Starting a work thread for debugging multiple issues with my Quadra 950.
State Upon Receipt
I picked this unit up online and it came in a non-working state. The power supply was dead (various issues on both the primary and secondary sides).
ATX Power Supply
Put together an ATX conversion of the PSU using a Corsair HX1200 which drives the right amount of current for the +5V rail for the Q950. This allowed me to boot and things seemed mostly ok. But I would get a very loud speaker pop on cold boot and I would get some odd crashes (illegal instruction, bus error, bad address, question mark icon even with a known good disk, etc).
Loud Speaker Pop
I heard Q950 owner do experience a speaker pop on boot but I didn't think it was supposed to be this loud. I borrowed a friend's stock working Q950 power supply and the pop was there but much quieter. I put both PSUs on a scope and the only difference I could see is that on the stock PSU, the +5V/+12V/-12V rails ramp up at the same time. The ATX PSU however has +12V coming first about 12-14 ms before the +5V/-12V rails. This seemed problematic. ATX power supplies don't have a requirement to bring up the rails at the same time as they utilize a ready signal to the logic board when the rails have stabilized.
Loud Speaker Pop Resolved
I put together a small circuit that gates +12V against the ramp up of the +5V rail. This got all three rails really close during ramp up. With that change, the loud speaker pop was much more muted, like the stock PSU.
OS Crashes
I was hoping the PSU fix would resolve the crashes but it did not. On boot (and sometimes on restarts), I get a lockup shortly after the "Welcome to Macintosh" screen appears. This means that the System file has loaded and the machine is booting. It's not always the same error. Sometimes I get an illegal instruction error, sometimes a bus error, sometimes a bad address error. Sometimes, it stops finding the disk and shows the question mark icon. In these occasions, it looks like the disk image got corrupted and I have to replace it.
Again I borrowed my friend's PSU thinking it was the PSU but I was able to reproduce with the stock PSU as well.
Setup
Q950. System 7.6.1. 128MB of memory. ZuluSCSI for the disk. Machine does have the stock SCSI terminator at the end of the internal SCSI ribbon cable. And I am not using termination on the internal ZuluSCSI.
Debugging Attempted
Board looks super clean. No corrosion. Chips are clean. No missing components.
Reproduction
The reproduction is not easy. I get the issue about 1 in every 5-10 cold boots. I get it about 1 in every 50 warm restarts. I have never yet hit the issue when no disk is installed - I have so far always made it to the question mark disk screen when there is no disk. That seems to suggest data and address lines from memory/ROM/cpu are ok.
I'm narrowing on the SCSI subsystem. The fact that it happens on both the internal and external means it's most likely not specific to the SCSI chips (there's two). If I am able to get through boot, the system behaves ok after that, no crashes. Even running a few runs of MacTest Pro and MacBench doesn't introduce a crash. It feels like something on boot up, perhaps power related, is inducing the issue which is why I thought it was the PSU but given I can repro on a known working PSU, maybe I have to look at the caps on the logic board.
What I'm currently looking into...
I printed out @Bolle 's Q950 schematic which he put together to design his Q950 reloaded logic board project. There are no electrolytic capacitors on this logic board but I wanted to take a look at the tantalums.
Found something interesting... the stock logic board primarily uses the following tantalum capacitors:
The C26 capacitor that burned out in the thread above is a 10uF 16V capacitor that is on a +12V rail.. the capacitor should have been rated at 25V. We've seen the same issue on IIfx machines that Apple put underrated tantalums on. I should probably replace these tantalums. But given these are underrated on the Q950, I may have to go back and take a look at my Q700 and Q800...
Another thing I need to look at is to check the G2 48MHz clock. I believe this one's used for the SCSI chips and need to check if it's flaky although I wouldn't expect that to manifest as a problem on boots only.
State Upon Receipt
I picked this unit up online and it came in a non-working state. The power supply was dead (various issues on both the primary and secondary sides).
ATX Power Supply
Put together an ATX conversion of the PSU using a Corsair HX1200 which drives the right amount of current for the +5V rail for the Q950. This allowed me to boot and things seemed mostly ok. But I would get a very loud speaker pop on cold boot and I would get some odd crashes (illegal instruction, bus error, bad address, question mark icon even with a known good disk, etc).
Loud Speaker Pop
I heard Q950 owner do experience a speaker pop on boot but I didn't think it was supposed to be this loud. I borrowed a friend's stock working Q950 power supply and the pop was there but much quieter. I put both PSUs on a scope and the only difference I could see is that on the stock PSU, the +5V/+12V/-12V rails ramp up at the same time. The ATX PSU however has +12V coming first about 12-14 ms before the +5V/-12V rails. This seemed problematic. ATX power supplies don't have a requirement to bring up the rails at the same time as they utilize a ready signal to the logic board when the rails have stabilized.
