Atari 7800 only black, white and yellow in 7800 mode

[Cmherndon79]

Hi there,

 

Got an NTSC 7800 in from someone that wanted it modded for composite video and audio output.  Board is CO25233-002 Rev A.  This is just the simple and cheap mod using the 2.2K, 3.3K resistors and the 2N3904 transistor, mounted on a PCB.  Done many of these on 2600s (maybe 150+ by now) and probably 10 7800s before.  When I got the unit in, I just booted up Frogger in 2600 mode over RF channel 3 and sound and video were working fine.  Foolishly I didn't check to see that the MARIA chip in 7800 mode was working before I did anything further.  I took the board out and replaced the 3 electrolytic capacitors from a kit from Console5 that I had on hand and replaced the audio polystyrene caps at C4 and C5 with so 820pF MLCC equivalents.  I removed the RF modulator from the board and I installed the composite mod.

 

I booted Frogger back up and the screen looked normal, colors were correct.  TIA audio was playing fine on the startup sound in Frogger.  Next, I put in Ballblazer since that had a POKEY chip in it, it is typically the first one I'll use.  When it switched into using the MARIA for the 7800 I got almost all black and white in the Atari rainbow, except for yellow.  When I watch Ballblazer in attract mode I see all black and white, but after the ball is hit, I see it turn yellow and after a second grey again.  No problems with POKEY audio.  Put in Food Fight and Centipede (7800) and same behavior. 

 

Started looking for problems on the board.  Noticed the rear leg of C57 was sitting in the mounting hole for the power connector, seemed unlikely this was the problem, but I connected it to where it was supposed to go and no change.  Checked all the component connections around MARIA and didn't find any problems, but did reflow a few things that looked like they could be marginal.  Checked for bad solder joints everywhere else and same thing, nothing apparent but reflowed a few points anywhere I thought things could be marginal.  Double checked the polarity of the electrolytics I installed and they were right.  Reflowed the solder on the 220uF cap sitting next to the MARIA just in case.  Did a search and found someone had B/W only here on a 7800, which was a bad solder connection at the slot, so I checked that, but didn't find anything wrong there.  Can't see any cut traces and checked continuity if I wasn't sure.

 

Just looking to see if anyone else has run into behavior that fits this description.  I've got a known working MARIA in a socket on another 7800, but I really don't want to desolder the one from this board, unless I'm fairly confident that is the problem.

 

Thanks!

 

 

[+-^CrossBow^-]

I did run into this with a 7800 that was sent to me for a UAV upgrade. The 7800 had one of the comp mods in it like you described but arrived with only a scrambled picture screen. I took out the simpler comp mod and restored the system back to factory (This one still had the RF modulator too as I recall). Anyway, I had to replace the small 74x logic chip on the left that sits just above maria first. That got me from a scrambled picture to one that was missing colors but at least the luma was working enough to see a picture. Finally traced it down to something in the maria itself. Couldn't see anything wrong with the board but once I found that out, I reinstalled the original in case if was a lose solder point..etc and same results. I think blues and purples were missing on this one. Anyway it was a bad maria in the end along with that 74x logic not providing a proper sync for the TV to latch onto. First and so far the only 7800 I've worked on that has done this and required the maria to be replaced.

 

And the missing colors only on 7800 games is kinda the giveaway here since the 2600 stuff is still all processed through the TIA. So..yeah...look at that maria and see if you have another on hand to swap into it to see if it makes a difference.

 

Also..if you haven't already make sure the color pot isn't to blame? Although that usually just gives a black and white or messed up colors but not just one color normally.

 

[Cmherndon79]

29 minutes ago, -^CrossBow^- said:

I did run into this with a 7800 that was sent to me for a UAV upgrade. The 7800 had one of the comp mods in it like you described but arrived with only a scrambled picture screen. I took out the simpler comp mod and restored the system back to factory (This one still had the RF modulator too as I recall). Anyway, I had to replace the small 74x logic chip on the left that sits just above maria first. That got me from a scrambled picture to one that was missing colors but at least the luma was working enough to see a picture. Finally traced it down to something in the maria itself. Couldn't see anything wrong with the board but once I found that out, I reinstalled the original in case if was a lose solder point..etc and same results. I think blues and purples were missing on this one. Anyway it was a bad maria in the end along with that 74x logic not providing a proper sync for the TV to latch onto. First and so far the only 7800 I've worked on that has done this and required the maria to be replaced.

