PSU advice

[rthorntn]

Hello you helpful bunch.

I have a dead PSU (try to guess from the image what device it came from? ) I disassembled it, basically 3 PCB's, a fan and an IEC AC filter thingie

I don't believe there is a circuit diagram for this PSU.

Now its not the beginners section but I'm a beginner and so I'm not going to be plugging this in to mains, in saying that, I do have access to a jaycar variac, rigol oscilloscope and micsig differential probes.

I read somewhere standard SMPS troubleshooting order applies: input, fuse, varistor, rect. bridge, capacitor, switchers, PWM, transformer, then secondary semiconductors, feedback circuit and output filtering, does that make sense, can somebody elaborate on that?

I see two riser PCBs with semiconductors on them.

The only issue I can see is at the red X, a power resistor maybe, the outer casing has come off at the top?

The fuses look OK.

Really just looking for SMPS advice, tips, tricks.

I have heard that the electrolytic capacitors go bad on them, I have 3 of these PSU's, 2 dead (one of them produced the magic smoke), 1 alive.

Thanks for looking.

Richard

[Zucca]

That's a big beast for a beginner. SPS could not be a begninner job.

Oh you spotted a broken component, that's fine. You could replace it and pray but it's not the common procedure.
You need to measure stuff, take notes and post here information.

For example the fuses just look ok or did you measured them?

What I will do is to power it up with the variac at low voltage AC and try to understand who it works and do some reverse engineering and some DMM/scope.
If you don't know where to put your hands just keep them in the pocket, you already mentioned it could be dangerous.
More picture with overlay measured value with your DMM could be helpful to nail donw the problem.
Internet is full of general troubleshooting procedure, but here we can guide you to the problem if you provide more information.

Also please provide details regarding the device where it come from, maybe repairing it is not worth.

Take care, and be safe don't play with your life.

[Specmaster]

Power resistor, highly unlikely since they need to dissipate the heat generated and that plastic casing will just the opposite. They are more likely to filter caps dumping any spikes to ground.

Photos taken from above that show the legends on the PCB would be a great help at least in identifying what that it is if nothing else.

[coromonadalix]

Is it an server supply

[rthorntn]

Thanks all, it's an SGI Octane 623w (Lucent) PSU, PSU p0rn!

What voltage would i use on the variac?

I will perform a continuity check on the fuses.

I will post some better pictures.

OK so I'm thinking out loud here, the top board is AC maybe to two 24v rails and it connects via molex to the left board which actually outputs maybe 24v, 12v and perhaps 5v to the motherboard for peripherals, then the bottom maybe outputs all the smaller voltages for the ICs, I don't understand why there are so many transformers on the bottom board doesn't that imply AC?

On the top board the two giant heatsinks what are they cooling, are they Switches as in Smps, like a mosfet?

On the right hand side of the picture what are the black rectangular plastic connectors P401 and P402 and the same connectors on the left board, they have 20 pins (I can't really see any traces coming off the pins), its actually four sockets with the corresponding pins on the Octane mainboard, elcon, a contact pin, maybe its as simple as a method to make sure you orientate the PSU properly.

Interesting voltages... two 3.45v rails, two 24v rails, 5v, 12v and 3.3v.

[coromonadalix]

Read the web about " smps topology and workings "

Simply put : the ac main voltage is converted into an dc voltage, then with an high frequency switching 50 khz and up to 200 khz or more, it is converted into an ac voltage again, to be rectified, and converted back into dc again for the outputs

or if put this way: if you have an sized transformer at 60hz, doubling its frequency will divide its size by an factor of 2, 4x times 60hz will cut its size in 4   etc ...

That way you can have very powerful smps, not taking huge space and doesn't weight a ton.  I have an linear psu "864watts" the psu weight 57 pounds, if i have the same version in smps it will be cut 1/3 of the size and weight 4 pounds
 

[coromonadalix]

Words of advice

If you probe in it with an scope, be sure to have it plugged into an isolation transformer, you will save tons of problems and maybe some damage ...

[Mr. Scram]

Words of advice

If you probe in it with an scope, be sure to have it plugged into an isolation transformer, you will save tons of problems and maybe some damage ...

If you isolate an oscilloscope you really should know what it does and what not to do afterwards. There are a few pitfalls.

[coromonadalix]

I have absolutely no problems with that, i use it all the time,  i have seen traces damages and destroyed scopes inputs because of ground loops.

The problem in smps  you can have or may have multiple grounds sections, it depends of the design ... even an positive output can be used as an ground reference if you have floating voltages regulators ...

@rthorntn    if you don't have any schematics of your smps, i would not play in it,  i could check the 4 fuses, check any evident/visual damage, the rest is up to you.

You have some for 20$ usd untested on ebay ...

What do you plan to do with those smps battery chargers    you have rcgroups forums who can give you tons of help about some smps conversions.

