pincushion correction" is the effect of pot 3537. You see that the right end of resistor 3533 is connected to the width pot 3534 to set a baseline picture width. The amount of current that goes into the circuit around transistor 7530 sets the picture width. As you see, the baseline picture width is combined with a signal coming from "above", through capacitor 2537. That signal dynamically adjusts the picture width to remove any pincushion distortion.

Deunan wrote on 2024-12-30, 21:48:

The 3537 trimmer pot. Basically the left and right sides should be straight. From top to bottom. Any inward or outward curve will make the screen resemble a pillow/pincushion depending on which way it goes, rather than nice rectangle. Top and bottom are affected too but this is not really a sweep issue but more of the magnetic field, so it's often corrected with small magnets around the CRT neck.

Thanks for the explanation, it's now understood. I already tuned 3537 trimmer, as you can see in the recent pics, the borders are now seem to be straighter then before.

Deunan wrote on 2024-12-30, 21:48:

My bad. Got the pots on the rear mixed up, I was thinking of 3534 potentiometer having wrong value. But funny thing, the 3550 is listed as 220k on the part list, but as 220 ohm (not kilo) on the schematic. I think it should be 220k but if you still have the original pots that were replaced it would be nice to double check. I wouldn't worry about that too much, this one controls vertical and that seems to work just fine.

Actually the vertical position tuning acts weird also. Sorry for not mentioning this before. I wasn't focusing it as it wasn't an issue for me, but now when you talk about it, it makes me think it might be something to consider.
All the test I've done and posted screenshots of so far, were with 3550 POT turned all the way to the left. The vertical size with this setting is reasonable, not too high and not too short.
When I start turning it clockwise, the pictures starts to fold at the bottom and the picture starts shrinking. Then, at about 50% of it's way, the picture stops shrinking and starts to expand vertically, in a weird way.
I'm attaching screenshots of the process so you can better understand what is going on here.

Deunan wrote on 2024-12-30, 21:48:

Now that is odd. The way I understand this circuit the horizontal coil of the yoke is biased by DC, and the (nominal) B+ voltage that is charging it sits at 128V. Well the waveform there is anything but DC, it's a big nasty high voltage spike but in terms of energy stored it's not that big. So by having this bias voltage you regulate how much energy can be stored in the coil, thus how wide/narrow the deflection will be. IMO there should be way more control here.

So it might be the 7526 transistor going dead open for example. Can you measure the voltage on its base, or the collector of 7530 - same thing, might be easier to access. Same deal as before, check voltage in both opposite positions of 3534 pot. If you again see little difference repeat this test but at the base of 7530 transistor. But that one should be OK, it's protected by resistors and probably the only thing that is actually making the slight changes you see now.

EDIT: Perhaps I misunderstood, but which pot did you turn to measure these values? 3264 horizontal position or 3534 horizontal size? You need to turn 3534 for this test, the size pot. The issue with the position might be the size is just too big and thus you can't fit it properly on the screen.

My fault, I adjusted the horizontal position (3264 POT) instead of the horizontal size (3534 POT).
Repeating the test and adding 7526 and 7530 transistors brought the following results:

2526 capacitor:
Widest horizontal display = 1.2V
Narrowest horizontal display = 20V

7526 transistor:
Narrowest horizontal display - Emmiter = Base = 19.6V
Widest horizontal display - Emmiter = 1.1V, Base = 0.4V

7530 transistor:
Narrowest horizontal display - Collector = 19.8V, Base = -1.8V (Yes, negative).
Widest horizontal display - Collector = 0.4V, Base = 0.5V

All the measurements above taken with negative terminal of 2445 capacitor as reference.

x86_guy wrote on 2025-01-01, 01:33:

2526 capacitor: Widest horizontal display = 1.2V Narrowest horizontal display = 20V […]
Show full quote

2526 capacitor:
Widest horizontal display = 1.2V
Narrowest horizontal display = 20V

7526 transistor:
Narrowest horizontal display - Emmiter = Base = 19.6V
Widest horizontal display - Emmiter = 1.1V, Base = 0.4V

7530 transistor:
Narrowest horizontal display - Collector = 19.8V, Base = -1.8V (Yes, negative).
Widest horizontal display - Collector = 0.4V, Base = 0.5V

All the measurements above taken with negative terminal of 2445 capacitor as reference.

