Jump to content
IGNORED

Slight smear in all forms of video output


nick3092

Recommended Posts

Since I've started using my Jaguar with a scaler and on my large TV, I've noticed some slight smearing/ghosting on the video image.  It's very obvious with RGB (more predominant on the white, but also noticeable on the red):

 

JagRGBSCARTscaler.thumb.jpg.2b31c788ebe263f00ddc501f7eab59f6.jpg

 

Hooked up to a Commodore CRT monitor via RGB, its harder to see (the CRT mask tends to make it less obvious), but you can still tell it's there:

JagRGBSCARTCRT.thumb.jpg.adf2c22624cdfc9cd91800a895526db7.jpg

 

I can also see it using s-video through the scaler as well:

 

JagSvideoScaler.thumb.jpg.3e8a8b02339d20d389a02b90d74738f1.jpg

 

Looking at the schematic and talking with a friend of mine who knows a wide array of electrical topics, we came to the conclusion that the since both s-video and RGB are affected, the issue has to be coming before the issue is coming from before the PAL/NTSC encoder (U3), since both RGB and s-video are affected (and RGB by passes the encoder).  Based on the schematic, he had me check a couple test points with my scope.

 

Most notably TP17 (which is right before the R signal goes through a 75ohm resistor and out the A/V port on the back and also right before it branches off to the encoder):

TP17200ns.thumb.png.7045ab45c6598992fac99ae973069cc7.png

 

And TP89-92 (which is the digital R signals going into the RDAC (U7).  Here is 89 (91 and 92 looked very similar and 93 was a flat line, which he said was most likely normal based on the grid test pattern I had on the screen):

 

TP89.thumb.png.f202d6e0b8e4f3b5e7de647a486d4301.png

 

From what I understand from him (some of what he was telling me was going a bit over my head), is that TP 89-92 looked clean.  But TP17 should really look similar to what I was seeing for the 89-92, with straight lines on both sides of the pulse.  It should just have a lower amplitude/voltage than the digital pulses.  And that the curved line on the right side of the pulse is the smear.  But he didn't really have any idea on what was causing it.  Outside of the smear, the picture looks quite good.  And colors all seem bright and correct.  I had recapped the Jaguar back in 2019, using a kit from Console5.  Unfortunately I wasn't using a scaler back then, so I have no idea if it was smeared before that (since its only very apparent when using RGB on a large display).

 

I'm hoping maybe someone here that is more familiar with the inner workings of the Jag video path might have a better idea as to whats happening and how to fix it.

 

Thanks!

 

-Nick

 

 

Link to comment
Share on other sites

A couple more data points, if it helps.

 

Given that white seems to have a larger smear than red alone, I theorized that blue and green are also smearing (since white is full R,G,B). I checked the equivalent points of TP17 and TP88 for blue and green, and those pulses look identical to the red pulses I posted above. The smear from each color seems to be summing up and creating a larger smear with white than an individual color. 

 

As a very long shot, I pulled c14 to see if maybe something in the encoder circuit was somehow back feeding into the RGB circuit. No change in the red pulses at TP17/88. And as expected, I lost all red color when hooked up via svideo. 

Link to comment
Share on other sites

Sounds like the issue is somewhere between U13 and C14/C17/C18 if I am reading this right? Maybe the transistors Q2/Q3/Q4 between the resistor array and video encoder could be looked at but it feels like this might be an inherent characteristic of the Tom ASIC package. Or something's wrong with your Tom. 😕

 

Your best bet at narrowing this down would be to get a second Jaguar to measure its test point readings against yours. I kinda want to volunteer scoping my personal Jag but I wouldn't be able to schedule this anytime soon.

Link to comment
Share on other sites

7 hours ago, john_q_atari said:

I notice smearing on yellow text in Gravitic Mines. My video path is jag -> hd retrovision genesis rgb cables -> ossc-> 4k video monitor. I wonder if all Jags have the issue you described.

I don't know for sure, but I don't think so.  At least not all Jags.  FirebrandX is known for dialing in scalers like the OSSC and Retrotink 4k for various classic consoles.  He did a video a couple years ago of the Jag and the OSCC.  If you jump to about about 11:30, he has the scaler dialed in and super burnout paused.  it looks incredibly crisp and clean.

 

 

2 hours ago, Bratwurst said:

Sounds like the issue is somewhere between U13 and C14/C17/C18 if I am reading this right? Maybe the transistors Q2/Q3/Q4 between the resistor array and video encoder could be looked at but it feels like this might be an inherent characteristic of the Tom ASIC package. Or something's wrong with your Tom. 😕

 

Your best bet at narrowing this down would be to get a second Jaguar to measure its test point readings against yours. I kinda want to volunteer scoping my personal Jag but I wouldn't be able to schedule this anytime soon.