 

And the missing colors only on 7800 games is kinda the giveaway here since the 2600 stuff is still all processed through the TIA. So..yeah...look at that maria and see if you have another on hand to swap into it to see if it makes a difference.

 

Also..if you haven't already make sure the color pot isn't to blame? Although that usually just gives a black and white or messed up colors but not just one color normally.

 

Thanks! I'll give the MARIA a shot. I did check the color pot and reflowed it as well just in case, so fairly certain that isn't the issue.

[Cmherndon79]

K, so desoldered the MARIA from the board I'm working on.  Removed the known working MARIA from the socket of the other working 7800 I've got here.  Installed the desoldered MARIA and we're getting color which appears normal as I can tell.  Thinking it must be something else.  At least it isn't an Atari chip that isn't made any longer, so that's a relief.  

[Cmherndon79]

Put the previously socketed MARIA into the 7800 lacking color, other than yellow and the problem follows the board.  Sure is an odd defect.  If chroma were not being mixed at all it would make more sense, but I'm getting yellow, and it isn't the combination as far as I know of Luma/Chroma into composite, at the high level, or else TIA graphics would have the same/similar issues, I'd think at least.  Might just have to set this one aside and think on it...  Thanks @-^CrossBow^- for the suggestion, as it gave me the confirmation I needed to think I wasn't crazy and it might be the MARIA.  Also, appreciate the video you did on the UAV for the 7800.  Had someone request S-Video on a 7800 and used your video to help me along that project.  It was fantastic and got everything wired up without any issues at all. 

[+DrVenkman]

Okay, if it's not the MARIA, it's gotta be one of the discrete components between MARIA and the points where you pick up signals for your video mod. Time to break out the schematics. But as a clue, the only 7800 game you tested with is Ballblazer, correct? Well, that playfield is predominately shades of green with a blue sky. So if you're seeing yellow and black, then you're missing the blue component of the output. 

 

Do you have any other 7800 carts to try before replacing more components on the board? Might help limit your work.

 

EDIT: Per the schematic, MARIA's LUM signals all get passed through U3 (I think that's the 74LS32 chip). That's probably the problem child.

 

EDIT EDIT: $0.61 each at Mouser. Cheap fix, aside from minimum order quantities and/or shipping charges. 

 

https://www.mouser.com/ProductDetail/Texas-Instruments/SN74LS32N?qs=q2XTDbzbm6DA9Mnew5GiLA%3D%3D

[Cmherndon79]

Thanks @DrVenkman.  Sounds logical.  Don't have any 74S32s on hand, mainly just logic chips for the C64, which the 64 doesn't use that model.  I usually place an order with Mouser about every 10-14 days or so, but may need to really take some inventory, since I just got my last order in yesterday.

 

I had tried Food Fight and Centipede (7800) earlier and the behavior is the same there.  Thanks for checking the schematics, as I don't have the 7800's readily available as I do the 2600.  I appreciate it!

[+-^CrossBow^-]

@DrVenkman is that the 74x logic that I was thinking of? I've had to replace that before at least 2 other times but not for color issues. As I said in my reply I had one of those go back before and it just resulted in a scrambled picture but it was doing that on both 2600 and 7800 games in my case.

 

I guess I need to break out my test 7800 and hook up the maria colorburst to my scope so a comparison might be possible?

 

[+DrVenkman]

6 hours ago, -^CrossBow^- said:

@DrVenkman is that the 74x logic that I was thinking of? I've had to replace that before at least 2 other times but not for color issues. As I said in my reply I had one of those go back before and it just resulted in a scrambled picture but it was doing that on both 2600 and 7800 games in my case.

 

I guess I need to break out my test 7800 and hook up the maria colorburst to my scope so a comparison might be possible?

 

Yeah I think it must be. It’s a Quad Or gate that passes either the MARIA  *or* the TIA LUM and SYNC signals on to the resistor ladder and then out to the RF modulator in an unmodified machine. If any of the four internal OR gates fails, it will stop passing signals properly for either TIA or MARIA.

 

Here’s the relevant chunk of the schematic:  

 

 

[Cmherndon79]

Thanks @DrVenkman.  Ordered a few 74LS32s from Unicorn Electronics.  Needed some 4164-150 DRAM as well, so they covered both needs.  Will update once the part comes in.  Thanks again.