These measurements look fine. The horizontal width control seems to be working as intended. The schematic indicates that 13.4V at C2526 is "usual" for a properly tuned monitor. As you can easily go above and below that value, it seems the circuit that controls that voltage is fine, but the horizontal deflection circuit generally operates at an excessive amplitude. This might indicate that the new flyback is not a 100% match for the old flyback, or that something is still damaged in the horizontal deflection circuit.

The primary idea of the horizontal deflection circuit is that it is a kind-of resonant LC circuit, which gets heavily kicked every horizontal retrace. The main components are the inductance of the yoke and inductor 5522, as well as the capacitors 2517, 2518 and most prominently 2519. Did you already check the capacitance of these caps? Loss of capacitance in these caps also might explain why the old flyback failed: The idea of the flyback transformer is to push around a considerable amount of energy. If that energy is not properly caught in capacitors of sufficient capacity, voltages may rise to an undesired level. Excessively high voltage from the flyback might also explain the high sensitivity of the X-Ray protection circuit (which is supposed to sabotage the high-voltage generation if there is an apparent danger of excessive X-Ray generation due to excessive voltage applied to the tube). I'm with Deunan on the theory that the reaction to excessive B+ is likely the X-Ray protection (trying to) shut down the horizontal oscillator / HV generator.

x86_guy wrote on 2025-01-01, 01:33:

All the test I've done and posted screenshots of so far, were with 3550 POT turned all the way to the left. The vertical size with this setting is reasonable, not too high and not too short.
When I start turning it clockwise, the pictures starts to fold at the bottom and the picture starts shrinking. Then, at about 50% of it's way, the picture stops shrinking and starts to expand vertically, in a weird way.

Hm. That is obviously not correct behaviout but I checked the datasheet for both TDA chips and it seems the nominal resistance for that current source is 330k. And there is already 270k resistor connected to pin 4 of TDA2579 so the pot must have the range to go up to 330k in total (and a bit beyond). It would probably be better to have 100k pot here but I'm not going to argue with the designer, they tested it that way and it was done for some reason. The point is the pot should be in kiloohms so the parts list has the correct value.

So this might not be related but then again the TDA2579 is the one controlling the vertical shape and frequency, the other chip TDA3653 is just an amplifier/driver for the yoke. So I was thinking if the IC can't control the vertical properly and there is a semi-related issue with horizontal sweep... maybe it's the chip that's the problem after all? Something wrong with its power supply or noise? That will need to be investigated.

x86_guy wrote on 2025-01-01, 01:33:

Repeating the test and adding 7526 and 7530 transistors brought the following results: […]
Show full quote

Repeating the test and adding 7526 and 7530 transistors brought the following results:

2526 capacitor:
Widest horizontal display = 1.2V
Narrowest horizontal display = 20V

Yep, that's more like it. See, the more DC bias the horizontal yoke coil has the less voltage difference it sees during charging. This results in less current going through it, and less energy stored. Less energy = less deflection and more narrow image. Which is exactly what you get with 20V. Now I did expect it to go a bit higher but this is kind of resonant circuit that feeds on the flyback pulses, the DC bias is generated by the coil itself. It could be due to the differences in the flyback transformer and having a bit different resonance frequency.

That being said you should check the 2526 and especially 2524 capacitors for loss of value and high ESR. Stricly speaking the ESR should not be that critical here due to the 5523 coil but it might just be making the difference.

The cap that you already replaced and was 2515, it's very close to the HOT heatsink so that's why it degraded faster (that and the pulses it has to work with). But there are a few other caps in the area that should be checked too, if you haven't done so already. So here's the list, better to check them now and be sure, otherwise we'll be just guessing:

2520: It's that 10uF that filters 128V to the flyback. It's located between flyback and the PSU transformer where is't quite warm, and it too works with nasty spikes. It needs to be in good shape.
2524 and 2526: In the horizontal DC bias circuit. Explained above.

2560 and 2561: These are two 220uF caps connected in series near TDA7560 on schematic. I do not understand why they did that, a single 100uF would work just fine. It just doesn't sit well with me and these are important to correct operation of the vertical amp so check those. Note the parts list is wrong here, lists them as 22uF instead of 220. Both sit in the corner of the PCB near HOT.
2571: It's the bigger cap nearby 2560 and 2561. Also very important for vertical shape and general behaviour. Now typically any issues with it would cause vertical to not reach full deflection but with the ICs correcting the shape all kind of weird things can happen. So it needs to be tested for value and ESR as well. This one must have good ESR.
2563: Also nearby, it filters power supply for the vertical amp. Important as well.