Whatever the issue is, it affects all 3 colors the same way as the curve on the falling edge of the pulse is present on all 3 colors.  Which is what I don't understand.  There is no common component/path between them, other than they all come form the Tom, and they all pass through the DAC (which is just an R-2R resistor ladder).

 

I don't see how it could be the Tom, as the digital RGB signals going into the resistor ladder/DAC are completely vertical.  If the Tom was at fault, I would expect to see some distortion on the digital pins of the DAC.  Which would point to the DAC.  But even that seems unlikely, since from what I understand the colors all have their own path in the DAC.  No resistors in the DAC/ladder should be common to all the colors.  I could be wrong though, as I admit I'm not very well versed in any of this.  I'm ;earning as i go here.

 

I also pulled C14 to eliminate any interference coming back from the Y-C encoder circuit, no change on the red pulse.  And the odds of all 3 transistors on the RGB lines all failing in the exact same fashion at the exact same time is almost improbable.

 

Looking at Ebay pricing, a second Jag is not in the budget.  But I'd be curious to see scope traces from anyone else with the proper equipment and time (whenever that may be).

Link to comment
Share on other sites

The curve on the trace looks similar to a CR discharge so it may be a capacitive issue. If your recap included soldering in the area of Q2, Q3 & Q4 and you did not do so at the time try using flux cleaner on that area of the PCB as flux residue linking pins can potentially capacitively couple them.

If that does not make any difference or you cannot do that, how does the RGB drive signal look on the inputs to to Q3, Q3 & Q4 (check it both ends of R30, R37 & R46 respectively) is it clean or have the slope? 

  • Thanks 1
Link to comment
Share on other sites

Perhaps the Commodore CRT test is not valid.  I thought CRT TVs were known to smear to the point that software/hardware developers counted on the smearing.

 

If you have a video capture card, you could connect the upscaler output to video capture card.  Then view/compare the captured results to TV output.  That would completely eliminate the TV as a possible source of the problem.  Some minor chance your tv is applying a spatial filter, resulting in the smearing, but you would think the tv would allow all filtering/processing to be turned off when used with external upscaler.

 

I banged my head against wall in the following thread, where i did not have an upscaler in the mix:

Conclusion was the TV was applying spatial (and likely temporal) filters to the Jag output.  I had no external upscaler in the mix, and relied on TV upscaling.  Filters may still possibly be applied, even if input to TV is already upscaled.

 

Link to comment
Share on other sites

14 hours ago, Stephen Moss said:

The curve on the trace looks similar to a CR discharge so it may be a capacitive issue. If your recap included soldering in the area of Q2, Q3 & Q4 and you did not do so at the time try using flux cleaner on that area of the PCB as flux residue linking pins can potentially capacitively couple them.

If that does not make any difference or you cannot do that, how does the RGB drive signal look on the inputs to to Q3, Q3 & Q4 (check it both ends of R30, R37 & R46 respectively) is it clean or have the slope? 

The 1uf caps that carry RGB to the Y-C encoder are kind of near the transistors.  Except the transistors are SMD on the top and the caps solder trough hole from the bottom.  But to be safe I just used 99% IPA to clean the top around the transistors and around the bottom where the 1uF caps went through.  Made 2-3 passes on each side of the board.  No change in the smear or the scope image.

 

I checked both sides of the R30, and one side of the other two resistors (37, 46).  I didn't really see much of a difference if any from side to side of R30.  But they not only all have the curve on the falling edge, but they have what looks like a curve on the rising edge too.  Basically like a shark fin:

RR30.thumb.png.654b7f05d0530c2eb3a261529d5f3f5c.png

 

8 hours ago, jguff said:

Perhaps the Commodore CRT test is not valid.  I thought CRT TVs were known to smear to the point that software/hardware developers counted on the smearing.

 

If you have a video capture card, you could connect the upscaler output to video capture card.  Then view/compare the captured results to TV output.  That would completely eliminate the TV as a possible source of the problem.  Some minor chance your tv is applying a spatial filter, resulting in the smearing, but you would think the tv would allow all filtering/processing to be turned off when used with external upscaler.

 

I banged my head against wall in the following thread, where i did not have an upscaler in the mix:

Conclusion was the TV was applying spatial (and likely temporal) filters to the Jag output.  I had no external upscaler in the mix, and relied on TV upscaling.  Filters may still possibly be applied, even if input to TV is already upscaled. 

 

I don't believe the modern TV is causing this.  I have hooked up several other consoles using RGB into the scaler, and their images are always clean and crisp.  I know the effect of CRTs you are referring to though.  It is more of a blending, rather than a smearing.  Making pixels look almost rounded in some cases, and also can be used with dithering to create some transparency effects.

