[kaigan]

Hey, folks,

As a part of my continued repair work on my Onyx2 deskside, I just picked up a pair of node boards from Raion. (Thanks again! ) However, after installing them, I was still getting general exception errors from one of the boards, and I can't get it to boot fully with a single board installed, either.

After disassembling everything again, I found that the midplane is actually missing two SMD components that look like either capacitors or perhaps a resistor pack. It wouldn't surprise me if this was the cause of my current woes. The two that are missing are labeled as R1F0B and N4E4B on the board. Both are located in clusters with others, on the side of the board with the power connectors. The areas are above the two white power connectors, close to the large group of big electrolytic capacitors.

I'm hoping that someone might be able to identify the ratings of these parts from their own midplane, so that I can obtain replacements. There are enough different parts around the board that I can't really guess exactly what they should be and just winging it seems like a bad idea.

I've attached two pictures showing the missing components and the ones surrounding them.

Thank you!

[jan-jaap]

I'm hoping that someone might be able to identify the ratings of these parts from their own midplane, so that I can obtain replacements. There are enough different parts around the board that I can't really guess exactly what they should be and just winging it seems like a bad idea.

I have a spare backplane so I can have a look. PM me if I don't report back in a couple of days.

[kaigan]

I have a spare backplane so I can have a look. PM me if I don't report back in a couple of days.

Thank you so much! I really, really appreciate it!

[jan-jaap]

OK, I had a quick look and on my spare backplane these two positions are not populated.

So it looks like you didn't break anything; it's simply supposed to be like that.

[weblacky]

Hi Jan-japp,
I can't speak directly to these boards. But trying to think up avenues that you might investigate to find a solution, can you tell us more about these "general exception errors"? Do they give any codes or anything?

I'm sort of wondering if this is a functional issue or a failing board power issue from the ONYX system's viewpoint.

Perhaps a fair way to start further investigation is something as simple as comparative analysis? That is, find the grounding plane of two identical boards (one that someone claims works) and gently probing using either a multimeter in diode mode or Ohm mode and check resistance or voltage drop values between board to system interfaces paths first (solder joint for board connectors/ports to board ground) when the boards are outside the system. Draw up a comparative list and see if there are substantive differences? That would give paths to investigate (local board regions that may be failing that directly affect interface ports/connectors) if you see big differences on the interfaces first.

My thinking is perhaps you do have a coupling capacitor failure or the like and to inspect the board-to-system interface side first, if that proves fruitless, then doing a chip by chip conductance measurement to ground, then after find grounding pins on each chip, do a pin to grounding pin diode mode check on the same chip, checking for shorts. Tracking carefully, you may find a coupling capacitor in a logic path (chip to chip) that's shorting and affecting functional operation, but is not affecting board general power. If the chips have data sheets, download them to find pinouts to help finding where power goes in and out.

It would also maybe be a good start to find the main power rail(s) for the board (assuming a single voltage...perhaps there several inputed by the backplane) and check them each to ground for shorts. This is actually the easiest place to start, if you end up knowing that the input to board is DC only (I would assume it is, but YMMV) you can use like 1v-2v with like 1.5 - 2A current limiting power supply to try to see if there is an overdrawing component (failed) on the main power rail(s) for the board. If there is little to no immediate amperage draw...then there is no main rail short and the short may be on the other side of any chips or buck/boost circuits on the board.

If we assume it's not a software thing, or a system connector issue, then my guess would be that it's a failure of some small part that would drag voltage to ground. I'd like to think it's not a chip issue or EPROM damage or something like that...seems improbable, it's much more plausible that it's a voltage drop or power issue given the age and board condition. A voltage drop due to short on the inner logic areas of the board would be seen by the system as a functional error, but have no complaints about the main system power supply because the damaged area is likely current limited, so it can't run away and just stays hot.

If that fails, doing a massive check of all resistors would be next (again resistors in circuit can measure different from marking due to layout and age) but if you compare them with another board you should see something very off, not a little off. So again comparative analysis of all resistors would be last on my list.

Maybe you'll get lucky with one of the first big tests.

Let us know if you decide to pursue any of these ideas and what results you found.

[jan-jaap]

Hi Jan-japp,
I can't speak directly to these boards. But trying to think up avenues that you might investigate to find a solution, can you tell us more about these "general exception errors"? Do they give any codes or anything?

Hmmm, my Onyx2 is fine. The OP is kaigan.

To investigate problems on the O2K/Onyx2 series, start with the https://archive.org/details/Origin2k_Hardware_Guide
Put the MSC jumpers in 'heavy diagnostics' mode, hook up a serial console to the MSC and see where it goes.

If it throws memory related exceptions: take out the memory, clean with pure alcohol and re-insert firmly. Reduce to a single bank of known-good RAM if necessary
If it throws errors related to an ASIC on the nodeboard you'll need to replace it.
Same for problems related to the PIMM (CPU/cache module).

Once you get it to pass with one nodeboard (in slot N0) you can expand the configuration.

[kaigan]

Put the MSC jumpers in 'heavy diagnostics' mode, hook up a serial console to the MSC and see where it goes.

I've put the system into heavy diagnostics modes before. It will usually make it to DEX mode and then stop responding via console shortly thereafter. It will still accept a CTRL-T and MSC commands, but nothing else.

I did find one more spot that might be missing a component. I try to be meticulous about this sort of thing, but I totally missed it. This one is labeled "M8N2B". It's on the side of the board with the power connectors as well. It's above the large heat sink, just below the start of the logo on the board. Again, I've attached a picture.

If we find that the new area is also okay (the pads do look a little rougher on this spot, though), then I can grab more detailed error messages from diagnostics, as well as running through the procedures Weblacky suggested, too.

Thank you both for your help!

[kaigan]

I did find one more spot that might be missing a component.

Okay, make that two spots. This one is above and left of the other, by the first of the drive connectors on the board. It's labeled "N5R1B". Again, a picture is attached.

Thanks!

[Irinikus]

You would be able to see the points where the component sheered off. I don't see any sheer points here.

This is the damage you would see if a component such as a surface mount cap or resistor was missing:

[jan-jaap]

Okay, make that two spots. This one is above and left of the other, by the first of the drive connectors on the board. It's labeled "N5R1B". Again, a picture is attached.

Not populated.

There are quite a few empty pads if you look for it. On the backside, roughly where the nodeboards mate (K/D quadrant) there are at least three ICs and a crystal not mounted.

I'd like to see the debug output you get when you start the system with the MSC jumpers in 'heavy diagnostics' position.

FWIW: I've seen several nodeboards and GE boards go, but never a backplane.