[Xav101]

So at some point within the last week or so while my machine was off, it seems that my Octane2's 747W Cherokee PSU decided to die. The machine just won't power on, no fanspin or anything. After pulling the PSU out, it seems that the death was caused by some sort of a short to GND by these 3 pins. There 3 pins are connected to some sort of +V red wires internally, although I'm unsure of the voltages.



As for the internals of the PSU it doesn't seem like any of the capacitors have been damaged and they look rather clean. No leaking capacitors. However, I did notice that one of the two fuses on the AC input had been blown, as it had a resistance of 490kOhms when measured in-circuit.




As for what actually caused it, I don't know. All the components on the frontplane look fine. The one exception to this is this tantalum capacitor, which had a fair amount of black residue on it, most of which I scrubbed off with isopropyl alcohol in an attempt to read the bottom number. Conductivity tests with a multimeter show that this capacitor is tied to the pins that were burned (through what, I don't know, however). Perhaps this capacitor developed an internal short of some sort? I'd have to pull it off the frontplane to measure the capacitance.

I've got a few different ideas so far:
1. Buy a new PSU. This is probably the simplest option. My main concern is that if there's something wrong on the frontplane, it'll burn out the new PSU.
2. Try and repair the PSU and/or replace the cap on the frontplane. Other than the fuse, I'm not sure what needs to be replaced yet - possibly a MOSFET?
3. Just replace both the frontplane and the PSU. I really don't want to have to do this and would prefer if I could fix at least one of them.

Does anyone have any suggestions of what I should try and look for when trying to repair this? It's a fairly good Octane2 so I'd really like to try and save it.

[weblacky]

Yo,
I'm a little apprehensive to speak up in this scenario. I think I can safely say we've not seen this before, at least published online (even in the old Nekochan days). This is serious and likely may or may not be fixable (or even worth the time to look into).

That being said, because no other suggestions have been posted, I'll post my two (or more) cents.

I'd push the octane PSU aside for a moment and focus on the backplane. Here's what I would try on that backplane to try to assess the real damage an whether I'd try to repair or replace the backplane if it were my machine.

I'd assume that cap is bad, but I wouldn't remove it just yet. I'd take a Bench current-limited DC power supply, set it 1 volt @ 1A to start (then step up to 3 volts @ 3 A maximum for this test if I couldn't get good readings). I'd attach the (-) negative DC PSU to the ground plane of the backplane. I'd touch the (+) positive DC Power to one of the three backplane-side) pins that burnt on that backplane connector and watch the current on the bench PSU ramp up (and watch for voltage drop). If the voltage drops to nothing and/or the Amps go to max, then I'd know there is an active short on the backplane and needs fixing. I'd then ramp up one notch to 2 volts @ 2 Amps and try it again, then use either the freeze spray, lighter fluid, finger touch, or a thermal imager method to find the "hot" component(s) on the board taking that amperage (shorted). I'd first go right for that "burnt" cap on the board, if it's super hot (burns you), then I'd stop the test and desolder it. Do the same test again with the cap removed, watch the Amps, then volts. Does it still drop voltage and "draw amps", then find the next hottest component on the backplane and repeat removal. Remove until less then a couple (10s) of milliamps or less are being drawn, ideally nothing is being drawn.

Replace the cap with a new one and retest, there should be NO appreciable draw, if there is, check where it's drawing, and keep removing and checking components. Keep replacing components you've removed in last action and retest after each replacement for the short coming back.

This basically outlines a "power supply short test" to find the parts that are passing current to ground in a short. You MAY find it's just that burnt cap, after replacing the cap, if you find no voltage drop and little to 0 current draw doing this test, then likely...the backplane is OK and now repaired. If you come across a lot of mosfets overheating during this test then you'd have to replace those mosfets AND the power IC that controls them (if these leads on the octane PSU were feeding voltage to do anything of the sort).

You can test any FETs you find using diode mode on a multimeter (confirmed by desoldering and testing out of circuit).

As for the PSU, we're slowly working towards SGI power supply repair. But far enough away that you'd need to just find another one for now.

Be aware that shorts can really travel, so when you do this test, I'd suggest you leave everything in place (as much as you can). If you get shorts doing the bench PSU test and if you can't find the components doing this, then remove your mainboard and retest and if you find the short's gone...you now know...there something in that mainboard to check for shorts and repair as well. If you test parts individually you MAY find they have shorts, but don't assume you have all the shorts figured out when you reassemble PCBs together, as the load from one PCB could have damaged another. So I'd suggest doing a low-voltage drop test first with everything. You can remove modules and retest and see if suddenly shorts go away or change. Then you'd know the module you removed is related to that circuit or shorted itself.