[Cmherndon79]

HI @DrVenkman and @-^CrossBow^-  Thanks for all your help thus far.

 

Finally got my shipment in this evening with the 74LS32s.  Went ahead and swapped a new logic chip in, but unfortunately the problem still persists, with the behavior looking identical to before swapping the chip out.  Must be something else.  

 

Thinking back I had not checked the power adapter.  The one I currently have I built from a Lite-On switching power adapter with the 7800 plug installed.  I tried the same power adapter on my 7800, that has the same mod as the one I've installed installed and the color came in full.  Also, doesn't seem to have any affect at all when I turn the potentiometer at R57 on the color but I do see different resistance values from 35-195k ohms with a meter.  No matter the pot value it's always black, white and yellow.  Perplexing...

 

Any ideas?  Thanks!  Took some pictures of the TIA mode running Frogger and the Maria running Food Fight and Centipede, where you can see the behavior.

[+DrVenkman]

That’s truly bizarre. You’ve already ruled out MARIA, and there’s just not that much else between the outputs of the chips (MARIA and TIA) except the 74LS32 and the resistor ladder. I’d think the next step is you have a burned out resistor that’s not passing one of the LUM signals, except that TIA colors go through the same resistors. 

 

I’m out of ideas.

[+Mitch]

The 7800 chroma signal is on a different resistor than the 2600 chroma signal.

 

Mitch

[+-^CrossBow^-]

As Mitch stated, there are two different color signals on the resistor ladder. One for the MARIA and the other for the TIA. The only other options if those resistors check out is that there has to be a bad trace somewhere in the mix on the MARIA side to the resistor ladder and you would have to go through the schems to start tracing from each pin to each next component in the mix and so on.

 

MARIA COL is on R16 (#7 from the left on the ladder) and is a value of 4K3

TIA COL is on R17 (#8 from the left on the ladder) and is a value of 4K7

 

You might have to take them out of circuit to get an accurate read on them though. But just in case that helps.

 

*Update* - I just looked as the schems and there isn't anything between the MARIA and R16. Pin #43 off the MARIA is the color output into the RF section. So if all of that checks out there has to be something else before being sent to the MARIA causing those colors to be off kilter. Interesting that you stated changing the top trimmer doesn't seem to affect it at all? 

 

So there really isn't that much in the mix component wise on the 7800 side. You had the data lines from the cartridge port, going into both the BIOS, RAM, CPU and MARIA. There is of course stuff mixed in from the TIA side for the audio and the like. Wonder if one of the RAM chips could be to blame? Do you have one of those bodge resistors attached in the middle between the two RAM chips that is tied to ground I believe? I've never bothered to really study that, but I do know that I see that resistor added in on more 7800s vs not.

 

 

 

[Cmherndon79]

So, checked out R16 and we're reading 3.928k there, so that seems ok.

 

And yeah, moving that pot from one side to the other, the screen doesn't change one bit.  Everything that was black is always black, everything that is white is always white and everything that is yellow is always yellow and it's solid while I turn the pot from one extreme to the next.  No flickering, brightness or color changes at all.

 

I do have that bodge resistor installed between pin 4 of the lower RAM chip to ground installed.  Early on I removed that resistor to see what would happen and the video quality degraded further, entire parts of the image would be missing, etc.  I reinstalled it onto the bottom of the board afterwords and the screen reverted to the previous black/white/yellow state.  Resistor reads at 1k, which I think is correct.

 

Guessing trace hunting is my next step.

[+-^CrossBow^-]

Well..the fact that the video image did change when you removed that resistor could indicate that the RAM is part of the issue? I've not personally seen a RAM IC go bad in a 7800 yet, but I'm sure it is possible. Again, the schems don't show much in the mix on the 7800 side. Might need to check the traces from the trimmer since you stated you are getting different values on the trimmer when you adjust it in measurement but nothing changes on the screen.

 

R57 in the schematics is the trimmer color for the 7800 side. See the snippet below. R57 goes into Pin 9 on the MARIA. And the trimmers are 500k so the readings seem off on yours? Maybe the trimmer is part of the problem?

 

 

[Cmherndon79]

Well.  I got color!

