CPD1604S Vertical Collapse

[vol.2]

It is interesting with 78xx regulators, in that over time they tend to have a small voltage shift in the downward direction, never up it seems.

That circuit probably measures just over 12V (12.3) at the regulator output pin with respect to GND because of the small voltage drop across the 0.45uH L909, L912 inductors and after L905 is drops a little again to 12V. Interesting that the current return via those two inductors is proportional to that in the +24 and +78V and +160V supplies. The DCR of the two 0.45uH chokes must be very low.

It is interesting. Thanks for the insight.

I replaced the 7812, and I indeed got back my 12V, so that's a win.

I ran a couple of test patterns to see if I could spot anything unusual, and I noticed that it's not doing very well on the High Voltage regulation test. This is a test that flashes a white box in the middle of the screen, and a thin white border towards the edge. When I show this test pattern, the white box significantly grows and shrinks, but only in the Horizontal direction.

I wonder if this might mean there's an issue with the horizontal output transistors?

 

[Hugo Holden]

Unlikely, the H output transistor acts as a saturated switch. If not saturating, apart from heating up, the R sided scan linearity would be degraded.

Since its only the H size that is changing, the EHT is probably stable . Possibly the voltage supplying the H output transformer primary, via Q401 is not stable during that test. Its a chopper power supply to about 60V controlled by IC401.

On the other hand there are more possibilities, such as a defect remaining in the E-W modulator where there was a lot of electrolytic corrosion, or it could be the EHT regulation is fluctuating during the test, but something is happening in the vertical section at the same time, partially cancelling the increase in vertical size as the EHT drops.

 

[vol.2]

Since its only the H size that is changing, the EHT is probably stable . Possibly the voltage supplying the H output transformer primary, via Q401 is not stable during that test. Its a chopper power supply to about 60V controlled by IC401.

Okay. I think I found the issue. There's some problem with the 160V source. I see 155V on the 160V line. There must something blown up in there?

This was measured on P903. I see the 24V dead-nuts on Pin 1, Pin 8 is 155V.

I tested and replaced any of the bad electrolytics months ago, so it's unlikely to be bad caps.

Spoiler: Voltage Sources

 

[vol.2]

Alright. I took a scope reading of the collector of Q403, which is where the chopped supply that is supposed to be 60.9VRMS. There is a waveform on the Service Manual that shows it's supposed to be 660Vpk, and it's more like 1.1kVpk!

This seems very not good. I also took a snapshot of Q402's collector with the same probe, just to make sure everything was setup correctly. Unless there is some huge difference between the 140V reading and the 660V reading insofar as my probe or scope is concerned, it's way too high. Or I guess the service manual could be wrong, but that seems like a lot of wrong.

As you can see here, the first reading is spot on, and the second one is too high. Same probe, same settings, same scope.

Spoiler: What they should be.

Spoiler: What Wave 1 is:

Spoiler: What Wave 2 is:

 

[Hugo Holden]

The drive waveform looks ok. The peak voltage on the HOT's collector may well be ok . It is typically around 1 kV in color sets. (make sure to use a 1500 to 200V rated x100 scope probe though, a standard x10 probe only puts up with about 600V, over that it risks damaging the probe or the scope's input FET's and attenuator area)

The peak voltage here is set largely by the tuning capacitors in the collector circuit. If one goes low in uF value value or open then the peak voltage will increase. What the peak of the wave corresponds to, it occurs halfway through flyback, is all of the energy of the magnetic field that was present, when the beam was on the extreme R side of the CRT (this being I^2L/2) is converted to electric field of those tuning capacitors and self capacitance (CV^2/2) which peaks when the magnetic fields are depleted (beam at CRT center half way through flyback). So for the same energy value, a constant k as in CV^2/2 = k then the Voltage peak = root (2k/C), meaning that the peak voltage is inversely proportional to the square root of the total value of the tuning capacitances + self capacitances of the windings of the yoke & transformer. So if the capacitance gets smaller, the peak voltage gets bigger. Its worth checking the capacitors. In some sets if these capcitors go open, it causes the EHT to increase (if the EHT is derived from the H scan output stage). In multi-scan VDU's they keep the EHT generator separate- but I can't see the rest of the schematic. And it can cause failure of the HOT, by exceeding its collector voltage.

