Indy Doesn't See Its Graphics Card

[Silicon-Surfer]

I'm hoping someone might have a clue as to what the problem is with this Indy. I picked up the machine about 5 years ago with a failed Nidec PSU. Recently I decided to bite the bullet and fix the power supply, and after much work and replacement of failed components, it's now working.

The problem is that I can't get any graphics output from the 24bit Newport graphics card. When I start up with a console connection, the message "Cannot Open Video(0) For Output" is displayed, and "hinv" doesn't list any graphics card present. I've cleaned all the contacts and reseated everything, carefully looked at the motherboard and graphics card top and bottom for any damage, but everything looks fine. When powered up there are voltages present on the 13w3 connector and the heatsink on the graphics processor is warm, so it seems to be operating, but the system can't see it.

I guess best way to troubleshoot would be to try a different graphics card, but I'm in Australia and they are not exactly easy to find. Plus the price being asked even for 8 bit cards is a bit crazy. Does anyone have any ideas on what I could try to diagnose/fix the problem?

 

[Elf]

That's a tough one unfortunately

If it isn't seeing the card and the connectors are clean, then I expect something has gone wrong at the hardware level. A fast logic analyzer might give some more indications if there is any communication going on, but still I have to expect that even with international shipping being involved, a replacement might be easier. Or you could use it as a Challenge S

 

[Silicon-Surfer]

Thanks for the reply Elf!

My reservation with purchasing another graphics card is that the fault might be on the motherboard. It's crossed my mind that when the PSU failed, it could have fried something on the motherboard that's caused the problem, so I was hoping to figure out a way to narrow it down without having to just throw parts at it. But maybe that's not possible...

 

[weblacky]

Unfortunately we don't have a community of measured "good" values for comparison at this time (hopefully in the future we will) for SGI parts/pins/ports. If you'd like to try to narrow it down (without buying either a duplicate mainboard or a duplicate graphics card first).

I'd use a multimeter and do a basic voltage measurement of the graphics card while it's in the Indy (with top off) while the Indy is powered up. I'd try to find:

Is there any power at all going to the card?

If so, what voltage and which interface pins does it look like it's coming through on the mainboard connector.

Does the voltage "stop" after probing sections of the card? Some parts have voltage, other parts have just nothing or next to nothing.

Is the voltage extremely low?

Then I'd power off the Indy and separate the boards, then do a diode test and an Ohm test on EACH pin of the card (internal daughterboard) interface and record the values in a spreadsheet. Then you can compare those values with someone else or a duplicate part to see if they are similar or dissimilar, then you have something to go on.

I mean test both the Mainboard's interface ports and the graphics card's interface ports. You'll need both for comparison of whatever part you decide to get values from a duplicate.

I have a Indy right now that was powered up (against my advice) with an original PSU and it started but a cap on the mainboard started to smoke. The owner then shut it down (it was still running I was told) by pulling power and I bought the system for cheap from them due to add-ons in it being great value. I will be attempting to restore it once my additional equipment arrives for PSU testing/rebuilding.

As PSUs die (due to age I mean) they often have incredibly high ripple voltage and that tends to take out decoupling caps and possibly diodes (especially aged caps and diodes). That may be what you have, a part was shorted and now you have a current-limited grounded out voltage source on your graphics card. Or something on the graphics card went (I understand the deliemia).

If you don't have any obvious burnt parts, then I'd try to trace power and then when it's powered off using diode & Ohm checks on the interface pins with both boards separated.

Whichever part you DO decide to buy (gfx or MB), perform a diode test on the SAME interface pins of whatever you buy and check/compare with the "bad" part. If you find a good discrepancy, that's likely a place to look!

 

[Silicon-Surfer]

Hi weblacky, thanks for your reply.

I have done a bit of probing on the graphics card and there are definitely voltages present. The heatsink on the graphics processor is noticeably warm when the machine is powered up, so there's power on the board for sure and it seems like it's at least partially functional. I'm just not sure what voltages should be where. I've tested voltage on some of the caps on the graphics board and they seem to be sane (5v, 3.3v etc).

Your point about caps and diodes possibly being damaged is a good one, I might try to track the power to the graphics board and also on the board and test components that could have been damaged by the bad PSU.

 

[weblacky]

Try this for pinout info (starting on page 2-17): https://honk.sigxcpu.org/doc/Indy/gio64.pdf

 

[Silicon-Surfer]

Very helpful, thanks!