 

I walked back all the traces and didn't find any issues.  Just out of not knowing what exactly to do next, I powered it up and of course nothing changed.  Figured I'd turn it back and forth some more and I did that for a few rotations and suddenly as I hit the furthermost counter clock position the Food Fight logo changed to a shade of red, from grey previously, turned it to the previous setting per a mark and it looked normal.  Powered it off and back on and its back to black, white and yellow again.  Rotated completely counter clock wise and again, color returned, moved it back to the correct position and let it sit and the color kind of pulses various shades periodically and then after some time the color went out again. 

 

Appears I need to order a trimmer.  

 

Or perhaps I could just pull one from a 2600 parts board?

Edited June 2, 2020 by Cmherndon79

[+-^CrossBow^-]

Well, you could also try spraying the trimmer down with some contact cleaner, rotate it back and forth a bunch to get it all in there. Follow that with an ISO rinse...and see if anything stabilizes?

 

[Cmherndon79]

Cleaned out the trimmer as suggested with some WD-40 Electrical Contact Cleaner, rotated from both extremes about 10 times, sprayed in some 99% isopropyl and did about 5 more rotations.  Walked away for a few hours to make sure everything dried out and the behavior was unchanged.

 

Pulled a part off a 2600 parts board and did the same cleaning on it. I measured the points to see that it would indeed go from 500k to somewhere in the 300s.  Removed the original trimmer and installed the donor part and, well it was the same as before.  Looked down at the board and back up again and color sprang to life, I made the proper adjustment and it appears solid in color, vs drifting in shading.  Tried a power cycle and the color survived, which didn't ever happen with the old trimmer.

 

Letting it run for a while.  Unsure about the initial results with the donor trimmer.  Perhaps in my haste I had not given proper time to let the isopropyl alcohol dry?

 

Things are looking up at least.

[+-^CrossBow^-]

I would have to look over the schems some more, but I wonder if the fact that the 2600 trimmer is in circuit along with the 7800 trimmer with each going to their own chips, if something in the 2600 trimmer isn't causing an issue. Since I have a good way to desolder components, I would likely be inclined to remove the 2600 trimmer completely just to see if it makes a difference or not. Because if the trimmers are good and everything else is the way it should be, you should be able to get colors nearly the entire time you are adjusting the 7800 trimmer and go from normal to way to cold to warm and all the psychedelic palettes in between.

[Cmherndon79]

Might need to do that. Certainly not out of the woods. I just returned to it and we're back in the old state.

 

Edit:

 

Pulled out R56 and it's still showing the same behavior.  Was able again to rotate it R57 completely counter clock and color came back.

 

Still got the trimmer from the 2600 at R57.

 

So, not some behavior introduced by R56 and not the trimmer at R57.  

 

Hmm...

Edited June 2, 2020 by Cmherndon79

[Cmherndon79]

Pulled out R16 and measured outside and it's reading 4.26k, so that looks good.  Don't have any 4.3k resistors, but did try 2 2.2k in replacement and the behavior looks mostly the same, just seem to get a few more yellow pulses during the Atari logo startup.

[+-^CrossBow^-]

You have tried this with a different TV right? Because aside from the trimmers for the color and few passives, there just isn't much between that and the MARIA and you already changed that out with the same result. Next would be to check all the traces above and below the board that the MARIA uses to make sure there isn't a broken trace. If there were, changing the Maria wouldn't change the condition.

 

Could also still point to a possible issue with the RAM as it seems that is only used for the 7800 games if I remember correctly.

 

[Cmherndon79]

Thanks for coming back around on this one.  I'd set this hardware aside for a while to think and had not come up with anything...

 

I have not tried a different TV.  It would be a good point to try.  I did get another 7800 in last week that I refurbed and modded and it worked fine on the same TV and I've never had this problem before now, but I should hook it up to my tube TV and see if anything changes.  The TV on my rework bench is just some cheap LCD that I hung from the ceiling for space reasons from a thrift shop.

 

Are there any diagnostics for the 7800 that are capable of testing the RAM?  And yeah that is correct, it's full color when using the TIA 100% of the time.

 

Thanks!

[+DrVenkman]

1 hour ago, Cmherndon79 said:

Are there any diagnostics for the 7800 that are capable of testing the RAM? 

Well, there's a 7800 Diagnostic Cart but unless you happen to own a physical copy of it; or own a very rare and expensive Cuttle Cart 2; there's not a way to load it on your console.  The lack of a readily-available 7800 flash cart is a real pain sometimes.