Quadra 950 Not Booting

[tt]

The machine's power supply turns on, powers the drives, but no boot chime sounds and I think there is no video. This Quadra 950 has worked in the past, but it had a logic board that was not fully seated in its plastic frame when I received it and I guess it shifted again when I moved the machine. I noticed the reset and interrupt switches were being pushed into the end of buttons. I am wondering if they got strained from being pushed on for a long time.

Can someone please measure the resistance of their S1 switch (marked on the board / triangle button)? I measured it across the two top exposed conductors, so it is not necessary to remove the board to take a measurement. I measured S1 and the resistance was on the order of 130 ohms. I feel like that value is too low, but I measured S1 on two different SE/30 logic boards and it is around 350 ohms (so some lowish resistance might be normal?). S2 is fully open when not pressed.

Other checks I did: Measure the voltages coming off the PSU which look good. Swapped out the RAM. Disconnected the hard drives. Smell test for burned components.

 

[David Cook]

Powered down with power plug unplugged but pram battery installed.

Triangle button: 145 ohm, down to about 2 ohm when pressed. No voltage.
Circled seagull button: 22 megohm, also down to about 2 ohm when pressed. However, 3.5V is present. I assume 22 megohm reading is false due to voltage applied by circuit.

I had to exercise the buttons a little bit before they would fully drop to 2 ohm. The contacts must be a little oxidized.

 

[tt]

@David Cook thanks for checking. Those readings are similar to mine so I guess the buttons are not the issue.

Does the Q950 need a PRAM battery installed to function properly? There isn't one installed, but the power supply seems to function ok without the battery installed.

Maybe the -12V drops under load? I haven't tested that voltage under load. I don't know, not sure what else to check at this point.

 

[joshc]

Does the Q950 need a PRAM battery installed to function properly?

No it doesn't need one, it can work fine without it.

So you're sure the reset/interrupt buttons are not being pressed in by the case if the logicboard is too far forward? That's what last caused this issue when I had a Q950.

 

[tt]

So you're sure the reset/interrupt buttons are not being pressed in by the case if the logicboard is too far forward? That's what last caused this issue when I had a Q950.

Yeah, there is a slight gap between the end of the case button mechanisms and the button faces on the logic board. There is some travel, ~ 0.5 mm before it contacts the switch.

One thing I thought was odd was that it was really tough to slide the logic board toward the back of the case to get the middle latching tab that prevents the board from sliding backward to clear. I don't know if that indicates anything. I have thought about trying to remove the board, but I don't want to risk damaging some part if it's not necessary.

When I first opened it up after I had put it away, I was surprised to see the speaker was disconnected. I doubt I had done that before boxing it up and thought maybe the loose SCSI cable near it somehow tugged it loose when it was moved. I guess it's possible the SCSI cable knocked some component near there. I looked around there but so far haven't noticed anything unusual. The board is fairly clean so it is easy to inspect. I tried connecting another speaker in case the original speaker happens to be dead, but no sound came out.

 

[joshc]

Probably something you've already checked but what position is the key in?

 

[tt]

I am not as familiar with this machine, so it is a good question. To turn it on, I turn the key to the 12 o clock position and then press the power button on the keyboard. The machine turns on, PSU fan runs, the green light from the logic board is on, and the hard drives spin up. No boot chime.

If I turn the key all the way to the 2 o clock position, it will just turn on right away without touching the keyboard.

 

[joshc]

Right, forgot you mentioned that it is powering on, just no boot chime.

Try reseating all the RAM. Also spray contact cleaner into the RAM slots.

Does the 040 get warm?

 

[tt]

I put a new set of 4 16MB SIMMs in since I happened to have those handy. There were 2 sets of 4 sticks in there before when the machine was working. I took a photo to put them back in the same way, but the angle I took them at they all look the same. So I put just one set of 4 matching sticks back. I am guessing they are all low capacity RAM, they have two chips per stick. I don't have contact cleaner at the moment.

The 040 seems to get warm if I place my palm over the heatsink; I feel a little bit of warmth coming off from it.

 

[David Cook]

I just upgraded my 950 to 256 MB of RAM. These are new SIMMs (or at least newly assembled) from eBay.

After the upgrade, the Quadra showed the same symptoms as yours. It powered on (fan, drives, etc), but no chime. I isolated it to a faulty SIMM. It is really hard to see even with a magnifying glass, but two of the chips are too near together, shorting two of the pins. Compare the gap between those chips to the gap of the pair of chips to the right.


A profile view makes it obvious, as light passes through the middle of a good pair of chips, but light doesn't pass through the middle of the pins of the bad (too close) pair.



Rather than generating the Sad Mac tones of death, this fault seems to even prevent the ROM from being readable or something like that.

I'm not suggesting your memory is bad (although you should try booting with no memory to see what happens). However, since the symptoms are identical it might be a clue for you. I assume you've checked the CPU reset line and clock signals are good.

- David

 

[Phipli]

I just upgraded my 950 to 256 MB of RAM. These are new SIMMs (or at least newly assembled) from eBay.

