Bad composite video signal

[Gominalo]

Hello!
I am trying to fix the image of a sony dvd player (model dvp-s536d)
I checked voltages and they are ok, then i tried to tune a little pot to achieve the 1Vpp of the signal but the nearest I can be without loosing it is 1.5Vpp. Even so seems to work better with higher voltages. I am suspecting on a bad video chip. What more can i try?

Thank you.

[james_s]

Do you have access to an oscilloscope to look at the video waveform? First thing I'd suspect is bad electrolytic capacitors, but it could be any number of things. Can you find a datasheet for the video IC you suspect?

[inse]

Did you measure the video level with a multimeter or an oscilloscope?
Anyway, the composite video is the least quality signal, are there other video interfaces you can use?
RGB, S-Video?
It looks like there is some HF modulation on the video signal, check it with an oscilloscope as already proposed.
Maybe power supply ripple...

[vk6zgo]

Did you measure the video level with a multimeter or an oscilloscope?
Anyway, the composite video is the least quality signal, are there other video interfaces you can use?
RGB, S-Video?
It looks like there is some HF modulation on the video signal, check it with an oscilloscope as already proposed.
Maybe power supply ripple...

Yes, I wouldn't trust a DMM reading to be accurate looking at video.
The default device should always be an oscilloscope for such checks.

[Gominalo]

Yesterday i measured the power supply with the oscilloscope, but today the oscilloscope doesn't seems to work well.
Here i have the video signal, it is super messy.
I think i have some 50hz noise comming into the oscilloscope, but i don't know how to clean it.

[Gominalo]

Okay, i had a bad ground.
In the supplies, the worst of all had a ripple less than 100mV.
Don't you think that composite signal is too noisy?
The second picture is the luminance component of the s-video
The third is chrominance.

[vk6zgo]

Your Luminance display shows your problem.

There should be no signal superimposed on the luma step pattern.
If you look at the composite signal, you will see exactly the same signal on the parts of the colour bar which don't normally have chroma, & a mixture of chroma & the interference in the parts that are coloured.

"Counting squares" on your display, shows approximately 6 cycles of the interfering signal in 10us, making its frequency approx 600kHz. (You can probably find that frequency a bit more closely with your cursors).

That frequency seems a bit high to be from a SMPS.
It might, perhaps, be a harmonic, or possibly some circuit oscillating, perhaps due to a faulty bypass.

[wasedadoc]

That interference is on the chroma too.  Third photo in the 3-photo post above.

[Gominalo]

Your Luminance display shows your problem.

There should be no signal superimposed on the luma step pattern.
If you look at the composite signal, you will see exactly the same signal on the parts of the colour bar which don't normally have chroma, & a mixture of chroma & the interference in the parts that are coloured.

"Counting squares" on your display, shows approximately 6 cycles of the interfering signal in 10us, making its frequency approx 600kHz. (You can probably find that frequency a bit more closely with your cursors).

That frequency seems a bit high to be from a SMPS.
It might, perhaps, be a harmonic, or possibly some circuit oscillating, perhaps due to a faulty bypass.

What is the best way to find the possible bad cap, ESR meter?, Curve tracer?
Those are supossed to be tantalum caps, right?
Thank you.

[james_s]

Finding them can be tricky, it's probably easiest to find them indirectly by looking at the power rails with the scope. Do you have another working DVD player? If so you should look at the signal from that using the same settings on the scope to give you a baseline and confirm that the scope is set up and connected correctly. I think the other posters are correct about the noise though.

[vk6zgo]

That interference is on the chroma too.  Third photo in the 3-photo post above.

It also appears on the sync tips & back porch, which should be totally clear of anything but pure DC levels.
It seems to be common to every component of the signal.

The 'scope pix seems to show that it is synchronous with video, too, but maybe that is just an artifact of the DSO, or the camera.
Some analog TV equipment does try to phase lock the SMPS to the video waveform to minimise the visibility of any interference patterns.
I tend to wonder if there is some kind of spurious oscillation on the original crystal clock for the whole process.

[inse]

Some analog TV equipment does try to phase lock the SMPS to the video waveform to minimise the visibility of any interference patterns.
I tend to wonder if there is some kind of spurious oscillation on the original crystal clock for the whole process.

Really? I haven't seen this kind of trick in consumer products except those TV's with combined mains and flyback transformer.

[Gominalo]

That interference is on the chroma too.  Third photo in the 3-photo post above.

It also appears on the sync tips & back porch, which should be totally clear of anything but pure DC levels.
It seems to be common to every component of the signal.

The 'scope pix seems to show that it is synchronous with video, too, but maybe that is just an artifact of the DSO, or the camera.
Some analog TV equipment does try to phase lock the SMPS to the video waveform to minimise the visibility of any interference patterns.
I tend to wonder if there is some kind of spurious oscillation on the original crystal clock for the whole process.

