XEGS red/brown screen of death.

[deffroe]

Afternoon all.

A few weeks ago I got an XEGS from ebay sold as Spares or repairs, in the hope that I could resurrect it, but I'm starting to hit a brick wall.  I started off swapping out GTIA followed by ANTIC with no change.  Researching online I've seen mentioned that it could also be ROM, CPU, PIA, RAM and MMU.  I already have a GS, which was with a friend for a couple of weeks until this weekend, so was able to continue swapping out chips.  So, swapped out CPU, ROM, RAM, MMU and finally PIA.  Each chip has been done one at a time, desoldered, cleaned and checked before adding a socket and then swapping chip with/on known working machine.  Each try the result has been the same, murky red/brown screen.

The board itself looks pretty clean, there was no signs of any working being done, although C56 to C59 are radials not axial they don't look like rework.  I'll add photos in a bit.

I've checked out various websites and youtube videos, Tynemouth Software http://blog.tynemouthsoftware.co.uk/2018/10/atari-xe-system-xegs-repair.html was looking like the same situation as mine, right up to replacing PIA and now I'm not sure what to try.

 

Any suggestions graciously accepted.

 

Regards

 

Phil

[deffroe]

Here's a couple of motherboard photos..

Phil

[Beeblebrox]

I had the same with an xegs last year. The ram was definitely failing as I tested the chips in a dram tested. Replaced them but still had issues. I socketed the board for those chips you also have socketed. 

 

Remember the xegs has its own custom MMU and os rom. 

 

I'd revisit the socket work and check the MMU and os rom sockets and chips are OK. 

 

Also when left on do any particular chips still get almost too hot to touch? 

 

 

Edited June 11, 2023 by Beeblebrox

[deffroe]

Been checking voltages to the main chips and checking temperatures too.  Noticed the MMU in the poorly GS is a little hotter then my working machine. 30c Vs 28c, only 2 deg.C Not sure if that would affect things significantly.

Also took some screenshots while I'll add below, the brighter screens are RF the darker is composite.  In person they are not so bright, the composite shot is really dark.

 

One thing I noticed this time is the vertical lines, it was as if the were moving in and out of focus when connected by RF.

 

[deffroe]

3 minutes ago, Beeblebrox said:

I had the same with an xegs last year. The ram was definitely failing as I tested the chips in a dram tested. Replaced them but still had issues. I socketed the board for those chips you also have socketed. 

 

Remember the xegs has its own custom MMU and os rom. 

 

I'd revisit the socket work and check the MMU and os rom sockets and chips are OK. 

 

Also when left on do any particular chips still get almost too hot to touch? 

 

 

MMU and OS have only been swapped between XEGS', likewise with ram. 

I've learnt the hard way how fragile XE boards are so take a lot of precautions and check every step.  But will give all the sockets a check over.

 

Thanks

 

Phil

[xrbrevin]

did you swap out the chips individually? its possible you have more than one culprit so it worth trying multiple substitutions at once

[Beeblebrox]

11 minutes ago, xrbrevin said:

did you swap out the chips individually? its possible you have more than one culprit so it worth trying multiple substitutions at once

Yes it looks like this was the case. 

 

Also the ram chips swapped in from the other xegs work absolutely fine in the other xegs swapped back into the working one? 

 

Generally it's not a good idea to swap good chips from a working system into a bad one. You put the bad machines chips (one at a time) I to the working machine. 

 

I am not sure re temps but that sounds OK. If a chips is running hot you'll know about it. 

[deffroe]

16 minutes ago, xrbrevin said:

did you swap out the chips individually? its possible you have more than one culprit so it worth trying multiple substitutions at once

Yes, one by one.  Each chip removed was tested in my working GS and then the chip from that swapped to the none working GS.  Each location was cleaned with IPA(not the beer) and checked before socketing.

 

Phil

[deffroe]

4 minutes ago, Beeblebrox said:

Generally it's not a good idea to swap good chips from a working system into a bad one. You put the bad machines chips (one at a time) I to the working machine. 

 

I am not sure re temps but that sounds OK. If a chips is running hot you'll know about it.

Yeah all done one by one.  Thanks regarding temps, think I was clutching at straws  

 

Phil

[deffroe]

56 minutes ago, Beeblebrox said:

I'd revisit the socket work and check the MMU and os rom sockets and chips are OK. 

Finished tracing out the RAM, ROM and MMU sockets to various points on the motherboard, all good there.

 

Phil

[TZJB]

3 hours ago, deffroe said:

Here's a couple of motherboard photos..

Phil

 

Unfortunately your images are out of focus so we are not going to see anything. However it seems that you have done all the right things so far. Do you have an oscilloscope to check the clocks with?

 

This document has the relevant frequencies to check starting at Freddie. OSC from Freddie pin 37 going to GTIA pin 28 is the most important clock for the machine to sync from.