If the peak changes so does the with of the base of the wave, in that as the peak gets higher, the width gets narrower. This is because it is a 1/2 cycle of oscillation being tuned by the capacitances. And the width sets the flyback time, as it corresponds to the time taken to get the beam from the R side of the screen to the L. But in sets capable of multi-standards it is not uncommon to modify the total value of the tuning caps circuit to control the flyback time, so I'm not sure if the recording of over 1kV you got is higher than the 660V in the manual because the set is on another video standard so check that too.

 

[vol.2]

So if the capacitance gets smaller, the peak voltage gets bigger. Its worth checking the capacitors. In some sets if these capcitors go open, it causes the EHT to increase (if the EHT is derived from the H scan output stage). In multi-scan VDU's they keep the EHT generator separate- but I can't see the rest of the schematic. And it can cause failure of the HOT, by exceeding its collector voltage.

Okay. I can check the capacitors. I'm assuming it's all those high voltage caps 408, 409, 412, 410, etc.

This is the whole D board:

Spoiler: Whole D PCB

If the peak changes so does the with of the base of the wave, in that as the peak gets higher, the width gets narrower. This is because it is a 1/2 cycle of oscillation being tuned by the capacitances. And the width sets the flyback time, as it corresponds to the time taken to get the beam from the R side of the screen to the L. But in sets capable of multi-standards it is not uncommon to modify the total value of the tuning caps circuit to control the flyback time, so I'm not sure if the recording of over 1kV you got is higher than the 660V in the manual because the set is on another video standard so check that too.

The manual states that all called-out voltages are taken with a color bar signal, H:31.47 V:70.1, which is 640x400@70Hz, and that's exactly what I have taken every and all voltages at. It also states that all voltages are with respect to ground.

I also checked the voltage here when the monitor was displaying 1024x768 to see if the horizontal breathing showed a voltage fluctuation here (it did not), and the 1kVpk signal was exactly the same at that frequency, so if it's supposed to change then something's wrong in there. Well, it's almost exactly the same. Obviously the duty cycle is different because it's a 57kHz signal, but the amplitude of the waveform is still about the same, maybe a tiny touch higher at about ~1.14kV instead of ~1.08-1.1kV.

Edit: I tested all the tuning capacitors and they are all perfect. Nary a nanofarad off.

 

[Hugo Holden]

Okay. I can check the capacitors. I'm assuming it's all those high voltage caps 408, 409, 412, 410, etc.

This is the whole D board:

Spoiler: Whole D PCB

The manual states that all called-out voltages are taken with a color bar signal, H:31.47 V:70.1, which is 640x400@70Hz, and that's exactly what I have taken every and all voltages at. It also states that all voltages are with respect to ground.

I also checked the voltage here when the monitor was displaying 1024x768 to see if the horizontal breathing showed a voltage fluctuation here (it did not), and the 1kVpk signal was exactly the same at that frequency, so if it's supposed to change then something's wrong in there. Well, it's almost exactly the same. Obviously the duty cycle is different because it's a 57kHz signal, but the amplitude of the waveform is still about the same, maybe a tiny touch higher at about ~1.14kV instead of ~1.08-1.1kV.

Edit: I tested all the tuning capacitors and they are all perfect. Nary a nanofarad off.

In that case the 660V figure in the manual is probably incorrect.

 

[vol.2]

In that case the 660V figure in the manual is probably incorrect.

Thanks for mentioning it. I would have been worried about it otherwise.

Actually, after checking all those caps and going over a couple spots on the D board, the instability seems to have decreased a lot. It's no longer breathing when I turn it on. I'm still getting some horizontal breathing with the flash test, but it seems to be slightly less.

Of course, the better I tune it in, the more critical I seem to be of the picture lol. Now I'm worried about some smearing I'm getting with too much contrast. I know that I am used to the brightness of modern monitors, but I can't help but wonder if there is something I could do to push it a little bit further.

Basically, what happens is when I turn up the contrast, at a certain point the bright whites start to smear red towards the right. I have seen this behavior in other VGA monitors, but it typically happens at a higher brightness with the others I own.

