So I have a friend's 8600 in which I recapped the psu on. It was working fine as I was testing it until it locked up and crashed out of nowhere. At first it turned on with no video but then it acted like it was shorting out. So I pulled everything out of the motherboard including connectors, and probed for shorts and I found a 28ohm short on the 3.3v side of the power connector. Is the 3.3v side of the motherboard supposed to be this low resistance or did I find something actually shorted?
Power Mac 8600 shorted 3.3v on the motherboard?
- Thread starter Kothnaaken
- Start date
Bit of an update, I pulled the board out and tested all the non tantalum caps and c369 and c372 are shorted, c369 just so happens to be a 28 ohm short as i was measuring it. C452 seems a bit marginal as well. I'm going to say it's safe to assume at least the first two would be the issue here? SMD soldering is not exactly my favorite thing to do so if anyone has one of these boards and can measure across these three caps on the back that'd be great.
It could also be a short elsewhere that's causing those caps to appear shorted. Also, you should test the tantalums here. It's very rare, but they can short out.
I tested I believe every single cap on that board including the tants. I'm going to be pulling those three today and checking since I ordered replacement tantalums. I'm hoping since c369 is exactly the resistance that I'm measuring from the power connector on the board that it's the main issue, otherwise I'm probably going to have to go schematic hunting.
I unfortunately don't think there is an 8600 schematic around, although there have been some new ones surfacing recently, so who knows.
That's especially unfortunate as it turns out none of the caps were causing a short and I'm a bit at a loss as to what it can be. Unfortunately (though also thankfully) since the PSU has OCP I can't run it and see what gets hot to try and find a short. Both C369 and C372 have the same resistance shorted across both pads though they're on opposite sides of the board which makes it all the more confusing. In fact it actually appears they're somehow connected and shorted together.
This board may very well be beyond hope without a schematic. I pulled up a datasheet for the 74f573 chips on the board (which I figured would be the easiest chips to find datasheets on and probe ground from) and sure enough there's still a direct short from the 3.3v pins on the power connector to ground on it. No telling what could have failed short with how much there is on it.
There's one additional thing you could try, which is voltage injection and the "owee" check.
Step 1: Use a current-limited supply and inject about an amp of current
Step 2: Touch stuff until you find something really hot (hence owee)
Alternately, do the current injection and use a thermal camera to see what's going on. I have a thermal camera for this sort of thing, to avoid the "burning yourself" part.
Step 1: Use a current-limited supply and inject about an amp of current
Step 2: Touch stuff until you find something really hot (hence owee)
Alternately, do the current injection and use a thermal camera to see what's going on. I have a thermal camera for this sort of thing, to avoid the "burning yourself" part.
That's great and all until you touch one of the LC III socketed DIP chips you reseated incorrectly and give yourself a nasty peeling burn!
Don't ask me how I know.
Seeing as this board isn't mine (and I've already had to sink a lot of time into this 8600) it sounds like I'm going to simply be swapping in a known working board and ending up with a 9600 to replace my 8600 I'll be parting out. Seeing as I don't yet own a benchtop power supply I can't do a ton more until after I move in a couple months and perhaps then I can pick one up and revisit it if he doesn't want the dead board back.
There was a topic similar to this recently either here of on 68Kmla. Turned out the resistance between power and Ground was normal for that board. You must account for leakage in all the chips that connect between 3.3V and Gnd. 3.3V/28ohm = 100 mA, which does not sound out of line.
What you really need is for someone with a good working 8600 to measure the static resistance between 3.3V and Gnd so you'll know if you're chasing the correct squirrel.
In that other thread someone did the test on a good board and the resistance was about what the original poster was seeing.
You might also check resistance between 3.3V and 5V and 12V if you have not already. I once had a board with a dead short between 3.3V and 5V. It was in one of the main chips.
Is this an original 8600 or a "Enhanced" 8600, AKA Kansas or Mach V? If it's an original 8600 you could try a 150MHz of slower PPC604 CPU card and see if it will boot. This is old memory but as I recall, the older PPC604 chips run on 5V and the newer on 3.3V. Most of the motherboard runs on 5V, but some of the PCI circuitry needs or may need 3.3V. So with an older CPU card, one can often boot one of these machines even if there's a problem with 3.3V.
What you really need is for someone with a good working 8600 to measure the static resistance between 3.3V and Gnd so you'll know if you're chasing the correct squirrel.
In that other thread someone did the test on a good board and the resistance was about what the original poster was seeing.
You might also check resistance between 3.3V and 5V and 12V if you have not already. I once had a board with a dead short between 3.3V and 5V. It was in one of the main chips.
Is this an original 8600 or a "Enhanced" 8600, AKA Kansas or Mach V? If it's an original 8600 you could try a 150MHz of slower PPC604 CPU card and see if it will boot. This is old memory but as I recall, the older PPC604 chips run on 5V and the newer on 3.3V. Most of the motherboard runs on 5V, but some of the PCI circuitry needs or may need 3.3V. So with an older CPU card, one can often boot one of these machines even if there's a problem with 3.3V.
No shorts between rails. Too many parts seem to be shorted to the 3.3v rail on their ground pin to make me believe its just leakage anyways. How do I tell the difference between the two boards?
Look at the four ROM chips on the back. They're 44 pin chips about 1.1" X .5". The original 8600 will have chips labeled 343S0280 - 343S0283. (343 might be 341.) The Enhanced 8600 has chips labeled 343S0380 - 343S0383.
Also, I believe the Enhanced lacks a cache slot. There should just be an unpopulated set of pads or holes there.
You're going to read the same resistance everywhere you measure between GND and 3.3V. The 3.3V plane is a near zero resistance connection between all the 3.3V supply points. The same for the GND Plane. So if you're probing a connection to those planes, it doesn't matter which point you're using as the probe point.
Also, I believe the Enhanced lacks a cache slot. There should just be an unpopulated set of pads or holes there.
You're going to read the same resistance everywhere you measure between GND and 3.3V. The 3.3V plane is a near zero resistance connection between all the 3.3V supply points. The same for the GND Plane. So if you're probing a connection to those planes, it doesn't matter which point you're using as the probe point.
Neither the faulty board or the one I have which is going in is the enhanced one. In fact, mine had a cache module on it. However after swapping boards it seems to be fine now. Depending on if I end up hanging on to the faulty one or not I might try a few theories I have about it, though it's not my say if I do or not yet.