I do not expect to find any issues here, except maybe 2524 / 2526, but since we now have two sweep issues the investigation should be more systematic. It's best to first check and eliminate any easy to test (and failure prone) parts before digging into more convoluted scenarios I think.

Here are the measurements I did following your last posts:
Component number (SCH value) = capacitance, ESR
2560 (220uF 20%) = 213.8uF, 500mOhm
2561 (220uF 20%) = 208.4uF, 560mOhm
2563 (470uF 20%) = 419uF, 250mOhm
2571 (680uF 20%) = 647.2uF, 130mOhm
2518 (22nF 10%) = 22.05nF, 37.18Ohm
2519 (560nF 10%) = 560.6nF, 2.5Ohm
2517 (8.2nF 5%) = 8.237uF, 1.5Ohm
2524 (??) = 5.068uF, 800mOhm
2526 (1uF 20%) = 1uF, 8.629Ohm
2520 (10uF) = 10.58uF , 1.645Ohm

Parts 2560, 2561 defined as 22uF on part list, but on SCH it's 220uF.

Any way to check if it's a flyback not matching 100% to the old one issue?

Deunan wrote on 2025-01-01, 12:29:

Hm. That is obviously not correct behaviout but I checked the datasheet for both TDA chips and it seems the nominal resistance for that current source is 330k. And there is already 270k resistor connected to pin 4 of TDA2579 so the pot must have the range to go up to 330k in total (and a bit beyond). It would probably be better to have 100k pot here but I'm not going to argue with the designer, they tested it that way and it was done for some reason. The point is the pot should be in kiloohms so the parts list has the correct value.

So this might not be related but then again the TDA2579 is the one controlling the vertical shape and frequency, the other chip TDA3653 is just an amplifier/driver for the yoke. So I was thinking if the IC can't control the vertical properly and there is a semi-related issue with horizontal sweep... maybe it's the chip that's the problem after all? Something wrong with its power supply or noise? That will need to be investigated.

This is a test that I put on my to-do list. Any other things suggested to test for this chip, other then power supply? I may try to replace 3550 POT with one of another value?
Are any more tests I can do to narrow down the issues?

x86_guy wrote on 2025-01-05, 00:15:

Here are the measurements I did following your last posts:

All of them look good. So good in fact that I would not bother replacing any of it.

x86_guy wrote on 2025-01-05, 00:15:

2524 (??) = 5.068uF, 800mOhm

Huh, so there is a reason why it's so bloody big for 4.7uF. Apparently it does need to be low ESR and tolerate high current pulses.

x86_guy wrote on 2025-01-05, 00:15:

Any way to check if it's a flyback not matching 100% to the old one issue?

Not really if you don't have another working monitor with the original flyback transformer to compare with. The exact specs of these parts are not provided.

x86_guy wrote on 2025-01-05, 00:15:

This is a test that I put on my to-do list. Any other things suggested to test for this chip, other then power supply? I may try to replace 3550 POT with one of another value?
Are any more tests I can do to narrow down the issues?

So before you start swapping other parts I'd like you to change one particular resistor first. I went back to your earlier post where you measured the 3264 horizontal position pot and noticed you also checked the 3271/3274 voltage divider, and you got 3271 = 1.15M. Well the parts list wants this 1M resistor to be 5% and that 1.15 is no longer in spec. I not so sure that is very important but it's a feedback so...

You could also scope test points 10 and 26 (both are vertical sweep) - both when the picture has correct height, and when it's collapsed on the bottom when trying to adjust height. This is to see what is going on there. Maybe also add test point 22 to that. BTW the screenshots from your scope are nice but the menu on the right seems to obscure some readings, like the V/div for the vertical scale. Is it possible to get rid of it for the screenshots?

mkarcher wrote on 2025-01-01, 10:59:

These measurements look fine. The horizontal width control seems to be working as intended. The schematic indicates that 13.4V at C2526 is "usual" for a properly tuned monitor. As you can easily go above and below that value, it seems the circuit that controls that voltage is fine, but the horizontal deflection circuit generally operates at an excessive amplitude. This might indicate that the new flyback is not a 100% match for the old flyback, or that something is still damaged in the horizontal deflection circuit.

