Another Indigo2 psu died...

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
Interesting; I guess it behaves a little differently with capacitative loads versus just resistive loads. But, at least it confirms (without involving the motherboard) that the supply is indeed not working, which was the only real purpose of that test anyways. :) Unfortunately, the test is not further diagnostic in any way, it just removes the motherboard as a variable.

Hopefully for consistency the known good supply works with your test rig, just to confirm that it's all set up properly?
 

marmotta

Member
Feb 22, 2021
62
6
8
Interesting; I guess it behaves a little differently with capacitative loads versus just resistive loads. But, at least it confirms (without involving the motherboard) that the supply is indeed not working, which was the only real purpose of that test anyways. :) Unfortunately, the test is not further diagnostic in any way, it just removes the motherboard as a variable.

Hopefully for consistency the known good supply works with your test rig, just to confirm that it's all set up properly?
I have not done this test, today I try ;-)
 

marmotta

Member
Feb 22, 2021
62
6
8
Update! If I put a 2,2 ohm resistor to 3,5v graphic rail (blue cable) it continue to no start. If I put a 1 ohm resistor it start normally with good voltages.

I’am still talking about the good power supply ;-)

Edit: Work also with a higher resistor (tried 2,2 and 6,8) but the resistor on graphic rail is mandatory.
 
Last edited:

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
Odd! Usually these will start if that graphics power rail is terminated even without a load as they can be used without Impact graphics.
 

weblacky

Active member
Jan 13, 2020
180
45
28
Seattle, WA
It might make sense if the voltages were too low. If you assume the terminator card is either a dead short or 1 ohm or less resistance then there could in fact be no voltage to register during termination on a faulty PSU.

Under normal operation yes, but if his voltages are too low or weak, it’s possible the termination would look like it’s approaching zero instead of a nominal voltage.

In other words, by limiting the current taken at termination he preserved a minimum voltage that registered to the internal logic of the PSU. Still speaks to out of spec voltage internally + PSU protection.
 

marmotta

Member
Feb 22, 2021
62
6
8
Odd! Usually these will start if that graphics power rail is terminated even without a load as they can be used without Impact graphics.
Yes is right... I think is the first sign of old age. I have others 8001 psu but I remember it works without graphic. For sure teal Indigo2 work without graphic board.

Now I want to repair the 8002, with one way or another :-D
 

marmotta

Member
Feb 22, 2021
62
6
8
It might make sense if the voltages were too low. If you assume the terminator card is either a dead short or 1 ohm or less resistance then there could in fact be no voltage to register during termination on a faulty PSU.

Under normal operation yes, but if his voltages are too low or weak, it’s possible the termination would look like it’s approaching zero instead of a nominal voltage.

In other words, by limiting the current taken at termination he preserved a minimum voltage that registered to the internal logic of the PSU. Still speaks to out of spec voltage internally + PSU protection.
When the 8001 psu is up the voltages is perfect, when is inside the Indigo2 work without a problems, no shutdown, no errors... and the Indigo2 have the max impact!

Is only a probable startup issue, it can help for understand the complicated logic of this psu.
 

marmotta

Member
Feb 22, 2021
62
6
8
If I not found another solution, I go for the b-plan! Swapping the LV/HV section of the two psu for find the issue without doubt :p

Is the occasion for see if 8001 is the same of 8002 (for me are identical)

C-plan is atx conversion :-D
 
Last edited:

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
It might make sense if the voltages were too low. If you assume the terminator card is either a dead short or 1 ohm or less resistance then there could in fact be no voltage to register during termination on a faulty PSU.
I covered it a bit more in the Indigo 2 PSU page, but all the terminator does is connect the relevant sense lines to the graphics power rail. It actually does not provide any load.
 

weblacky

Active member
Jan 13, 2020
180
45
28
Seattle, WA
I covered it a bit more in the Indigo 2 PSU page, but all the terminator does is connect the relevant sense lines to the graphics power rail. It actually does not provide any load.
So you measured the actual draw on graphics power rail when attaches to the terminator card, and saw “no draw”?

Because your statement just sounded like you agreed, the terminator has no limiter itself. Simply dead shorts graphics power rail to sense. So sense has some sink limit itself. That sink limit would equal the draw from the graphics rail at terminator card.
Sounded like this statement just confirmed the terminator is a dead short (confirmed) to sense (which governs the load of this shorted connection). So the user shifted the load control from sense to terminator by further restricting load with another resistor.

Am I missing an important point in the statement?
 

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
I did not measure current on the sense lines, but would expect no substantial current draw. Usually sense lines are high impedance (and further, the wires are pretty small gauge). Certainly there is negligible resistance between the rail and the sense lines, but I'm not sure if "dead short" is the best way of describing it in this context; it would suggest that the rail itself was shorted, or alternatively had a very low impedance load across it. Rather (although again I haven't measured it) the opposite is true, when connected to the terminator, the graphics power rail basically has no load.
 

weblacky

Active member
Jan 13, 2020
180
45
28
Seattle, WA
Perhaps this situation means there is something wrong with his sense line? Hence won’t turn on in an Indigo2? If sense line circuit impedance has changed then this further limiting brought it back into tolerance?
 

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
Could be, I forget what exactly happens without the sense lines although I did try it before. I think it goes out of regulation a bit, but I don't really remember.
 

marmotta

Member
Feb 22, 2021
62
6
8
B-plan started! At bottom the 8002, on top 8001

Practically the boards are the same, the 8001 have more last time fixup and bridges/wires. 8002 are more clean. Change only the revision of the boards. ;-)
 

Attachments

marmotta

Member
Feb 22, 2021
62
6
8
I don’t like this plan! You’re risking too much. Use the clues you have to check more board items.
I not have the schematics and diagnostic tools for use another way... this is the only way for exclude HV or LV and circumscribe the problem.
 

marmotta

Member
Feb 22, 2021
62
6
8
First fast test:

HV 8001 + LV 8002 = try to start, then stop after a second

HV 8002 + LV 8001 = start! But it need the 3,5v grail resistor

Conclusion:

The two HV boards are ok
The 8002 LV board don’t work
The 8001 LV board work good but it need a load on 3,5 grail.
 
Last edited:

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
Whew, glad it worked out alright; any differences between the revisions could have created some spectacular fireworks. But I guess as it is, you did discover some things!
 

marmotta

Member
Feb 22, 2021
62
6
8
Whew, glad it worked out alright; any differences between the revisions could have created some spectacular fireworks. But I guess as it is, you did discover some things!
Yes! No fireworks :)

8001 and 8002 are the same. Same spec, same pcb and components... the 8001 have some last minute fixes, the 8002 not.

The only difference I see is one capacitor at low amp 3,5v: on 8001 is 1000uf 10v, on 8002 is 2200uf 10v

Now I at work for replace all capacitor on 8001 LV, but now one by one and testing it after any replacement ;-)
 
  • Like
Reactions: Elf

About us

  • Silicon Graphics User Group (SGUG) is a community for users, developers, and admirers of Silicon Graphics (SGI) products. We aim to be a friendly hobbyist community for discussing all aspects of SGIs, including use, software development, the IRIX Operating System, and troubleshooting, as well as facilitating hardware exchange.

User Menu