SGI Power Supply Tester

SGI hardware problems, solutions, tips, hacks, etc.
Forum rules
Any posts concerning pirated software or offering to buy/sell/trade commercial software are subject to removal.
User avatar
Elf
Donor
Donor
Posts: 66
Joined: Wed Oct 19, 2016 9:54 pm
Location: Pacific Northwest (US)

SGI Power Supply Tester

Unread postby Elf » Wed Apr 05, 2017 10:18 pm

So, I am still working on the large stack of SGIs. I was warned that some of the power supplies might be in an unknown condition (e.g. worthy of starting up in series with a lightbulb), so the first thing I am doing is testing them. My main concern is not damaging any of the computer hardware due to a bad power supply.

Since there are a large quantity of them, the idea is to build a power supply tester to handle:
  • At least Indigo, Indy, and O2 PSUs, in some sort of modular way
  • Testing all possible voltage rails simultaneously under load. Somewhat arbitrarily: 3.3V @ ~11W, 3.3Vsb, 5V @ ~30W, 5Vsb, 12V @ ~20W, -12V @ ~2W. Not full load, but good enough to tell whether it would actually behave while powering a machine
  • Providing visual indication of low voltage, high voltage, or OK status, per rail
  • External outputs for measuring voltage and ripple per rail, with some amount of idiot proofing
  • Repeatably surviving a short-through of the high-side switching voltages or mains to the PSU output
  • To try and protect the PSU from itself, in terms of its AC input current
  • Providing an excuse to start doing electronics again after a decade or so
Ideally I want it to be something nice that I can hold on to, rather than just hacked together.

Anyways I am at the step where I have designed the load boards and have just sent one off to OSH Park. The load boards are per rail, and:
  • Provide input protection (somewhat excessively)
  • Indicate the range of the voltage
  • Connect it through to a resistive load bank

Once the load boards and main board are working in a satisfactory way, I will put them in a nice custom metal enclosure from Protocase.

I will try to post updates here occasionally so that it doesn't look like all those SGIs will just be gathering dust forever! Meanwhile, some of them are also getting cleaned off.

Image
Image

Positive load board: top / bottom. I think it turned out nicely.
Image Image
:Indy: :Indy: :Indy: :Indigo2: :Indigo2IMP: :O2: :O2: :Octane: :Octane: :Fuel: :Tezro:
:Indy: [x19] :Indigo: [x7] :O2: [x4]

User avatar
jan-jaap
Donor
Donor
Posts: 4732
Joined: Thu Jun 17, 2004 11:35 am
Location: Wijchen, The Netherlands
Contact:

Re: SGI Power Supply Tester

Unread postby jan-jaap » Thu Apr 06, 2017 12:57 am

Interesting! So you've got external Rload resistors to dissipate the power, right? If so, the value here compared to, let's say, a load resistance and an oscilloscope is to protect the PSU in case of a mishap? How would this scale to bigger PSUs, which can put 1000VA into a single rail?
:PI: :Indigo: :Indigo: :Indy: :Indy: :Indy: :Indigo2: :Indigo2: :Indigo2IMP: :Octane: :Octane2: :O2: :O2+: Image :Fuel: :Tezro: :4D70G: :Skywriter: :PWRSeries: :Crimson: :ChallengeL: :Onyx: :O200: :Onyx2: :O3x02L:
To accentuate the special identity of the IRIS 4D/70, Silicon Graphics' designers selected a new color palette. The machine's coating blends dark grey, raspberry and beige colors into a pleasing harmony. (IRIS 4D/70 Superworkstation Technical Report)

User avatar
Elf
Donor
Donor
Posts: 66
Joined: Wed Oct 19, 2016 9:54 pm
Location: Pacific Northwest (US)

Re: SGI Power Supply Tester

Unread postby Elf » Thu Apr 06, 2017 7:50 am

jan-jaap wrote:Interesting! So you've got external Rload resistors to dissipate the power, right? If so, the value here compared to, let's say, a load resistance and an oscilloscope is to protect the PSU in case of a mishap? How would this scale to bigger PSUs, which can put 1000VA into a single rail?