I would like to know how did you get that the horizontal deflection circuit operates at excessive amplitude? Is it according to one or more of the test results I've posted or by understanding the monitor behavior symptoms I described?

mkarcher wrote on 2025-01-01, 10:59:

The primary idea of the horizontal deflection circuit is that it is a kind-of resonant LC circuit, which gets heavily kicked every horizontal retrace. The main components are the inductance of the yoke and inductor 5522, as well as the capacitors 2517, 2518 and most prominently 2519. Did you already check the capacitance of these caps? Loss of capacitance in these caps also might explain why the old flyback failed: The idea of the flyback transformer is to push around a considerable amount of energy. If that energy is not properly caught in capacitors of sufficient capacity, voltages may rise to an undesired level. Excessively high voltage from the flyback might also explain the high sensitivity of the X-Ray protection circuit (which is supposed to sabotage the high-voltage generation if there is an apparent danger of excessive X-Ray generation due to excessive voltage applied to the tube). I'm with Deunan on the theory that the reaction to excessive B+ is likely the X-Ray protection (trying to) shut down the horizontal oscillator / HV generator.

I get it. I tested those components, you can see the results in my previous post.
Prior the flyback replacement the B+ voltage wasn't adjusted, it was after I replaced the 128V voltage trimmer, as well as all other trimmers on this board.
I don't really know what was the voltage before, as I adjusted it only after turning this trimmer to see how it impacts the image. Then, after the screen when black and the high pitch sound came back (probably the X-Ray protection as you and Deunan mentioned, same as I had with the previous flyback), I realized I need to tune it for exactly 128V per service manual.

Deunan wrote on 2025-01-05, 16:51:

Not really if you don't have another working monitor with the original flyback transformer to compare with. The exact specs of these parts are not provided.

I'm thinking about the possibility to return the old flyback. What are the chances it's still in working order?

Deunan wrote on 2025-01-05, 16:51:

So before you start swapping other parts I'd like you to change one particular resistor first. I went back to your earlier post where you measured the 3264 horizontal position pot and noticed you also checked the 3271/3274 voltage divider, and you got 3271 = 1.15M. Well the parts list wants this 1M resistor to be 5% and that 1.15 is no longer in spec. I not so sure that is very important but it's a feedback so...

I took the 3271 off now for another test. This time I measured it with my ESR tweezers, those should be even more accurate than my DVM. I got a 1.18M, so it's definitely out of spec. I'll search for a compatible resistor with high accuracy and will let you know what impacts the display.

Deunan wrote on 2025-01-05, 16:51:

You could also scope test points 10 and 26 (both are vertical sweep) - both when the picture has correct height, and when it's collapsed on the bottom when trying to adjust height. This is to see what is going on there. Maybe also add test point 22 to that. BTW the screenshots from your scope are nice but the menu on the right seems to obscure some readings, like the V/div for the vertical scale. Is it possible to get rid of it for the screenshots?

I'll scope those TPs and post the results.
For the scope display, you're right. I noticed it yesterday while reviewing the tests I made. I'll fix it in the future measurements.

Another thing I want to mention. Prior to opening this thread, I did an initial debugging session when the monitor didn't display anything. During this this session I found that resistor 3512 served a complete open loop. So I replaced it, and the monitor came alive.
The thing is, I didn't have the exact same 4.7Ohm resistor as the original one, so I soldered two 10Ohm resistors in parallel. Measuring it giving 4.85Ohm. I think it's OK as it's within the 5% tolerance service manual wants, but on the other hand, it's not the exact resistor and exact P/N as service manual wants (it's marked in /!\ , which means "replace with identical parts for safety reasons"). Any impact it can have related to the issues I described?
I'm attaching pics of the old 3512 resistor and the new one, and also the old 3550 pot (if it helps for something).

x86_guy wrote on 2025-01-05, 23:07:

I would like to know how did you get that the horizontal deflection circuit operates at excessive amplitude? Is it according to one or more of the test results I've posted or by understanding the monitor behavior symptoms I described?

The "amplitude" of the horizontal deflection is the picture width. If the picture is too wide, the amplitude is too high. Well, most likely. Another cause for excessive deflection can be insufficient anode voltage (I'm talking about the ~25kv now). If the electrons are not accelerated hard enough, they spend more time near the deflection coils and thus get deflected more. This effect applies to both horizontal and vertical deflection. As you don't seem to have the issue that the picture is generally to high as well, I'm currently not expecting low anode voltage to be the root cause.

x86_guy wrote on 2025-01-05, 23:07:

I'm thinking about the possibility to return the old flyback. What are the chances it's still in working order?