Link to comment
Share on other sites

8 hours ago, nick3092 said:

I checked both sides of the R30, and one side of the other two resistors (37, 46).  I didn't really see much of a difference if any from side to side of R30.  But they not only all have the curve on the falling edge, but they have what looks like a curve on the rising edge too.  Basically like a shark fin:

Without another comparing against another Jaguar displaying the same image if they all look the same then we can only assume that is like normal.

As for the curve on the output of the transistor at TP17, again without a direct comparison with another Jaguar it difficult to know for sure whether or not that is normal or indicative of a fault, but if the latter (i.e. Blue and Green are not similarly affected) then I would suspect a failing Q2 may be the problem.

 

I noticed that your scope probe setting is set to x10, if Blue and Green are square without the curve then this is irrelevant but for completeness did you remember to calibrate the probe when in its x10 setting before taking the measurements? I expect that anyone using such an expensive scope would know to do that, I ask just to be sure that what you are seeing is not the effect of an incorrectly calibrated scope probe which is sending us all on a wild goose chase.

Link to comment
Share on other sites

4 hours ago, Stephen Moss said:

Without another comparing against another Jaguar displaying the same image if they all look the same then we can only assume that is like normal.

As for the curve on the output of the transistor at TP17, again without a direct comparison with another Jaguar it difficult to know for sure whether or not that is normal or indicative of a fault, but if the latter (i.e. Blue and Green are not similarly affected) then I would suspect a failing Q2 may be the problem.

 

I noticed that your scope probe setting is set to x10, if Blue and Green are square without the curve then this is irrelevant but for completeness did you remember to calibrate the probe when in its x10 setting before taking the measurements? I expect that anyone using such an expensive scope would know to do that, I ask just to be sure that what you are seeing is not the effect of an incorrectly calibrated scope probe which is sending us all on a wild goose chase.

All 3 RGB analog pulses look the same before and after their respective transistors. So whatever it is, is happening across the board.

 

I would have calibrated the probes back when I got the scope, probably 2018 or 19. I don't use it a lot, just to try and help chase down oddball issues like this. Plus I would expect the digital pulses to look off if the probe wasn't calibrated. Just to be sure, I put the probe back on the calibration terminals, and saw a nice clean square wave. But noticed the calculated voltage was off. Looked at the probe (which I normally keep at 10x at all times), and found it was switched to 1x. I must have bumped the switch either while using it or putting it in/taking it out of the storage bag. 

 

I put it back to 10x and the calibration was still good. Rechecked the two test points on red (17 and 88) and the waves looked very similar (if anything, the curve on the rising edge maybe looked a little more pronounced with the probe on 10x). So I don't think the 1x probe was loading the circuit (or at least not too much).

Link to comment
Share on other sites

Posted (edited)

I found a somewhat reasonably priced (at least based on what working ones are going for) "won't power on" Jaguar and ordered it. I hoped it was simply the seller not using a cartridge. Unfortunately it still did not power on with a cart. Opened it up to find not only what looks like a bulge or ring of some kind on u38, it was also installed backwards. And whoever did this also removed the current sensing resistor and didn't reinstall it. Sigh.

 

Hopefully when the replacement parts show up mid to late next week there is nothing else wrong and I can compare the output of this Jaguar.

 

IMG_5510.thumb.jpeg.062c59e8e6a62c3376b3c1147de9f0e1.jpeg

Edited by nick3092
  • Sad 1
Link to comment
Share on other sites

3 hours ago, nick3092 said:

Opened it up to find not only what looks like a bulge or ring of some kind on u38, it was also installed backwards. And whoever did this also removed the current sensing resistor and didn't reinstall it. Sigh.

omg-oscar.thumb.gif.29605bb880104d366ae2630c3cdb77b8.gif

Link to comment
Share on other sites

On 5/17/2024 at 3:48 AM, nick3092 said:

JagRGBSCARTscaler.thumb.jpg.2b31c788ebe263f00ddc501f7eab59f6.jpg

My Jaguar looks very similar when connected to a video capture card using composite or S-video. I've never tried to plug it into a RGB CRT screen myself, but I've seen other people doing so, and it doesn't appear to have that issue. So I thought it was caused either by the video capture card or the non-RGB video output doing that.

 

I'll have to take a look with my oscilloscope to compare.

Link to comment
Share on other sites

All the parts showed up and I was able to get the 2nd Jag up and running. And it also has the smear. The scope traces at tp17 and 88 pretty much match my first jag, with the falling edge having a curve. I didn't look at the digital pins on this one, as I don't think it matters at this point. 

 

I am far from a Jag expert. But as I understand it, there were M and K versions of the jag, based on the serial number. My original is an M, while this one is a K (assuming no one swapped shells on either for some reason). My original has a sticker saying it's Rev R/P. This second on says Rev R. I thought I read there were 3 different Toms. Both these units have a Motorola Tom, both labeled v1.0. So maybe it makes sense they both have a smear if it's coming from the Tom. 