After the upgrade, the Quadra showed the same symptoms as yours. It powered on (fan, drives, etc), but no chime. I isolated it to a faulty SIMM. It is really hard to see even with a magnifying glass, but two of the chips are too near together, shorting two of the pins. Compare the gap between those chips to the gap of the pair of chips to the right.
View attachment 62764

A profile view makes it obvious, as light passes through the middle of a good pair of chips, but light doesn't pass through the middle of the pins of the bad (too close) pair.

View attachment 62765

Rather than generating the Sad Mac tones of death, this fault seems to even prevent the ROM from being readable or something like that.

I'm not suggesting your memory is bad (although you should try booting with no memory to see what happens). However, since the symptoms are identical it might be a clue for you. I assume you've checked the CPU reset line and clock signals are good.

- David

I've had this but thankfully noticed before installing them. I adjusted them a little with a knife.

 

[David Cook]

I've had this but thankfully noticed before installing them. I adjusted them a little with a knife.

Good tip. I will try that before going through the hassle of a return.

Lucky for me, these chips are making full contact, rather than being microns apart. For anyone having intermittent issues that go away / occur when the computer heats up, check the spacing of your chips.

 

[joshc]

David beat me to it but I was going to suggest this does seem RAM related. Some Macs are weird with how they treat these sorts of scenarios. The IIfx, upon which the Q900/950 is somewhat based, will give a good boot chime even with bad RAM!

 

[tt]

I'm not suggesting your memory is bad (although you should try booting with no memory to see what happens). However, since the symptoms are identical it might be a clue for you.

Thanks @David Cook, @joshc, and @Phipli. I will take a closer look at the RAM. I was assuming that since the machine booted with the old RAM before that it should still be good. I will pull out the old RAM out of Bank A and put in the other set of 4 SIMMs in Bank A. If it still doesn't work, I'll get some RAM off an SE/30 board that boots and try that.

I assume you've checked the CPU reset line and clock signals are good.

I don't think so, how are theses checks performed?

 

[dougg3]

Rather than generating the Sad Mac tones of death, this fault seems to even prevent the ROM from being readable or something like that.

Ahh, this makes total sense. On a lot of machines (likely this one too) the ROM and RAM are attached to the same data bus. If any of the data pins are shorted to something else, it's going to interfere with the ROM reads too...

 

[tt]

I swapped in and out the memory groups and ended up "testing" the original RAM in my SE/30 with memtest bypassed using an OG @dougg3 8MB ROM SIMM. I also tried booting with no RAM installed, but there was no sound emitted in any case. I put back the 8 x 1 MB SIMMs that were initially in the machine when I received it.

I'm wondering if the ROM could be bad? Could a new ROM SIMM be installed on a Q950 to bypass the installed ROM?

I took out the board to take some photos of it in case there is something that can be seen that is wrong with it.

 

[dougg3]

Technically my old 8 MB ROM SIMMs don't automatically disable the onboard ROMs in Quadras properly, but people have reported success using them in the 700/900/950 -- so they are likely able to overpower the onboard ROMs. Since you already have an original ROM SIMM and programmer, there's basically no downside to at least attempting it as a quick test to see if the ROMs are the problem. You could flash a 950 ROM image onto the SIMM and see if it changes things at all...

 

[David Cook]

Wow. Okay, you've taken the board out. That will make testing much easier, assuming you can still connect the power supply. I think you'll need to jumper the keyswitch connector to allow it to power up.

Check pin 12 of U23 (ROM) to make sure it gets driven low (~output enable).



This is the pinout of the 68040 viewed from the bottom of the board.



Check pin S7 (~RSTI) of the 68040 to make sure this goes high. It will start low to hold the system in reset. When power stabilizes (milliseconds after startup), reset should go high to let everything start running from a known state.

If you have a scope or a meter capable of measuring high frequency, then check pin R7 and R9 of the 68040 to make sure the clock signals are going. If you don't have a scope/fast meter, measure those pins in voltage mode to make sure it is not almost 0V or almost 5V. Basically, a value in between suggest a changing signal.

Lastly, you could check the values or different parts of the bus address lines/data lines, R/W, etc with the board back in the machine from the pins in the PDS slot. See the attached Quadra 900 developer note page 32 for pin descriptions.

Measure the ohm of the speaker to make sure it is not open or shorted. I assume you've given the computer plenty of time to boot or show video, and haven't given up too quickly due to a lack of chime.

- David

 

[David Cook]

I adjusted them a little with a knife.

I used a razor blade as wedge to bend the pins away from each other. Worked great. It turns out two of the SIMMs had this flaw. Everything is working now.

Aside: I think MacTest Pro may leave the processor caches disabled after a memory test. I got horrible results from Norton System Info until I rebooted. Could be coincidence.

 

[tt]

You could flash a 950 ROM image onto the SIMM and see if it changes things at all...

Thanks for the guidance. I tried this with a 2MB ROM with soldered on flash chips (maybe one of your earlier builds) but unfortunately, no dice. The red pirate skull LED powers on at least.