I wish I had better news for you, but at this time...that's the only suggestion I have, Fix and check the backplane and everything else...then get a new PSU and cross your fingers. It's not really practical (because of the HUGE amperage draw of most SGIs) to build an AC current limiter with 300W light bulbs or the like. You'd have to make a large setup to current limit a octane PSU, so unfortunately it's not easy to do.

I'd also suggest that you might want to try this exact test on the octane PSU itself (unplugged of course). Use the PSU chassis ground and the burnt pins with like 2-3V @ 2A before changing that axial fuse in the octane PSU, then try it after changing the fuse and track down any shorted damage. My hope is you'll find the fuse is the only damaged part and your PSU may still work yet. But yes you can shove small voltages up the DC voltage lines, there should be no real amperage draw because the mosfets have no power and act as diodes, basically blocking any further ingress of power backwards through a switching PSU. It's fair way for checking secondary PSU output stages for shorted caps and the like!

From your description this sounds like a cap (or similar) failed in the backplane shorted to ground which over drew the PSU and the fuse protected the PSU from over taxing itself, because that cap became a wire between hot and ground. The burned pins show the jules affect on the connection between chassis and PSU. but after shining them up a little it's possible that the fuse saved the PSU and it might be back in service if you remove any short you find on both sides.

Also while Ohm meters are cool, don't use an ohm meter to confirm the LACK of the short. I won't go into explaining it but, a load drop test is true, and ohm test can produce false positives. It won't hurt anything to do an Ohms test...but don't rely on the results as "Safe to proceed".

Best of luck and please post any results you get.

[Xav101]

Hey, thanks for the reply! Hopefully I'll get a chance to try a bench supply on it by the end of the month - ideally I'd be able to fix this, as Xbow 1.4s don't seem to be super common or cheap right now without just buying a whole new Octane2 (if you know where to find one for not insane prices that would probably be helpful.)

Either way, can you elaborate more on the method of testing the PSU with the bench supply? It sounds like you're saying that you can connect the +V of the bench PSU to the +V burnt pins on the PSU, and attach the bench PSU's -V line to the PSU chassis, and then see if there's a current draw? My main question is what would happen if you had say, a diode and a MOSFET in series and only one of them was dead, or the dead one was behind the one that wasn't dead. Would you still get current through the PSU?

[weblacky]

Yo,
You understand the connections correctly, normally switching power supply output voltages will be "cold ground"/Chassis linked. You can shove at least the same voltage back through the DC output line as it was designed to output from the PSU under normal operation, when the PUS is off/unplugged. Though I’d suggest staying with low voltages.

EXAMPE: So if you had an ATX PSU, for example, from a PC, it outputs -12V, 3.3V, 5V, 12V. All of these just from a secondary windings from on a single transformer. I can connect a bench (current limited!) PSU to the 3.3V line and connector ground, set the bench PSU for 3V and feed the same voltage (same polarity) back into the connector of a (unplugged from wall) ATX PSU. The line was designed to output 3.3v…so why can’t I have 3V placed on the end of it…answer…I can. It doesn’t know the difference because everything is not running, it’s just like bring up 2 stations of a production line in a factory for testing while everyone’s away and everything’s shutdown. I‘m not fighting anyone. I’m testing if current leaks into places it shouldn’t because that line is broken in some way. That’s why I need a current limited supply to test this leakage.


The primary side's AC Neutral isn't connected to chassis (at least it's against electrical code to tie outlet ground to outlet neutral in any place but your primary electrical panel). Primary AC is normally rectified before the transformer and sent through the FET to the transformer and back through neutral (coupled with Larger filter Cap(s)). The diodes on the other side of the transformer are the "start of DC". Earth ground on the chassis should be the same voltage potential as the DC negative (they come from the same source).

If it helps, you can try to find a pins on the PSU connector that are tied to the PCB ground plane on the secondary side to verify this is true and treat that as DC output ground. That might be easier for you, see if any pins connect directly to the PCB ground on the same board with the PSU connector on it using an Ohm meter.

Again, while I can't predict every situation, there shouldn't be a reason to have a diode and MOSFET in series here. If I take your question as theory, you’re worried about not finding something. Yes, that happens but believe in “path of destruction” when it comes to things overloading. You could have 5 items shorting to a grounding plane from the same voltage rail. You keep looking and testing components until the entire short goes away, whether it’s one or many…they are all connected in some way to receive that damage in the first place. I don’t think you’ll find a bad component, then a good component, then another board component hiding like that. Fire burns everything in its path…so to speak.