Link to comment
Share on other sites

Apologies for the delay, I had some unplanned things to take care of.

 

I did some oscilloscope measurement on a K Jaguar (TOM v1.0) and got virtually the same results as the ones you've posted. So your Jaguar is likely working fine.

 

I also found out that measuring the output of the video DAC (on R30) reliably is tricky -- even the parasitic 15 pF introduced by an 10x oscilloscope probe is enough to increase the signals rise and fall time noticeably. So instead I probed one of the digital input lines, and TP17. Here's what I got:

image.thumb.png.a3ba7e6975e030f80445a179178c1076.png

(Ignore the slight ripple on the traces ; this is a grounding issue introduced by the scope, it only appears when probing both signal at the same time)

 

The yellow trace is one of the digital signals input to the DAC ; as you can see, it's almost perfectly square.

The green trace is the output of the DAC after being buffered by Q2. It's noticeably asymetric (rise time is 37.5 ns, but fall time is 214.5 ns).

 

So I believe the issue is either:

1) The output stages in Tom are unbalanced, and can source more current than they can sink ; or

2) The transistor-based buffering circuit is distorting the signal

 

It would be interesting to disconnect the output of the DAC from the transistor to see how it'd look like without the buffering circuit.

 

 

Link to comment
Share on other sites

Posted (edited)
16 hours ago, Zerosquare said:

Apologies for the delay, I had some unplanned things to take care of.

 

I did some oscilloscope measurement on a K Jaguar (TOM v1.0) and got virtually the same results as the ones you've posted. So your Jaguar is likely working fine.

 

I also found out that measuring the output of the video DAC (on R30) reliably is tricky -- even the parasitic 15 pF introduced by an 10x oscilloscope probe is enough to increase the signals rise and fall time noticeably. So instead I probed one of the digital input lines, and TP17. Here's what I got:

image.thumb.png.a3ba7e6975e030f80445a179178c1076.png

(Ignore the slight ripple on the traces ; this is a grounding issue introduced by the scope, it only appears when probing both signal at the same time)

 

The yellow trace is one of the digital signals input to the DAC ; as you can see, it's almost perfectly square.

The green trace is the output of the DAC after being buffered by Q2. It's noticeably asymetric (rise time is 37.5 ns, but fall time is 214.5 ns).

 

So I believe the issue is either:

1) The output stages in Tom are unbalanced, and can source more current than they can sink ; or

2) The transistor-based buffering circuit is distorting the signal

 

It would be interesting to disconnect the output of the DAC from the transistor to see how it'd look like without the buffering circuit.

 

 

No worries, this is far from life or death. And I'm not going to lie, most of this is over my head. But I do see the similarities in my traces both before and after the DAC. I did run your two theories past my friend who was originally looking at this with me. And he said he was leaning towards the same conclusions. And before I even mentioned your suggestion of disconnecting the DAC from the transistor, he basically said the same thing. He said he'd like to see red with R30 disconnected.

 

While I'd be willing to do that, I'm not comfortable in my skills to do it. I tend to shy away from SMD work (don't have a hot air station). I've done some minor SMD work with my iron when there is enough open space. But not when it's this cramped. The only reason I was able to replace the regulators was because you have a lot of room to work when you pull the filter caps. And putting some Fastchip alloy on them made short work of removing them. In the case of R30, I'd be worried about collateral same to surrounding parts, and then not being able to put it back in. 

Edited by nick3092
Link to comment
Share on other sites

Posted (edited)
18 hours ago, cubanismo said:

You go first 😁

Well, since you asked nicely...

 

First, another example of why you should never fully trust Atari's documentation: do you know what's under the R30 resistor?

image.thumb.png.af6d8f8a287009cb19edc6bfcde84750.png

That's right. A shunt. So the 1 kΩ resistor is completely bypassed in practice.

 

If the shunt and the resistor are both removed, here's how the output of the DAC looks:

image.thumb.png.fa581f88f57a2d5ab74b77e51b79ffeb.png

The pulse is pretty symmetrical: rise and fall times are both roughly 105 ns. So Tom (and the DAC) are no longer suspect.

Note that response is pretty slow (a single pixel is about 150 ns long), but the parasitic capacitance of the oscilloscope probe probably makes things look worse than they actually are.

 

With the shunt removed, what happens if the original 1 kΩ resistor is reinstalled?

The answer is... not very interesting, unfortunately. The output of the DAC gets slower:

image.thumb.png.4e4f95e72af31cb25f70e82abb77285c.png

And the output of the transistor buffer stage is still very asymmetrical:

image.thumb.png.251a31f8ca4e0fa5199d4a63a7626f63.png

 

I guess the next step would be to try replacing the transistors with a better video buffer. Texas Instruments has a few chips with 3 channels, I think they'd be well-suited for that.

Edited by Zerosquare
  • Thanks 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...