It's not in working order, unless you miscounted pins when making test measurements. I'm quoting an older post by you and the reply by Deunan:

Deunan wrote on 2024-12-10, 10:36:

x86_guy wrote on 2024-12-10, 00:09:

I tested the old flyback now. Pin 7 isn't shorted to any other pin. However, I get a resistance of 28.5Ohm between Pins 9,10 and 3,5. Per flyback datasheet, those pins need to be totally isolated, so I guess this flyback sure is bad?
The new flyback reads open loop between those pins.

That would be my assumption, a short inside. In some cases it might be just an autotransformer-like setup to generate additional voltages directly from primary (it is very common for so-called boost voltage) but not here. Pin 9 is connected to GND so there is no way we can have this winding be a part of the primary, which is directly connected to +128V. It's dead I'm afraid.

x86_guy wrote on 2025-01-05, 23:07:

During this this session I found that resistor 3512 served a complete open loop. So I replaced it, and the monitor came alive.
The thing is, I didn't have the exact same 4.7Ohm resistor as the original one, so I soldered two 10Ohm resistors in parallel. Measuring it giving 4.85Ohm. I think it's OK as it's within the 5% tolerance service manual wants, but on the other hand, it's not the exact resistor and exact P/N as service manual wants (it's marked in /!\ , which means "replace with identical parts for safety reasons"). Any impact it can have related to the issues I described?
I'm attaching pics of the old 3512 resistor and the new one, and also the old 3550 pot (if it helps for something).

Replacing a 4.7Ohm resistor with two 10Ohm resistors in parallel is perfectly fine by itself. The value of this resistor is not critical. But "identical part for safety reasons" is critical, and you probably defeated a safety mechanism in the monitor. The 4.7 Ohm resistor that was originally installed is built in a way that it will not incinerate if it gets too hot because a lot of current is passing through it ("flameproof resistor"), and most likely it is even more sophisticated: The resistor is built in a way that it also is a non-resettable thermal fuse ("fusible resistor"). So if there is too much current through the resistor, it will get too hot, and then the built-fuse will blow, resulting in a resistor that reads open circuit. So it is likely that 3512 opened for a reason. The reason is quite likely that the old flyback was broken. I don't see any proof that you still need to hunt for reason for 3512 to be damaged.

The compoent list says 4.7 ohms, 0.33W. 0.33W at 4.7 ohms is an average current of around 250mA at a voltage of 0.15 volts. I guess your replacement is able to dissipate a lot more power before blowing open, so even if the monitor doesn't catch fire, the chance of transistor 7510 getting damaged if something goes wrong is likely much higher now than it was before the original resistor failed.

x86_guy wrote on 2025-01-05, 23:07:

I'm thinking about the possibility to return the old flyback. What are the chances it's still in working order?

Slim to none considering that you found a low resistance path between pins 3 and 9. Unless I misunderstood the previous posts. This resistance would be connected between +128V from PSU and GND. Best case scenario the PSU will detect a short and trip but this one seems a bit dumb so it will try to restart rather then give up. So what's going to happen is one of two things is going to blow open: the winding in the flyback or the 3520 resistor. And/or the PSU fuse.

x86_guy wrote on 2025-01-05, 23:07:

I took the 3271 off now for another test. This time I measured it with my ESR tweezers, those should be even more accurate than my DVM. I got a 1.18M, so it's definitely out of spec. I'll search for a compatible resistor with high accuracy and will let you know what impacts the display.

The parts list wants a 0.5W resistor but that might be a special high voltage rated carbon composition type. Anything carbon will drift due to age and temperature, even soldering those too long will shift their resistance. You might want to get a 2W resistor if you can fit it in there (which is usually possible). If you need to stand if off the board use a piece of stripped wire insulation on the leads. At 2W I think a modern 1% metal film resistor will also survive here. See the 2W is not about the power rating but bigger size, which is what provides better max voltage rating.

x86_guy wrote on 2025-01-05, 23:07:

Another thing I want to mention. Prior to opening this thread, I did an initial debugging session when the monitor didn't display anything. During this this session I found that resistor 3512 served a complete open loop. So I replaced it, and the monitor came alive.