Okay this is what I got from de clock generator (Hint: all signals are bad)

The pictures are in order, also in the scope the signals waveforms are labeled from 1 to 6, in the service manual screenshots you can see the example waveforms and the only matching point are the frequencies (more or less)
The IC that generates all the signals i can get it from china, it is the cxd9572n, but maybe is the resonant circuit of the xtal the origin of the problem.
In my workplace I can measure the smd caps in circuit with a smart tweezers that we have.
Tell me your thoughts.
Thank you in advance.

[Gominalo]

Okay this is what I got from de clock generator (Hint: all signals are bad)

The pictures are in order, also in the scope the signals waveforms are labeled from 1 to 6, in the service manual screenshots you can see the example waveforms and the only matching point are the frequencies (more or less)
The IC that generates all the signals i can get it from china, it is the cxd9572n, but maybe is the resonant circuit of the xtal the origin of the problem.
In my workplace I can measure the smd caps in circuit with a smart tweezers that we have.
Tell me your thoughts.
Thank you in advance.

This is how i managed to probe those signals, because they are under the pcb

[vk6zgo]

Some analog TV equipment does try to phase lock the SMPS to the video waveform to minimise the visibility of any interference patterns.
I tend to wonder if there is some kind of spurious oscillation on the original crystal clock for the whole process.

Really? I haven't seen this kind of trick in consumer products except those TV's with combined mains and flyback transformer.

Luckily, those particular horrors weren't part of the domestic TV scene in Oz.

Some small Philips TVs (actually, Sharp TVs in drag) did use line rate SMPS, but it was a separate unit, just locked to synch pulses.

I seem to remember some of the Sony TV Broadcast level monitors synching the SMPS as well, although the big KX27PS1 "Profeel" units had a totally free running SMPS, relying on very good shielding ---If you ran them without the shielding, you could see an interference pattern on the screen.

[james_s]

Which signal is which? Many of those don't look bad to me, at least not without context. You're probing tens-of-MHz signals with those loooong wires stretched over loads of other parts and traces, that's going to add a lot of noise. How are you grounding your probe? That's critical at higher frequencies.

[inse]

It's a liitle hard to judge the measurements without having my own hands on the set.
Yet I wonder how the clock generator would interfere with the video signal.
The distortion on the CVBS signal was somewhere between 500 and 600 kHz, can you focus that more detailed?
What does the supply voltage of the video output chip look like?

[james_s]

I wonder if there's a linear regulator that's oscillating. I've seen that before when there was a worn out capacitor on the output. Perhaps there's an oscillating amplifier in the video signal.

[vk6zgo]

"Just for shits & giggles" look at (3) & (5) again,on pins (12) & (16) of IC105 respectively, with the 'scope set to a much longer time/div.
The poor triggering tends to give me the idea that, over time, the trigger point is shifting, as would happen if there was a lower frequency signal superimposed on the desired one.

It is interesting that both pins on IC105 which show this problem are connected to MB-86 BOARD 3-11

Another point worth checking  is the +3.3V "B+" supply.
This can be checked on pin (4) of CN102 as shown on your schematic.

I haven't played with a DVD player for years, & am trying to remember if the colour bar pattern is a built-in source, or if the one you are showing is on a DVD.
If the former, does the problem still exist if you are actually playing a DVD?

[Gominalo]

"Just for shits & giggles" look at (3) & (5) again,on pins (12) & (16) of IC105 respectively, with the 'scope set to a much longer time/div.
The poor triggering tends to give me the idea that, over time, the trigger point is shifting, as would happen if there was a lower frequency signal superimposed on the desired one.

It is interesting that both pins on IC105 which show this problem are connected to MB-86 BOARD 3-11

Another point worth checking  is the +3.3V "B+" supply.
This can be checked on pin (4) of CN102 as shown on your schematic.

I haven't played with a DVD player for years, & am trying to remember if the colour bar pattern is a built-in source, or if the one you are showing is on a DVD.
If the former, does the problem still exist if you are actually playing a DVD?

Hello again friend, sorry for the waiting.

Signals (3) & (5) are stable and i can't see any other superimposed signal.

Other interesting thing is when i touch the DVD ground with the oscilloscope ground alligator clip i can hear some noise in the switching power supply due to that touch, but when the clip is closed and good connection is done there is no noise. So that makes me think about ground noise but i think i need a diferential probes to measure that, right?

I also probed the cristal oscilator of IC102, it seems a bit deformed but still in specs.

Thank you.

[timenutgoblin]

I don't know if it's of any interest to the O.P., but would it be worth considering building a parallel resonant LC filter using ceramic capacitors tuned somewhere between 500kHz to 600kHz (as already mentioned) and then connect the filter in series between the CV output of the DVD player and the CV input of the TV/monitor?

The idea is that the filter would pose a high series impedance to the mystery superimposing signal while also passing signals necessary for luminance, chrominance and sync pulses. DC will pass through the inductor and the colour subcarrier will pass through the (ceramic) capacitor.