Is there some component responsible for this that I can tweak the value of in order to achieve higher picture setting without smearing, or is that pretty much set in stone?

 

[andy]

You generally don't probe the C of the HOT because of the high voltage pulses. It requires great care to avoid blowing up your test equipment. Often , service manuals say "do not measure" on the schematic. The fact that you have a full width picture with no weird linearity problems means you can forget about the HOT collector circuit. I would be looking at how well regulated the +160v line is. If it's sagging under load, this would explain your problem.

As far as smearing goes, that's caused when the video output transistors are driven into saturation (ie the signal is clipping). It's often a sign of a weak CRT if you have to drive the cathodes that hard to get sufficient brightness and contrast. A good CRT that's properly setup should never show smearing. Getting a little blurry at high brightness is normal, but not smearing to the right. Back when CRTs were all we had, you learned to balance a bright picture with a sharp picture, often reducing the contrast a little to make it sharper.

You should go through the CRT gray scale setup since improper adjustment can cause smearing at high contrast. For example, if the G2 is set too low, and you're compensating for it by over driving the cathodes, you will see smearing even thought the CRT is good.

 

[vol.2]

You should go through the CRT gray scale setup since improper adjustment can cause smearing at high contrast. For example, if the G2 is set too low, and you're compensating for it by over driving the cathodes, you will see smearing even thought the CRT is good.

Understood. That might just be the case. I did go through the white balancing act, but I didn't start with the backgrounds all the way down like the service manual said. I think the backgrounds were probably in the middle and not the bottom of travel as SM wanted me to. I didn't realize that could cause a smearing issue. I thought the G2 and the backgrounds were basically just doing the same thing, but the G2 was all of them at once and the backgrounds do it per color. I guess I need to learn and understand those relationships (between G2, the Background adjustments, and the Contrast and Drive adjustments) better than I do.

Thanks for the heads-up. I'll redo the white balance and see how that effects the overall condition.

It's often a sign of a weak CRT if you have to drive the cathodes that hard to get sufficient brightness and contrast.

To be fair, I can get it to at least 103, probably more like 105 Nits before I get the red ghosting on the brightest whites, so it's not dim by any means, but I'm used to seeing VGA monitors do a little better than that. My little 640x480 Mitsubishi can about 135 Nits before it get bad, and it's fairly high hour.

 

[Hugo Holden]

You generally don't probe the C of the HOT because of the high voltage pulses. It requires great care to avoid blowing up your test equipment. Often , service manuals say "do not measure" on the schematic.

I agree, which is why I often suggest using x100 probes working on TV's VDU's and any Tube based equipment where the voltages are often over 400v. The x100 probes now are very cheap on ebay and are usually rated in the range of 1500V to 2kV and they suit looking at the HOT's collector voltage in most color sets, without risking the scope's input circuitry. Pretty cheap insurance for the scope. It is a very psychologically painful experience blowing up the front end jFets in a Tek scope and probably leads to PTSD, I did it once, over 35 years ago now, a mistake I have never made since but still have bad dreams about:

100X P4100 High Voltage 2KV 2000V Oscilloscope Scope Passive Clip Probe 100MHz | eBay

We are Industrial Products Supplie. We have focused on various industrial parts for more than 15 years.

www.ebay.com

 

[vol.2]

I agree, which is why I often suggest using x100 probes working on TV's VDU's and any Tube based equipment where the voltages are often over 400v.

Thanks. Yes, I have a 100x probe that I used for this purpose with my oscilloscope. I also have a good isolation transformer for my DUTs.

 

[vol.2]

I was looking at the schematic yesterday, and I can't help wondering about the smearing/ghosting at higher contrast levels in higher resolutions.

The main video amplifier IC of this monitor is the Sanyo VPS07T, which maxes out at 70MHz, and is generally only really happy up to 56kHz Horizontal. Sony didn't really spec this thing with much headroom for higher resolutions. Looking at the datasheet for this IC, there is a Pin compatible drop-in IC (VPS10) that does 100MHz, but seems to have a slightly lower DC gain at about 20 dB vs the 22 dB of the original IC.