The transformer provides electrical separation to not only prevent damage and separate the two electrical forms but prevent dangerous AC voltage/Current from getting into the secondary side. The test I suggested (while not being the only test) will go from the output caps (PSU connector) to until the secondary output diodes (just one hop from the main transformer). It should STOP there, if the diode is shorted...it won't stop and you'll see a draw into your half of the secondary transformer wiring (inductor). This will all appear as current draw on the BENCH PSU, it doesn't mean your transformer is toast or not (that's another test), it means you're improperly seeing your transformer when you shouldn't be. You'd need to remove the diode that burnt and replace it to know.

But normally (outside the small 5v transformer and resistors used to create a "PSU standby" voltage to power the PSU main IC while asleep), the main transformer and the FET on the AC side control everything. There are multiple secondary windings, each connected to secondary fast output diode (that prevents back feeding the transformer under normal operation) and then proceeds into output Caps (the things in PSUs that often go first) for additional filtering and such to then output smooth DC voltage. This description doesn't take into account things like power factor correction, but those aren't needed in basic understanding on AC to DC current limited switching PSUs. There is also a feedback circuit to the power management IC to see the resulting voltages and change FET switch rates to keep those voltages steady as current draw changes on ALL output voltages.

Seeing as you have only one fuse blown, my assumption was your FET was likely fine as it seemed a single line DC output was overloaded, not all outputs. All DC outputs SHARE a single transformer and FET, when current demand causes voltage to drop the management IC in the PSU commands the FET to change switching rates to push the voltage UP (ohms law if you consider the FET a resistor - which by its rate changes “resistance”), this affects ALL DC outputs to some extent (not linearly).

My initial assumption from your description was, you had your XBOW over draw a single line of DC output from a multi-line PSU, one line drew over the max and a PSU fuse went. I don't know why over current protection didn't stop this, but I assume speed is the reason. The XBOW cap may have failed and drew so quickly that the management IC didn't have time to notice and limit/shutdown things. I find it little odd because we KNOW the Octane PSU has a relay it will hold open in the case of some internal short or overdraw. Why that wasn't an option in your case I don't know.

I'm not saying you don't have a burnt transformer, or FET, or anything else. I just recommended doing a first test on just the voltage line that appeared to overload. If you're good at finding them, all secondary diode outputs are a DC voltage line output. You could follow all the lines you saw to the connector to try to test all the voltage outputs you can find.

If the diodes are fine, they will not allow current flow back into the transformer, so you'll get no draw (no connection between ground and DC output for that pin) which is what you want. No draw. If you do get draw, you could have a burnt output cap on that voltage line, a burnt secondary output diode, also a burnt transformer winding, possibly more.

If you find your bench power supply doesn’t voltage drop and you don't have a short (after fuse replacement), it's honestly possible that the fuse saved you and it's replacement would put the PSU back in service (just as long as the XBOW doesn't overdraw it again :-). That’s why I focused on fixing your XBOX first…it likely caused this PSU damage.

I hope that answered your question.

[Xav101]

Yeah, that seems to have helped. I'll keep this in mind while poking around in it.

Another thing that I unfortunately seem to have forgotten to make clear in my original post was that the machine was off when it failed. I was using it for the weekend, and then turned it off, and when I tried to turn it back on the next weekend it was just dead. Perhaps this is why the overcurrent regulation didn't kick in?

Either way, hopefully I can revive this. I'm too invested in this now to just give up. That being said, if for some reason I can't save it, maybe I'll just get the biggest sun machine I can find to fill the gap

[weblacky]

Oh,
That was actually really important information, I'd change my assessment of what likely happened based on that info (do the same tests though).

That means you likely blew an output diode or your FET (or both) during last shutdown (bad filtering caps likely). If you started the system and nothing happened (it didn't fail while running or go POP during start), then it actually blew during shutdown (happens). Normally when a Diode/FET blows on shutdown you momentarily produce a much larger voltage than you're supposed to, that's what burnt your cap. The cap didn't fail, the PSU over-volted it, once above its voltage rating, it shorted. This is normally caused by very bad output filtering caps on that voltage line which stress those two semiconductors. Startup and shutdown depend the MOST on all filtering caps, idle running is the least dependent on them.

So yeah, if you perform the bench PSU test you should find your PSU line shorted, you can test the FET independently under a multimeter with diode mode, but you likely lost your output diode at least (maybe main FET). So yeah, PSU caused it. I either misread or didn't understand the conditions of your systems original symptoms (sorry).

If you want to fix this you'd have to not only identify the diode/FET failure but you'll need to recap at least the path for that voltage rail.

So, concentrate on the XBOW voltage top check (cap replacement) and get another PSU for now.