O2 Power Supply Basics

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
It's time for more Power Supply Basics! This time of the Silicon Graphics / SGI O2. This one had less to figure out than the Indigo 2 and Indy.

If referencing or duplicating this information, I ask that you maintain a prominent link to this original forum post for further updates, as well as a credit by name.

At least of the time of this writing, you won't get any better power supply information than at the Silicon Graphics User Group!

Here is the power supply pinout from the O2 front plane perspective, looking from the back into the recessed cubby hole for the PSU:
O2 power inlet.png
Power is supplied over a 24 pin Molex Mini-Fit BMI connector. This is like the Mini-Fit Jr. commonly found on wire to board solutions, but has guide posts and no clip. A 24 pin Mini-Fit BMI "plug" (machine side) is pictured to the right, with the Mini-Fit Jr. to the left for pin number reference purposes.

The O2 front plane (which connects the motherboard, power supply, and other accessories) has the male pinned connector, while the O2 power supply has the female connector.

+3.3V standby (Aux) power is supplied when the PSU is connected to AC Line power, but has not yet been turned on. In contrast to the Indy and Indigo 2 PSUs, the O2 PSU does not turn itself on by default, and the Run line must be grounded to start.

Pin #Wire Color / SizeFunction
1Red / small+3.43V rail feedback / sense
2Blue / smallRun (Ground to start)
3Red / large+5.1V
4Red / large+5.1V
5Black / large0V
6Black / large0V
7Black / large0V
8Black / large0V
9Black / large0V
10Yellow / large+3.43V
11Yellow / large+3.43V
12Yellow / large+3.43V
13Grey / small0V return feedback / sense (?)
14Brown / smallPower good signal (3.3V)
15Yellow / small+3.3V Standby
16Purple / small+12V
17Orange / small-12V
18Black / large0V
19Black / large0V
20Black / large0V
21Black / large0V
22Yellow / large+3.43V
23Yellow / large+3.43V
24Yellow / large+3.43V
Pin numbers are as per the mechanical drawing. Wire colors are internal to the power supply, as only the connector is exposed when fully assembled.

The unit under test is Sony model APS-90 / part # 8-68-1189-41, SGI part # 060-0022-001, with the following specifications:
RailRating
+3.43V31A
+5.1V10A
+12V3.2A
-12V0.25A
+3.3V Standby (Aux)10mA

Unlike the Indy and Indigo 2 which have most of their capacity on the +5V rail, the O2 is primarily +3.3V driven, with more of its power on the oddly spec'd +3.43V rail. The 3.43V regulation was confirmed. Standby power comes from a separate converter, and is at 3.3V, versus the main power rail.

Power factors were in the 0.6 area; bad as usual! Standby power is about 4 Watts. No precise numbers this time around, although they could be reported if desired.

O2 test wide.jpg O2 test normal.jpg

As you can see the O2 PSU underwent the usual test setup. Initially what gave me some trouble was the nature of the two sense lines. Without Pin 1 (red / small) and Pin 13 (grey / small) connected to the +3.43V rail and 0V return respectively, the power supply goes wildly out of regulation. The +3.43V rail will swing above 5V, and the power supply has a tendency to shut down under any load.

This made it difficult to test until I uncovered the nature of the sense lines by testing the O2 front plane. I am still uncertain why there is a feedback connection for the 0V return, but without this second sense line connected the PSU does not operate properly.

SCR01.PNG

(C1 - Pgood, C2 - +3.43V, C3 - +5.1V)

Pin 14 (power good) is asserted at a 3.3V logic level well after the +3.43V and +5.1V rails have stabilized. Notice the odd ramp up of the +5.1V rail, closely following the +3.43V rail until it gets a bit more of a boost about 150ms after +3.43V has stabilized. Strange! Examining the conversion stages would probably make more sense of this.
 
Last edited:

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
Also worth noting: while testing with the cover off and fan removed, my particular O2 power supply had capacitors that were starting to make bubbling noises. Given the age of these (and many other SGI power supplies), new capacitors may be in order for continued life.
 
  • Like
Reactions: Jacques

Jacques

Active member
Dec 21, 2019
166
65
28
Somerset, United Kingdom
I appreciate it, thank you! :)
Quick question, while my O2 PSU is open for a fan replacement ( putting the original back), I noticed a potentiometer on the daughter board labeled 'Fmin'. Does this regulate the minimum fan voltage? I only ask as my fan spins much slower in the O2 while it's connected to a dedicated 12v supply outside the psu.
 

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
I apologize, I wish I could say! I did not try adjusting that or tracing it out. If you feel adventurous, you might mark its current position and try changing it?

The only other thing that occurs to me that "Fmin" could specify is perhaps switching frequency.
 

drmadison

Member
Jun 30, 2020
33
20
8
I'm curious if the machine will run if "only" 3.3V is supplied over that admittedly odd 3.43V line. And same with 5V over the 5.1V lines. Maybe at least on more limited systems? Would lead to a nice upgrade to a more modern PSU for these machines, assuming you can find one with suitable amps on the 3.3/5V lines. The Corsair RMx line of PSUs goes up to 25A on the 3.3V. But it seems crazy to replace a 170W PSU with something capable of 750W in the same of a slightly improved power factor...

(and LOL, the 1200W Corsair HX1200 WILL push 30A over 3.3V. But overkill much?)
 