If you look at the schematic, the 3512 resistor provides the power to collector of 7510, to switch the primary side of the transformer that drives HOT. 7510 transistor base is biased high by the 3510 pull-up resistor, thus it's turned on by default - without signal from the TDA2579. This loss of signal would quickly burn out the primary of the transformer and/or the driver transistor itself. So this resistor is basically a fuse. Has a small value to not drop the voltage too much and a capacitor (2512) on the other side to provide even lower impedance for AC.

TL;DR: You want a 0.33W resistor here and the two you put in are probably 0.6W each, so now it's about 1.2W or so, way too high. The original one probably blew when the flyback shorted and this event confused the hell out of the control IC. It didn't turn the transistor off as it should. At least that's my theory on that.

BTW the reason I want to focus on the vertical issue now is it should be easier to figure out and perhaps will provide some clues. Now I also noticed, looking at the PCB photo, that the corner where the 7270 and 7560 ICs live has seen water ingress and there is some sort of residue left behind. Some of it might be a bit conductive and/or attract moisture, you want to clean that as best as you can with some brush. Maybe even remove the big capacitors for easier cleaning and also to check if there isn't any buildup between those and PCB. Some of the control currents these IC use are low, in tens of uAs, so even minor current leaks might be causing issues.

mkarcher wrote on 2025-01-06, 00:10:

The "amplitude" of the horizontal deflection is the picture width. If the picture is too wide, the amplitude is too high. Well, most likely. Another cause for excessive deflection can be insufficient anode voltage (I'm talking about the ~25kv now). If the electrons are not accelerated hard enough, they spend more time near the deflection coils and thus get deflected more. This effect applies to both horizontal and vertical deflection. As you don't seem to have the issue that the picture is generally to high as well, I'm currently not expecting low anode voltage to be the root cause.

I get what you say. But on the other hand, I'm thinking about how is it that the picture is to wide, if I'm actually able to tune it wider and the folding doesn't increase? The "folding" on each side of the screen just moves with the picture while making it wider and moving picture to the left / right to see the edges.
This leads me to think, it might be the picture signal that is to wide and/or not synced with the deflection system, or am I missing something?
As I mentioned earlier, CGA signal shows up just fine, maybe this is a direction?

mkarcher wrote on 2025-01-06, 00:10:

Replacing a 4.7Ohm resistor with two 10Ohm resistors in parallel is perfectly fine by itself. The value of this resistor is not critical. But "identical part for safety reasons" is critical, and you probably defeated a safety mechanism in the monitor. The 4.7 Ohm resistor that was originally installed is built in a way that it will not incinerate if it gets too hot because a lot of current is passing through it ("flameproof resistor"), and most likely it is even more sophisticated: The resistor is built in a way that it also is a non-resettable thermal fuse ("fusible resistor"). So if there is too much current through the resistor, it will get too hot, and then the built-fuse will blow, resulting in a resistor that reads open circuit. So it is likely that 3512 opened for a reason. The reason is quite likely that the old flyback was broken. I don't see any proof that you still need to hunt for reason for 3512 to be damaged.

The component list says 4.7 ohms, 0.33W. 0.33W at 4.7 ohms is an average current of around 250mA at a voltage of 0.15 volts. I guess your replacement is able to dissipate a lot more power before blowing open, so even if the monitor doesn't catch fire, the chance of transistor 7510 getting damaged if something goes wrong is likely much higher now than it was before the original resistor failed.

Deunan wrote on 2025-01-06, 00:55:

If you look at the schematic, the 3512 resistor provides the power to collector of 7510, to switch the primary side of the transformer that drives HOT. 7510 transistor base is biased high by the 3510 pull-up resistor, thus it's turned on by default - without signal from the TDA2579. This loss of signal would quickly burn out the primary of the transformer and/or the driver transistor itself. So this resistor is basically a fuse. Has a small value to not drop the voltage too much and a capacitor (2512) on the other side to provide even lower impedance for AC.

TL;DR: You want a 0.33W resistor here and the two you put in are probably 0.6W each, so now it's about 1.2W or so, way too high. The original one probably blew when the flyback shorted and this event confused the hell out of the control IC. It didn't turn the transistor off as it should. At least that's my theory on that.

Each resistor is 0.25W making this parallel resistor network 0.5W overall.

So I will be fine to change the resistor to to a standard 4.7Ohms 0.33W? Or I need to find a fusable one?