Spoiler: VPS07T and VPS10 Datasheet

I am thinking about just giving this a shot to see if I can get a little more headroom out of it before it ghosts. My assumption is that this IC group are current driven amplifiers, and that I would need to change the 22 ohm resistor on the color inputs for the RGB lines to something a bit lower like a 20 ohm resistor to increase the current at the inputs for the VPS10. I am not sure if this might effect some impedance matching in the circuit that is controlled by that input resistor though.

Can anyone confirm my approach here?

Spoiler: Video Amplifier Circuit

This is really more of an academic thing for me, but it would be nice if it succeeded of course.

 

[Hugo Holden]

You might be hard pressed to improve those.

They are a cascode stage driven (TR1 & TR2) class B amplifier (Tr4 & Tr3) etc, with a small initial bias to overcome cross over distortion.

You have some access to the high frequency compensation though, by way of the emitter of TR1,TR5 & TR9. As you can see they have bypassed that with the series 33R and 100pF caps to boost the HF response. You could try lowering the 33R resistors (even shorting them out) or increasing the value of the 100pF or both.

The input resistance to TR1,5, 9 at their bases is much higher than the 22R series resistors, though to peak the HF response you could try bypassing those with a capacitor in the range of 47pF to 470pF. But there is always a risk of istability at a very high frequency.

In any case if you go too far with HF boost in the video, you can get an effect where the outlines on edges etc get exaggerated, but you may be able to compensate the smearing somewhat before that happens.

 

[vol.2]

You might be hard pressed to improve those.

They are a cascode stage driven (TR1 & TR2) class B amplifier (Tr4 & Tr3) etc, with a small initial bias to overcome cross over distortion.

Thanks for your insight.

It sounds like after reading your explanation, I could easily substitute VPS10 for the existing VPS07T because the gain is set by the surrounding components and it is not fixed (and therefore the datasheet is telling me the maximum gain (22, 20) and not a fixed gain value.

You have some access to the high frequency compensation though, by way of the emitter of TR1,TR5 & TR9. As you can see they have bypassed that with the series 33R and 100pF caps to boost the HF response. You could try lowering the 33R resistors (even shorting them out) or increasing the value of the 100pF or both.

The input resistance to TR1,5, 9 at their bases is much higher than the 22R series resistors, though to peak the HF response you could try bypassing those with a capacitor in the range of 47pF to 470pF. But there is always a risk of istability at a very high frequency.

So then both of these things "lowering the 33R resistors (even shorting them out) or increasing the value of the 100pF or both" and bypassing the 22R input resistors would potentially have the effect of peaking the high frequency response?


Another issue I was hoping to address was the red smearing before the other two colors. One idea that came up was to slightly lower the frequency response of the Red channel on the output Bias transistors to lower the energy put into the Red phosphors for the duration of the scan and give them a chance to recover. The thought was that this might lower blooming at a similar voltage level at the expense of a little color resolution. Might be worth it if it works?

Looking at my schematic, I think perhaps the best spot to do that would be to increase C538 (0.0033uF) cap that might be handling HF roll off? Maybe try and swap it for a 0.0047uF?

I don't know, this might be a crazy idea.

 

[Hugo Holden]

Thanks for your insight.

It sounds like after reading your explanation, I could easily substitute VPS10 for the existing VPS07T because the gain is set by the surrounding components and it is not fixed (and therefore the datasheet is telling me the maximum gain (22, 20) and not a fixed gain value.

So then both of these things "lowering the 33R resistors (even shorting them out) or increasing the value of the 100pF or both" and bypassing the 22R input resistors would potentially have the effect of peaking the high frequency response?


Another issue I was hoping to address was the red smearing before the other two colors. One idea that came up was to slightly lower the frequency response of the Red channel on the output Bias transistors to lower the energy put into the Red phosphors for the duration of the scan and give them a chance to recover. The thought was that this might lower blooming at a similar voltage level at the expense of a little color resolution. Might be worth it if it works?

Looking at my schematic, I think perhaps the best spot to do that would be to increase C538 (0.0033uF) cap that might be handling HF roll off? Maybe try and swap it for a 0.0047uF?

I don't know, this might be a crazy idea.

Umm...not sure what you are getting at here. I think the bias controls do not relate to frequency response, they are DC level controls.