Checking for an oscillating linear regulator (as already mentioned) is a good idea.

[EPAIII]

In over 45 years I have not seen a composit video problem like that. What strikes me are two facts. First, it appears to be locked to the video's horizontal rate and second it is low at lower levels of the video signal (sync tip) and high at the higher levels (white level). I could be wrong but do not think that a power supply issue could do both of those. It should be simple to check this out by looking at the power rails with the scope. Be sure to trigger the display at the horizontal rate while doing this (Channel 2 probe).

This is coming out of that video chip you mentioned. Is it on any other outputs other than composit video? Component video? L-C? Or whatever? What about digital output(s)? Do they show it?

I think the next thing I would do us to probe all the pins of that video chip, keeping the scope triggered at the horizontal rate. See if you can see anything around 600 KHz on any of them.

BTW, does that video chip do the D-A conversion or is that elsewhere?

[EPAIII]

I have noticed something else strange in the three waveforms in Reply #5. The first one, composit video and the second one, luminance (no chroma) show the 600 kHz signal increasing with higher video levels and almost completely absent at sync tip.

But the third waveform, the chroma one, which also seems to be displayed at the horizontal rate and locked to horizontal sync (note below), shows the 600 kHz signal all the way across the display. There is no place where it goes nearly to zero at the tip of sync.

It is not there (at sync tip) in the composit signal, which is the combination of luminance and chroma but it is there in the chroma signal. That does not make sense. If it is present in the chroma and not in the luminance, then the sum of those two should show it present. You can't cancel the signal present in the one with the lack of signal in the other.

Is there something going on in the Rigol scope or in the probing? Could this be an artifact of the digital sampling in the Rigol scope. The diagonal pattern on the picture monitor photo seems to argue against the stripes seen there being locked to the horizontal frequency. If they were, they would be a series of vertical bars, not at an angle. Could a scope sampling problem be hiding the actual problem from us. You must understand your tools. Isn't there a way to change the sampling frequency in the Rigol?

Note: I feel fairly confident that the chroma trace is locked to horizontal sync because I can make out the two areas which represent the white and black bars. They have the 600 kHz signal but no chroma (3.579 Mhz) mixed with it. And those two areas, one black and the other white, seem to have the same level of the 600 kHz signal.

[wasedadoc]

1. All the analogue video outputs go through the video buffer chip IC101. Page 4-36 of the Service manual. There is also a DC-DC converter (not the main power supply) at the bottom of the schematic on that page.  Is that running at the same frequency as the interference?

(I once had a problem with a Hi-8 video deck giving distorted audio playback.  DC to DC converter for the vacuum fluorescent display was working very hard because the electrolytic on its output rectifier had dried up and gone really low capacitance.  The extra strong switching frequency from that converter was getting into the audio subcarrier retrieval and demod circuits.)

2.  Ignore the suggestion of a notch filter.  Even if it reduces the interference it will have a very bad effect on picture quality.

[vk6zgo]

I have noticed something else strange in the three waveforms in Reply #5. The first one, composit video and the second one, luminance (no chroma) show the 600 kHz signal increasing with higher video levels and almost completely absent at sync tip.

But the third waveform, the chroma one, which also seems to be displayed at the horizontal rate and locked to horizontal sync (note below), shows the 600 kHz signal all the way across the display. There is no place where it goes nearly to zero at the tip of sync.

It is not there (at sync tip) in the composit signal, which is the combination of luminance and chroma but it is there in the chroma signal. That does not make sense. If it is present in the chroma and not in the luminance, then the sum of those two should show it present. You can't cancel the signal present in the one with the lack of signal in the other.

Accurate DC levels are very important to the luma component of an analog TV signal.
If such a signal passes through an ac coupled amplifier or network, it loses the reference DC levels like black level, blanking, & sync tips.
These need to be restored by a process called "DC restoration".

Historically, this was done in various ways, but over time, the other methods have fallen into disuse, leaving the almost universal use of "keyed clampers" which produce a keying pulse synched to the line syncs, & either delay it a bit over 5us to be coincident with the "back porch", (the most common method) or somewhat less so it still falls within the sync pulse duration.

The keying pulses will cause the keyed clampers to hard switch either of those two video levels to a selected DC reference value during the duration of the clamp pulse, & would normally remove any signals riding on that DC level, however, keyed clampers used on the "back porch" of  composite colour signals  incorporate "traps" parallel resonant at the colour subcarrier (soft clamping), so as to not "mangle" the colour burst.
Sync tip clamping, which, for a standard 1V p-p video signal, clamps the sync tip to a DC level of around -300mv w.r.t zero volts (exactly that for PAL) does not require such traps,

Perhaps Sony decided on the use of sync tip clamping to reduce the complexity of having to provide LC "subcarrier traps", & further decided to clamp the luma component both before & after it was combined with the chroma signal.

The Chroma signal doesn't have a DC level to be clamped, so the interfering signal isn't "clamped out".