 

XEGS_Original Diagrams.pdf

[deffroe]

15 minutes ago, TZJB said:

Unfortunately your images are out of focus so we are not going to see anything. However it seems that you have done all the right things so far. Do you have an oscilloscope to check the clocks with?

 

This document has the relevant frequencies to check starting at Freddie. OSC from Freddie pin 37 going to GTIA pin 28 is the most important clock for the machine to sync from.

I only have one of those cheapie pocket scopes but should be enough for this.  I'll give it a go after work tomorrow.  thanks TZJB

[xrbrevin]

3 hours ago, Beeblebrox said:

Generally it's not a good idea to swap good chips from a working system into a bad one. You put the bad machines chips (one at a time) I to the working machine. 

yes good point. if the working XEGS then fails to operate, it will highlight a dud.

[deffroe]

7 hours ago, xrbrevin said:

yes good point. if the working XEGS then fails to operate, it will highlight a dud.

That's what I hoped...All the main system chips bar Pokey, Freddie and IC logic have been socketed and tested working.

[Rybags]

A non black background indicates that the OS isn't starting up which could mean CPU, OS Rom, MMU or memory select logic as prime suspects.

A diag mode cart such as Star Raiders can sometimes be helpful since it will run immediately bypassing the startup sequence.

The vertical lines mainly on the left are usually noise from Ram refresh.  Sometimes when the TV is out of tune or there is such interference you can actually see signs of life of the CPU as the noise can change depending on what's going on.

 

The Ram as used in XEGS is fairly common, even used in some earlier PC graphics cards and trying some others would be a good idea.

Of course the XE[GS] is mostly not socketed so not as easy as most earlier machines.

[TGB1718]

10 hours ago, deffroe said:

I only have one of those cheapie pocket scopes but should be enough for this. 

If that's what I think it is, it only really works up to a few 10's of KHz, certainly not in the MHz range,

hope yours is better than mine

[mimo]

I'd go for CPU , just my experience with 3 xegs it has been the crappy mexican Rockwell CPU every time 

[Rybags]

Maybe monitoring the Sync pin could be sufficient.  Does it still pulse if a lockup code has been executed?

[xrbrevin]

freddie could be worth a check, it handles RAM operations

be careful when desoldering - the XE models are not as resilient as their predecessors due to manufacturing cost-cuts

Edited June 12, 2023 by xrbrevin
syntax error

[TZJB]

4 hours ago, mimo said:

I'd go for CPU , just my experience with 3 xegs it has been the crappy mexican Rockwell CPU every time 

 

I am sure it has been changed, bar Pokey, Freddie and logic chips.

 

6 hours ago, deffroe said:

That's what I hoped...All the main system chips bar Pokey, Freddie and IC logic have been socketed and tested working.

 

[deffroe]

Evening all,

22 hours ago, TZJB said:

. OSC from Freddie pin 37 going to GTIA pin 28 is the most important clock for the machine to sync from.

I get 3.54Mhz at both pin locations.

 

11 hours ago, Rybags said:

Maybe monitoring the Sync pin could be sufficient.  Does it still pulse if a lockup code has been executed?

Is this pin7 on Sally?  If so, it just shows a steady 5v.

 

One thing I forgot to mention, and I wanted to wait til I got home to double check, inserting Star Raiders shows no difference.

 

Something else I've tried this evening, not sure if it help with diagnosing, I connected a Sophia 2 DVI.  The TV picks up a signal but only shows a black screen.

Edited June 12, 2023 by deffroe
Clarity

[flashjazzcat]

CPU is certainly hanging almost immediately, so if you already swapped the CPU from another machine and it didn't fix the issue, maybe you're unlucky and have a bad CPU plus faulty RAM/ROM or MMU, etc, which would prevent the machine from working even when you swapped ICs one at a time.

Edited June 12, 2023 by flashjazzcat

[deffroe]

21 minutes ago, flashjazzcat said:

CPU is certainly hanging almost immediately, so if you already swapped the CPU from another machine and it didn't fix the issue, maybe you're unlucky and have a bad CPU plus faulty RAM/ROM or MMU, etc, which would prevent the machine from working even when you swapped ICs one at a time.

Evening fjc,  I removed each chip one by one but as I swapped them and saw no difference I've been keeping them like that.

 

Which pins on Sally should be checked with a scope to show if it's hanging?

 

Regards

Phil

[deffroe]

Here's a couple of better less blurry photos.

[flashjazzcat]

5 minutes ago, deffroe said:

Which pins on Sally should be checked with a scope to show if it's hanging?

Grab a pinout via Google and check the address and data pins. You should see some activity there. We know the system crashed early on since it didn't get far enough through the OS initialisation to even reset the screen background colour to $00.