Ah, yes. For ones where the load is trivial (<5 W maximum before the crowbar kicks in) it is just a power resistor on the board, but past that, external power resistors connected via the P1/P2 spades. One or more 30W ceramic resistors per rail was what I had picked out.

I was trying to go for a few things with the input protection:
  • Preventing the power resistors from overheating
  • Protecting the power supply, connectors, etc. from excessive current draw if something has gone really wrong (e.g. transformer isolation broke down, some wire fell out and bridged through AC, dead bug dropped somewhere inappropriate, etc.)
  • Protecting the comparator, etc. (for voltage status) against truly high overvoltages
For the rest, I wanted to have more of an "at a glance" status reading across all rails to go through testing more quickly, vs. having to move a resistor and meter or scope around. At most, to turn a knob to get more detailed access to different rails on an external meter/scope.

For bigger power supplies like the 1000VA number you mentioned, I think the design would have to be reworked entirely depending on whether that was voltage or current:

In the case of high current/low voltage (>100A), I'd probably abandon the crowbar/fuse approach and maybe try to shut down the PSU on the AC input side with triacs/SSRs. You could either leave a (huge, heater coil) resistive load bank hooked up to the output all the time or use more expensive parts like IGBTs to attach and detach it from the rails. You might protect the voltage sensing circuit but not the load bank, and protecting the connectors generally wouldn't be as much of an issue with huge bus bars.

For higher voltage/low current (>400V) you could still keep some input protection methods with fuses (ditching the crowbar and relying entirely on the GDT, etc.) but you'd have to be careful about the component ratings, creepage, etc. You'd probably want to optoisolate a lot of stuff before it got closer to the user.
:Indy: :Indy: :Indy: :Indigo2: :Indigo2IMP: :O2: :O2: :Octane: :Octane: :Fuel: :Tezro:
:Indy: [x19] :Indigo: [x7] :O2: [x4]

User avatar
Elf
Donor
Donor
Posts: 66
Joined: Wed Oct 19, 2016 9:54 pm
Location: Pacific Northwest (US)

Re: SGI Power Supply Tester

Unread postby Elf » Sat Apr 15, 2017 3:38 pm

I just received the first set of boards from OSH Park today and they look great!

Unfortunately... @Wolfcry just ran our van into a rock wall today and it is looking like it will cost a lot to repair. So I cancelled my outstanding Digikey orders and it looks like I will have to put this off for a while. :(
:Indy: :Indy: :Indy: :Indigo2: :Indigo2IMP: :O2: :O2: :Octane: :Octane: :Fuel: :Tezro:
:Indy: [x19] :Indigo: [x7] :O2: [x4]

User avatar
Raion-Fox
Donor
Donor
Posts: 1118
Joined: Thu Jan 30, 2014 5:01 pm
Location: near King George, Virginia
Contact:

Re: SGI Power Supply Tester

Unread postby Raion-Fox » Sat Apr 15, 2017 4:45 pm

Ugh tough man. I would be happy to help repair if I wasn't across the country. I've been refurbing my Miata and I've helped with dozens of body damage repairs.
:O3x02L: R16000 700Mhz 8GB RAM kanna
:Octane:
:Indy: (Acclaim) R4400 100MHz 32MB RAM
:Indy: (Challenge S) R4600 133MHz (MIPS III Build Server)
:O2: R12000SC 300MHz 192MB RAM 15k 147GB HDD suzuha
Thinkpad W530 i7 3940XM 3GHz, 32GB, K1000M FreeBSD 11/Windows 7 rin
Acer Aspire 5157 (headless) 4GB, FreeBSD 11 kuran
Thinkpad R40 Pentium M 1.7GHz 2GB RAM kasha

User avatar
Elf
Donor
Donor
Posts: 66
Joined: Wed Oct 19, 2016 9:54 pm
Location: Pacific Northwest (US)

Re: SGI Power Supply Tester

Unread postby Elf » Sat Apr 15, 2017 8:04 pm

Thanks :)
:Indy: :Indy: :Indy: :Indigo2: :Indigo2IMP: :O2: :O2: :Octane: :Octane: :Fuel: :Tezro:
:Indy: [x19] :Indigo: [x7] :O2: [x4]


Return to “SGI: Hardware”

Who is online

Users browsing this forum: No registered users and 2 guests