Mac SE - Single Pixel Thick Verticle Jail-lines

[KGLlewellyn]

So I've got a Mac SE which until recently was working perfectly. However, I have installed a set of SIMMs totaling 4MB, having increased this from the 1MB SIMMs.

After installing the new SIMMs, the machine started getting these one-pixel thick vertical jail-lines. Outside of the annoying visual issue, the machine appears to continue to work perfectly. It boots up, runs applications without any issue.

Wondering if it was the new SIMMs at fault, I attempted to put the original ones back in, but to my dismay the issue persisted. In both instances the machine correctly detects the available RAM.

Are we looking at a RAM issue here, or something unrelated?

 

[Crutch]

This happened to me recently on an SE/30 - in my case it was a bad analog board. Probably needs a recap but so far I just swapped it out - all good now. Either way I doubt both your old and new SIMMs are bad!

 

[KGLlewellyn]

This happened to me recently on an SE/30 - in my case it was a bad analog board. Probably needs a recap but so far I just swapped it out - all good now. Either way I doubt both your old and new SIMMs are bad!

I have an SE/30 too which has a good analog board. I wonder if the SE board is compatible with it?

 

[KGLlewellyn]

Hmm...I'm thinking this is probably more of a logicboard issue than an analog board issue. There could be a damaged trace between the SIMMs and some other component that's involved with drawing the video. I don't really want to mess around with my SE/30 as it's got a lot of sentimental value, so I won't really be touching it.

I have tried cleaning out the SIMM slots and re-doing the solder joints on the analog board to no avail. I think I need to try the board in another SE to see if the issue re-occurs there. If anyone's got a schematic for the SE board, it'd probably be quite handy.

 

[BadGoldEagle]

You can install your SE board in your SE/30. They‘re compatible with one another. In fact my SE/30 was originally an SE FDHD that was treated to the Apple upgrade plan in late 90. 

The way the SE deals with video is way less complicated compared to the SE/30. RAM plays a big role here.

I see you‘ve just upgraded from 1MB to 4MB, did you remove the jumper on the logic board? If you have the older style of logic board, you need to clip one end of a resistor. 
 

Given that it was working fine previously (ie before the upgrade), the problem should go away if you install your 256kb SIMMs back in (and reconfigure the jumper/resistor).

 

[KGLlewellyn]

That's one of the interesting things. When I originally upgraded to the 4MB, I forgot to clip the resistor, but the machine did boot up albeit still with the vertical lines. So I reverted back to the original SIMMs, but the problem still persisted.

So I did try to remove the resistor, but no dice, still a problem. My guess something broke when I was doing the upgrade, but yeah, I'm not sure what circuits I should be verifying with continuity on my multimeter. I can't figure out how the video goes from the RAM to the analog board, if I can get that figured out, I might get an idea on how to fix this.

 

[quorten]

If you mention damaged trace on the logic board... let me put up my guess from what little I know of the Macintosh SE logic board.

The BBU chip is responsible for fetching video as a 16-bit word at a time from RAM by controlling *DTACK to prevent the CPU from accessing RAM at the same time.  (Unless a PDS expansion card requests priority over *DTACK via the *EXTDTK control signal.)  Once the 16-bit video word is grabbed, it is stored in an internal shift register and the data is shifted out verbatim to VIDOUT, most significant bit first.  Logic level zero is white, logic level one is black.  (The analog board logically negates the signal as part of its processing to generate the CRT drive signal.)

My guess is that if there were a short, it would cause an overall system malfunction since all RAM reads would have a dead bit.  I guess it could be possible the shift register in the BBU was damaged in a way that creates a "dead pixel," but I can't imagine improper configuration of the MBRAM and ROW2 RAM configuration inputs would cause this.

Nevertheless, you can check trace RDO13 (and adjacent traces) and see if it is shorted.

 

[quorten]

Regarding schematic, I've retraced the nigh unreadable incomplete fragment of a schematic and have it here in my GitHub repository.  Pages 2 and 3 are nowhere to be found on the Internet, but at least page 1 covers everything you need to know about your circuit in question.

 

[KGLlewellyn]

Righto, apologies for the massively delayed follow up.

@quorten, thank you for your input, your recreation of the SE schematic has been very helpful. So far I think I've been able to verify that RDQ0 to RDQ15 have continuity between the LS245s and the BBU. You mentioned regarding RDO13? Was that a typo and you meant RDQ13 or is there a different trace you're referring to?

I have also tried the board in my SE/30 chassis, and the issue persists, so we can be absolutely confident that the issue is on the logicboard itself.

I'm going through it with a microscope to find any damaged traces, so far I haven't found any, although there are quite a few that are discolored. Though my experience has been that's just the solder mask and the trace itself is okay.

Of note, the vertical lines are persistent, all the way through from the start of the RAM test with the grey screen to the booted desktop. Outside of the visual irritation, the machine operates just fine. I have a feeling I need to follow back from the board connector's display outputs back through the board to see what's up.

Just for giggles I have fully cleaned the board with IPA and rinsed with distilled water, it's looking great now, but the issue still persists.

I'm just hoping this isn't a busted IC that's causing this. :/

 

[rollmastr]

Did you set the jumper or clip the resistor after upgrading the RAM? Does the issue persist when going back to the original RAM?

 

[KGLlewellyn]

Yep, already done that - mentioned it earlier in the thread.