Deunan wrote on 2025-01-06, 00:55:

The parts list wants a 0.5W resistor but that might be a special high voltage rated carbon composition type. Anything carbon will drift due to age and temperature, even soldering those too long will shift their resistance. You might want to get a 2W resistor if you can fit it in there (which is usually possible). If you need to stand if off the board use a piece of stripped wire insulation on the leads. At 2W I think a modern 1% metal film resistor will also survive here. See the 2W is not about the power rating but bigger size, which is what provides better max voltage rating.

I don't have a 1M ohm 0.5W (certainly not 2W) resistor, but I have a 0.25W one. So I prepared two sets of 2 x 1M 0.25W resistors in parallel, to get a 1M 0.5W resistor values total, just for the test.
This setup measures 0.98M on DVM.
Installing it didn't seem to change the picture at all.

Deunan wrote on 2025-01-06, 00:55:

BTW the reason I want to focus on the vertical issue now is it should be easier to figure out and perhaps will provide some clues. Now I also noticed, looking at the PCB photo, that the corner where the 7270 and 7560 ICs live has seen water ingress and there is some sort of residue left behind. Some of it might be a bit conductive and/or attract moisture, you want to clean that as best as you can with some brush. Maybe even remove the big capacitors for easier cleaning and also to check if there isn't any buildup between those and PCB. Some of the control currents these IC use are low, in tens of uAs, so even minor current leaks might be causing issues.

I cleaned it. Still I get no improvement .

Deunan wrote on 2025-01-05, 16:51:

You could also scope test points 10 and 26 (both are vertical sweep) - both when the picture has correct height, and when it's collapsed on the bottom when trying to adjust height. This is to see what is going on there. Maybe also add test point 22 to that. BTW the screenshots from your scope are nice but the menu on the right seems to obscure some readings, like the V/div for the vertical scale. Is it possible to get rid of it for the screenshots?

I couldn't find the time to do it but I'll try to do it soon and post results.

BTW, are there any places I can fit a variable resistor so I can try and tune the vertical / horizontal circuits in a trial and error method, so I can narrow down the issue to a specific circuit I have to investigate further and then get to the root cause?

x86_guy wrote on 2025-01-12, 23:13:

BTW, are there any places I can fit a variable resistor so I can try and tune the vertical / horizontal circuits in a trial and error method, so I can narrow down the issue to a specific circuit I have to investigate further and then get to the root cause?

I'll address the other things when I have a bit more time but if you want to experiment then replace the 3550 pot with 220 ohm one (not 220k). Long story short it didn't sit well with me but the datasheet uses 220k pot and it confused me. What they use their pot for is not what the monitor does. And on top you get correct vertical deflection in both 0% and 100% positions, where the resistance of the pot is next to nothing to one of the side terminals. But it goes wrong in the middle of the range. So there is probably nothing wrong with this after all but the pot having wrong value.

If you don't have 220 ohms on hand just use two 100 ohm resistors in series, and connect those in place of the pot. In fact you can even leave the pot in place with the resistors, just tack them on the solder side for this. The middle point of the series to the wiper contact, the side leads to the pot side contacts. Set the wiper in the middle. Obviously the pot will no longer work (except the 0/100 positions) but if that gives you a stable picture with the image height somewhere in the middle and no fold then you'll have your answer right there - the pot is wrong value. And if so skip the scope tests of the vertical section, no longer necessary.

mkarcher wrote on 2024-12-30, 21:15:

Your current problem is no longer a "picture too far to the right" issue anyway, as you can make the folding disappear on either side of the image. You now have a "picture too wide" problem instead, and that problem only affects the (black) overscan area of the CGA image.

No, look at the two workbench screenshots, they are all too far right, even the one with the left-most pot setting.
The range of the horizontal pot seems to be a bit small, only +/-10 pixels or so.
But that should be very easy to fix, use a bigger potentiometer, and/or replace the resistors that are in series to the potentiometer (and also check if those resistors are still intact at all).
Then you should have at least have a properly centered picture.

Edit: Sorry, the left-most workbench picture seems to be already properly centered (it looks odd because the right border is half blue and half black).

mkarcher wrote on 2024-12-30, 21:15:

Compare your image to photos of actual CGA setups, and you will see that the image of Dangerous Dave is not supposed to stretch the full area between the left and right border. Is the "horizontal width" potentiometer 3534 already set to the minimum?