Loud Speaker Pop Resolved
I put together a small circuit that gates +12V against the ramp up of the +5V rail. This got all three rails really close during ramp up. With that change, the loud speaker pop was much more muted, like the stock PSU.
OS Crashes
I was hoping the PSU fix would resolve the crashes but it did not. On boot (and sometimes on restarts), I get a lockup shortly after the "Welcome to Macintosh" screen appears. This means that the System file has loaded and the machine is booting. It's not always the same error. Sometimes I get an illegal instruction error, sometimes a bus error, sometimes a bad address error. Sometimes, it stops finding the disk and shows the question mark icon. In these occasions, it looks like the disk image got corrupted and I have to replace it.
Again I borrowed my friend's PSU thinking it was the PSU but I was able to reproduce with the stock PSU as well.
Setup
Q950. System 7.6.1. 128MB of memory. ZuluSCSI for the disk. Machine does have the stock SCSI terminator at the end of the internal SCSI ribbon cable. And I am not using termination on the internal ZuluSCSI.
Debugging Attempted
- Replaced the memory, didn't help.
- Cleaned the memory socket, didn't help.
- Reflowed the memory socket, didn't help.
- Reflowed the PSU connector, didn't help.
- Reflowed both SCSI chips, didn't help.
- Reflowed the internal primary SCSI connector, didn't help.
- Used my own burned custom ROM, didn't help.
- Tried both an internal and external ZuluSCSI, didn't help. Issue reproduces on either SCSI chain.
- Messed with combinations of SCSI termination, didn't help.
- Tried System 7.5.5 and System 7.6.1, didn't help.
Board looks super clean. No corrosion. Chips are clean. No missing components.
Reproduction
The reproduction is not easy. I get the issue about 1 in every 5-10 cold boots. I get it about 1 in every 50 warm restarts. I have never yet hit the issue when no disk is installed - I have so far always made it to the question mark disk screen when there is no disk. That seems to suggest data and address lines from memory/ROM/cpu are ok.
I'm narrowing on the SCSI subsystem. The fact that it happens on both the internal and external means it's most likely not specific to the SCSI chips (there's two). If I am able to get through boot, the system behaves ok after that, no crashes. Even running a few runs of MacTest Pro and MacBench doesn't introduce a crash. It feels like something on boot up, perhaps power related, is inducing the issue which is why I thought it was the PSU but given I can repro on a known working PSU, maybe I have to look at the caps on the logic board.
What I'm currently looking into...
I printed out @Bolle 's Q950 schematic which he put together to design his Q950 reloaded logic board project. There are no electrolytic capacitors on this logic board but I wanted to take a look at the tantalums.
Found something interesting... the stock logic board primarily uses the following tantalum capacitors:
- 47uF 6.3V - many of them
- 10uF 16V - many of them
- 4.7uF 16V - a couple
- 22uf 16V - a few
- 2.2uF 25V - a few
- 1uF 35V - a few
Dear Members
Recently I noticed my Quadra 950 not responding correctly to power ups, it was somewhat intermittent - the PSU appears to work as expected however. I took it apart and pulled the logic board out and noticed that a tantalum capacitor had burned up, it had some white powdery substance on top and appeared cracked in half. I carefully cleaned the area in and around the cap and while doing so it literally fell apart. The pad came loose either before or during cleanup but continuity is good — for now I applied a drop of solder until someone here can guide me on which capacitor to...
Recently I noticed my Quadra 950 not responding correctly to power ups, it was somewhat intermittent - the PSU appears to work as expected however. I took it apart and pulled the logic board out and noticed that a tantalum capacitor had burned up, it had some white powdery substance on top and appeared cracked in half. I carefully cleaned the area in and around the cap and while doing so it literally fell apart. The pad came loose either before or during cleanup but continuity is good — for now I applied a drop of solder until someone here can guide me on which capacitor to...
- Alex
- Replies: 33
- Forum: Mac II, Quadra & Centris
The C26 capacitor that burned out in the thread above is a 10uF 16V capacitor that is on a +12V rail.. the capacitor should have been rated at 25V. We've seen the same issue on IIfx machines that Apple put underrated tantalums on. I should probably replace these tantalums. But given these are underrated on the Q950, I may have to go back and take a look at my Q700 and Q800...
Another thing I need to look at is to check the G2 48MHz clock. I believe this one's used for the SCSI chips and need to check if it's flaky although I wouldn't expect that to manifest as a problem on boots only.
