PM 7100/80 with distorted chime

[Zarwox]

I have a Power Mac 7100/80 with a distorted chime and no video output. Probing the onboard memory there is reasonable activity up until right after the end of chime when the CPU appears to crash with what looks like just noise on the address and data busses.

Machine does not respond to any key combinations at boot time that I've found except ctrl+apple+power for reset, but holding down NMI during reset does produce the silly car crash sound which does not sound distorted like the chime (although it's a bit hard to tell).

What is done so far:

• Verified power rails good
• Running the board without SIMMs, cache module, drives or other nubus cards
• Replaced dried up thermal compound under CPU heat sink
• Tried moving the ROM card to the cache slot since they are in parallel
• Tried a different known good power supply
• Replaced all capacitors. They did not appear to leak and non tested totally bad after removal but replaced them anyway
• Tried another known good PDS video card
• Replaced battery with a CR2032 (voltage high enough?)

No change.

Now I'm suspecting bad onboard RAM. So my question is if it's possible to run the 7100 without the onboard memory entirely using only SIMMs? Of course the onboard video probably will not work but it would be an interesting test. Just want to verify my assumptions before I start de-soldering. Anything else I should try? Is there maybe a quick way to disable the onboard RAM without desoldering chips?

 

[Durosity]

Out of interest did you try removing the onboard ram chips?

 

[Zarwox]

I actually recently did. Or rather I was a little reluctant to remove all that DRAM so instead I disconnected the shared RAS line from the bus driver and pulled it high (the SIMMs have their own dedicated RAS lines), and installed 4x8MB known working SIMMs. By doing that my idea was that the onboard DRAM would never be selected and thus invisible. That however resulted in neither chime nor video. Only repetitive clicks and pops through the speaker so I think the on board RAM really is mandatory.

What I did then was to tone out all traces from the onbord DRAM to make sure they made their way to the buffers, memory contoller and data path chips. Found no broken or shorted traces. Used the 8100 schematic for reference. Even though the designators are wrong the design appears to be the same. Also scoped all DRAM pins during reset/post/chime looking for anything weird but all signals were wiggling and at expected levels.

Now I'm a little bit out of ideas. My best theory is still bad DRAM. The problem in that case is figuring out which one(s). Since the chime is distorted I had an idea of capturing the serial audio data entering the AWACS during the chime. Maybe there is some pattern that can be seen there? If not then I may just replace all onboard memory and see what happens. Got a few NOS chips on the way but it may be safer to strip down a known working SIMM instead.

Ideas and thoughts are welcome!

 

[Durosity]

Got a few NOS chips on the way but it may be safer to strip down a known working SIMM instead.

Yeah I'd agree, at least you can test them elsewhere!

Alas I don't really have any further ideas on what may work, other than have you tried not having the HPV card installed?

I've got a similar issue with a lightly battery bombed 7100, it doesn't actually chime but it tries to. There's definitely a few vias near the battery that are knackered, but I couldn't find a schematic to tell where they went to, but I'll have a look at the 8100 one since it'll likely be similar! Do you have a link for that?

Let me know how you get along, might be helpful with my troubleshooting!

 

[Zarwox]

I made progress! I think the memory has been falsely accused. I found two broken traces in the IOData bus going from the data path chips to the rest of the system. In my case bit 1 and 3 were floating which matches the distortion in the chime.

Now the machine has a clean chime and produces video with a grey screen and a movable mouse pointer. But there it hangs. There is a short blipp of SCSI activity if I hook up a BlueSCSI but not much else. The usual cmd+opt+p+r shortcut behaves as expected and resets the machine if held down. Will look for more broken traces now I guess.

If you have even light battery damage on your 7100 I think chances are it may have the same issue with broken IOData bus traces. They have vias right under the battery holder. Unfortunately they dive down into an inner layer which complicates things.

 

[Durosity]

Now the machine has a clean chime and produces video with a grey screen and a movable mouse pointer. But there it hangs

I had a SE/30 that did exactly that, and it was due to bad contacts in the scsi controller. Might be worth reflowing that as a potentially easy fix!

you have even light battery damage on your 7100 I think chances are it may have the same issue with broken IOData bus traces. They have vias right under the battery holder. Unfortunately they dive down into an inner layer which complicates things.

Yeah that seems entirely likely as the vias in that area are utterly destroyed. I will check out the schematic you attached and I’ll see if I can work out where they go to! Thank you!

 

[Zarwox]

I had a SE/30 that did exactly that, and it was due to bad contacts in the scsi controller. Might be worth reflowing that as a potentially easy fix!

You're right! That was the problem. The REQ and RST lines were not connected. Hooking up a SCSI device to the bus got it stuck on a grey screen and booting without any device at all on the bus gave the usual ?-disk icon. With that fixed the machine now boots without issues

So even though there appears to be very light battery damage on these machines the routing of critical signals right below the battery socket can cause problems.

 

[Durosity]

You're right!

According to my wife I'm never right.

So even though there appears to be very light battery damage on these machines the routing of critical signals right below the battery socket can cause problems.

Damn those Apple engineers in 1993 not considering the potential issues caused by running those traces under the clock battery.