 

[vol.2]

Umm...not sure what you are getting at here. I think the bias controls do not relate to frequency response, they are DC level controls.

I wasn't at all sure about this idea. You're light years ahead of me with all this stuff, so I'm sorry if it seemed silly on my part.

The issue I was trying to address is the typical Red smear you get with older tubes. My understanding is that the red phosphors often exhibit ghosting and smearing before the others because the red phosphors take more energy to excite, and as all the phosphors age, the red ones will end up needing a bigger voltage swing to light up the same amount. This bigger voltage swing on the red leads to ghosting, smearing, blooming effects in the red before the other two colors.

The general idea was to see if I could somehow tweak the frequency response roll off of just the red channel by adding a capacitor somewhere. Perhaps messing around with the roll off would be a way to reduce red blooming at higher luminance levels, while introducing a controlled amount of red color resolution loss across all luminance levels. But I'm not sure where in this circuit I would try it, or if it would even work.


Honestly, I'm more excited to try swapping the VPS07T for a VPS10 and to try tweaking the values on the "TE" pins.

 

[Hugo Holden]

Lowering the video's HF response won't help any form of smearing effect, only exaggerate it.

On the other hand overdoing the HF response can appear to improve things because when the beam is bright and goes to a lower level on the edge of an image along a horizontal line, the voltage overshoot provided by the extra HF boost causes it to go transiently to a lower level. A similar effect going from a low level to a higher one, the overshoot has the effect of "sharpening up the edges" on the screen image. Going too far can make it look cartoon like. On some sets they had a control for the amount of HF boost in the video, called an Aperture control.

 

[andy]

As I said earlier, the smearing isn't a frequency response problem. It's caused when you try to over drive the cathode which causes the video output transistor to go into saturation. It then takes some time for the transistor to recover from saturation, which causes the smearing. The only way to fix this is to not drive the video output stage into saturation. You can turn up the G2, and reduce the bias as much as possible, but too much G2 will cause retrace lines to appear.

A strong CRT that's properly adjusted should never show smearing. If the CRT is getting weak, then all you can do is make the best of it with careful adjustment, and lowering the contrast. You're not going to be able to redesign the circuit to make a weak CRT look like new.

 

[vol.2]

Thanks for the reply.

A strong CRT that's properly adjusted should never show smearing. If the CRT is getting weak, then all you can do is make the best of it with careful adjustment, and lowering the contrast. You're not going to be able to redesign the circuit to make a weak CRT look like new.

I'm generally aware that the smearing is caused by pushing the cathodes too hard. This is all just academic tweaking to see what is possible. I certainly don't expect any miracles, but I'd like to learn more about the electronics and concepts by optimizing what I have to work with. I'm especially interesting in out-of-the-box thinking about the issues and finding novel solutions. I have plenty of working monitors to use, this one is a project.

As I said earlier, the smearing isn't a frequency response problem. It's caused when you try to over drive the cathode which causes the video output transistor to go into saturation. It then takes some time for the transistor to recover from saturation, which causes the smearing. The only way to fix this is to not drive the video output stage into saturation.

I have seen this demonstrated in slightly older monitors with push pull transistors on the drive outputs and that is easier for me to follow. However, in my monitor, the video output stage has two ICs and then some intermediary transistors. I don't entirely understand when you say it "causes the video output transistor to go into saturation," which parts you are referring to in this case. There are "drive" transistors Q511, 512 and 513 and "bias" transistors Q508, 509, and 510, but there is also an RGB preamp IC, and the RGB amplifier IC.

If I had to take a guess, it would be the RGB amplifier IC which is going "into saturation," but I don't really know. Maybe it's the drive transistors or some combination of them.

Spoiler: Neck Board Schematic

You can turn up the G2, and reduce the bias as much as possible, but too much G2 will cause retrace lines to appear.

So, I tried this, and it really made everything much, much worse. The picture is at it's clearest, sharpest and most bright when I turn the G2 down as far as I can and the Bias controls up as far as they will go. I seem to get a lot more headroom that way. I went back and forth a couple times with it, and it definitely seems to shake out this way.

Not to get my hopes up, but you seem to be pretty confidant that it should be the other way around, so perhaps that is indicative of some other electrical failure and not the tube?