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
My guess is that the 3.43 / 5.1V is used to compensate for a bit of voltage drop at higher current draw. The logic itself is almost certainly 3.3 and 5V. It might run if only supplying 3.3/5V at the connector, but you might be starving things a bit when there is a lot of draw on those rails. This could lead to instability, or it might just work!

Watch out on the ATX PSUs; with any relatively recent ones there is usually a maximum wattage between the 3.3 and 5V rails as they share a converter.
 

Jacques

Active member
Dec 21, 2019
166
65
28
Somerset, United Kingdom
There is a post on one of the old Nekochan threads suggesting there *is* a POT for the fan speed in the O2 PSU, I will trace it out today and report.

If I wanted to start the psu out of the machine do I just short the run line with ground? This way I should be able to test the voltage across the fan header.
 

Jacques

Active member
Dec 21, 2019
166
65
28
Somerset, United Kingdom
I didn't get round to tracing the POT, but I do know the fan header has 12V across the pins from power on. I was under the impression the O2 PSU had some way to regulate the voltage subject to temperature increase?

I took the opportunity to install a Cooler Master 80mm fan with 4.23mmH2O static pressure rating, it does very well keeping the innards cool. My original fan is now very noisy. The Cooler Master model is (was) often used on CPU coolers. The Arctic F8 has woeful static pressure, not really enough to draw heat out of a R10K/R12K case.
 
  • Like
Reactions: Elf

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
If I wanted to start the psu out of the machine do I just short the run line with ground? This way I should be able to test the voltage across the fan header.
Ah, you will also want to connect the feedback lines (pins 1 and 13) to their respective rails before starting it by grounding pin 2.
 

Geary morgan

New member
Dec 5, 2021
2
0
1
Hello, Thanks for the info. was wondering if you repair or would consider repairing O2 supplies. I have two that let the smoke out,LOL.
I do not have the means to do the repair my self. if you do not any suggestions on a possible vendor that would perform the repairs?

Thanks for your consideration
 

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
Ouch! I apologize, but I unfortunately wouldn't have the time or even the experience repairing these. My ideas have mostly centered around replacing them, but it's also a really low priority project for me :(

Others may post some repair shops they have had success with? I have only heard of repairs happening a small handful of times though so it may take a while before you find someone with experience doing so.
 

vvuk

Administrator
Aug 25, 2021
39
33
18
You may want to reach out to @weblacky -- I think he's done some of these types of repairs and may be willing to do them as a service?
 

weblacky

Active member
Jan 13, 2020
181
45
28
Seattle, WA
Yo, I plan to do O2 (I have two of them after all)..but I've not gotten to O2 PSUs yet. Still finishing Indy Nidec and Tezro PSU...then I need to move to Octane!. However I KNOW for a fact that Herb @ http://retrotechnology.com/herbs_stuff/mailto.html has several of them and will likely give you a very good price! Please contact him, I've purchased a few things from him before and plan on purchasing several dead PSUs from him for refurb. He's going to give you the best price around for a tested unit on short notice.

That should get you going for now. Hold on to your broken O2 PSU, I may do a call to buy them. I do have one extra O2 PSU but it's 13 years old, untested, and it's going to be the unit I inspect and rebuild first, so I don't have any O2 PSU to sell at this time. Hopefully in the future, once I've gone through all the station PSUs (or major ones) I'll have rebuilt inventory to sell consistently...it's still too early (apologizes, life happens and this doesn't really make any money as you cannot live off selling SGI PSUs).

I'll keep the groups updated. Also FYI, I'm extremely close to advertising rebuilt/tested/warranty Indy Nidec supplies here in the next few weeks!
 

Geary morgan

New member
Dec 5, 2021
2
0
1
Perfect! Thank you guys, I will followup with your recommendation. yes keep me posted on your PSU rebuild services when available.

big thanks all!
 

mgtremaine

Member
May 3, 2020
62
35
18
San Diego, Ca
www.stellarcore.net
Ugh I think I have power supply issues sadly. Was re-reading this post because it's good enough, and I might have take the PS apart to inspect it.

This poor o2 will no complete a boot, it often makes it past chime and even comes up with the monitor sometimes before just powering off. Sometimes it doesn't make past amber. I know the sensors have a few rules (Temp above 70c, Voltage below 20% max, and Fan Speed less then 80% think are the 3.) So I suspect it's voltage. (This is R5200 300mhz with 1Gb ram).

Sadness, we've had this since 2007 and the only problem has been the disintegrating plastics...

-Mike
 

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
Somewhat expected unfortunately, they are all getting to that age! You should be able to test the voltage of the supply outside of the unit (ideally with a load). And as usual, capacitors are often the first thing to think about replacing on things this old, especially before something else goes! :)
 

mgtremaine

Member
May 3, 2020
62
35
18
San Diego, Ca
www.stellarcore.net
Somewhat expected unfortunately, they are all getting to that age! You should be able to test the voltage of the supply outside of the unit (ideally with a load). And as usual, capacitors are often the first thing to think about replacing on things this old, especially before something else goes! :)
I agree, I was thinking about this late last night. I don't have a scope but I do have fairly good multimeter. So I guess I'll take it apart and try figure out if this is with in my abilities. Not sure how to put a load on it though. (Otherwise I think I'll reach out to Ian.)

-Mike
 
  • 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