Well maybe, but one could argue if there any clear rules whether CGA is or isn't supposed to be displayed with large borders on left and right side.
Most people are probably happy without large borders, at least if the aspect ratio is about right.
But if you want large CGA borders, then yes, answer the question about whether the "horizontal width" potentiometer is set to minimum.

Anyways, this is a commodore monitor, supposedly intended to be used with hardware like the amiga, so I would use the amiga as primary reference for how it should look like.
If it's flexible enough to display CGA pictures with proper centering and whatever desired border widths and heights, that would be a nice bonus.

Picture Height
That's clearly too small. Are you sure that you aren't trying to display a picture from a NTSC amiga on a PAL monitor?

Picture Width / Folding
The problem there seems to be that the monitor refuses to display pixels close to the screen borders.
If the picture gets zoomed or moved too close the borders, then it does instead draw those bright vertical lines.
Could that be some protection circuit intended to prevent excessive deflection, getting triggered a bit too early?

The amiga picture width looks about right, it's almost certainly not too wide.
If anything it could be a bit wider, then it would be perfect (apart the folding issue which would then probably appear on both left and right screen sides).

PS.
Some CRTs have a dot matrix that is a bit smaller than the visible glass surface.
Going by your photos, I couldn't tell if that's the case here (but looking at photos from other commodore monitors, it shouldn't be the case).
Anyways, please have a look at your tube. If the the edges of the tube are looking "non-transparent", then it might be simply impossible to display pixels in that region.

PPS.
Thanks for clarifying the reason of the red color. I had initially thought that it might be some hardware hack, or a hardware glitch happening when writing to the palette registers (which could have explained similar glitches about screen dimensions).

So I replaced the 220k POT in question with 10k POT and vaula! The vertical size control is now working properly!
So one thing solved.

For the horizontal size control - I may replace the POT also with another value, any recommendations? I don't want to mess things up there.

All of you, thanks alot for the help!

Well 10k is already 22 times better but if you can find a 220 ohm one it would be better yet 😀

I recommend not changing any part values in the horizontal deflection, there's just too high risk of causing HOT/flyback damage. Not to mention I do not think changing the 3534 pot would actually affect anything. I'd like the -27V rail to be a bit lower, closer to that but that too is flyback-generated so... You could check the parts in the rectifier/filter (everything connected to flyback pin 10), especially the 2541 capacitor. ESR is not critical but it should be close to 22uF. So far the other caps tested OK, except that one in the yoke current path, so I doubt you'll find anything wrong here. Replacing 2541 with a 33uF or even 47uF might lower the voltage a bit but not by much.

So first I'd like to comment on the resistors from previous posts. Yes, if you can replace 3512 with 4.7 ohm fusible one, even at 0.5W instead of 0.33W, it would be better then your current replacement. As for the 1M resistor 3271, while it does create correct resistance when you connect 2x2M in parallel, it does nothing for the voltage rating. In fact it now offers two possible failure paths. It's far better, if you can't get a higher wattage / bigger 1M resistor, to create a replacement out of series of 2 or even 3 resistors. Then their voltage ratings add and make the whole circuit safer. The problem is you wan to have these 2/3 resistors more-or-less the same value for even voltage split. For two it would be a 510k + 470k, both selected to be a higher value, or 3x 330k, also selected. With 330k it might be easier as even 1% is already 3k so finding 3 resistors that are a bit above 330k would be easier. Still the easiest thing would be to get a higher wattage 1M in there.

I'm also wondering if the horizontal issue is not some sort of 50Hz vs 60Hz signal difference. Depends on the timings but the 50Hz will appear wider if it has the same amount of lines as 60Hz. One way to work around a different flyback would be to try adding a very small resistance (ohms at most) in series with the yoke horizontal coil. It wouldn't even be that difficult seeing how the compensation chokes are all over the PCB, connected with jumper wires. But before you start messing with that it would be good to rule out all other possible issues.

This is how workbench looks on a 1084S-P (and I assume 1084S-P1 should look similar), with, say, 8mm borders on left/right and top/bottom,
https://www.amibay.com/threads/commodore-1084 … y-london.70989/

The folding issue will probably cut-off the leftmost 1mm and rightmost 12mm, but apart from that...
Can you adjust the width and height to match that? And adjuste the horizontal/vertical position to get it centered properly?