Oh  forgot to tell you this :   some smps absolutely need active cooling, if not they will over heat pretty fast and destroy themselves
Learned the hard way with an HP 12v 150amps psu, forgot that little detail, it lasted 5 minutes and bam ...  was bought for 20$ not a big loss

[Mr. Scram]

I have absolutely no problems with that, i use it all the time,  i have seen traces damages and destroyed scopes inputs because of ground loops.

The problem in smps  you can have or may have multiple grounds sections, it depends of the design ... even an positive output can be used as an ground reference if you have floating voltages regulators ...

@rthorntn    if you don't have any schematics of your smps, i would not play in it,  i could check the 4 fuses, check any evident/visual damage, the rest is up to you.

You have some for 20$ usd untested on ebay ...

A danger is that what's normally ground doesn't have to be that any more. Lethal voltages could be in places you don't expect or you could short parts of your circuit together through ground. You really do need to be aware what you are actually doing if you use an isolation transformer. It's an easy way of getting hurt if you just jump in.

[coromonadalix]

Some smps rectify the 120vac into 347 vdc for the main,  sure you have to be very careful, the main capacitors may keep a charge long enough to give you a big surprise

[rthorntn]

I don't need to fix it, I'm exploring, learning, it's quite a "pretty" SMPS!

Won't the faulty SMPS connected to the variac set to a low voltage and the oscilloscope with the differential probes connected give me what I need, salus cogito?

It does have a fan, need to check that, especially in the PSU that released the magic smoke, maybe the magic smoke was through overheating, it was a minute or so before the magic smoke appeared.

On the right hand side of the picture what are the black rectangular plastic connectors P401 and P402 and the same connectors on the left board, they have 20 pins (I can't really see any traces coming off the pins), its actually four "huge single pin" sockets with the corresponding pins on the Octane mainboard, elcon brand, a contact pin of some description, maybe it's as simple as a method to make sure you orientate the PSU properly but I'm thinking maybe it could be a power bus for high amperage between the two boards, strange to have the bus on the octane mainboard but it would explain why I don't have a cable that's suitable for high amperage connecting those two boards?

[rthorntn]

The only issue I can see is at the red X, a power resistor maybe, the outer casing has come off at the top?

The piece that came off is definitely ceramic, the PCB silkscreen says "R1" and it has two pins.  The visually identical component next to it "R4" says 5W on it.

[Mr. Scram]

I don't need to fix it, I'm exploring, learning, it's quite a "pretty" SMPS!

Won't the faulty SMPS connected to the variac set to a low voltage and the oscilloscope with the differential probes connected give me what I need, salus cogito?

It does have a fan, need to check that, especially in the PSU that released the magic smoke, maybe the magic smoke was through overheating, it was a minute or so before the magic smoke appeared.

On the right hand side of the picture what are the black rectangular plastic connectors P401 and P402 and the same connectors on the left board, they have 20 pins (I can't really see any traces coming off the pins), its actually four "huge single pin" sockets with the corresponding pins on the Octane mainboard, elcon brand, a contact pin of some description, maybe it's as simple as a method to make sure you orientate the PSU properly but I'm thinking maybe it could be a power bus for high amperage between the two boards, strange to have the bus on the octane mainboard but it would explain why I don't have a cable that's suitable for high amperage connecting those two boards?

What do you think is the benefit of using a variac set to a low voltage?

[rthorntn]

The fuse will blow before it kills me?

[eblc1388]

The PSU is rated from input AC 100~240V, so it will probably start working with just 60V or higher AC input.

If you don't have an isolation transformer handy, just get some transformers and have their secondary windings in series to obtain such voltage to test the PSU.

With an isolation transformer as the source, any single point on the PSU is safe to touch. However, once you connect the 0V black clip of a scope probe to the primary side of the PSU, then the whole PSU becomes live again and must be treated as such.

Don't be fool into thinking that such low voltage would be safer than the full mains voltage as the PSU Power Factor Correction circuit will immediately boosts this low voltage into 380V ~ 400V DC on the main smoothing capacitors.

[coromonadalix]

The fuse will blow before it kills me?

Nope  the fuse is the last thing to blow  lolll  many thing can happen before a fuse blow ...

The supply may start around 85 vac and up,  many good psu's i've seen are rated around 85vac ?, in my shop they all do.

[sean0118]

Yes, don't plug in into a variac, as others have said the power factor correction circuit won't start until the input is ~80Vac at which point it will start producing ~400Vdc on its output. That's what those big large 450V rated capacitors are for...

I would start off by putting your multimeter in resistance measuring mode and look for shorts with the power supply turned off. Specifically, check all transistors for shorts, to do this you must look up the part number to see what type they are (BJT, FET etc) and the pinout. Also check all capacitors/diodes for shorts, including smaller SMD ones.

edit: Oh and of course check the output rails for shorts, if you haven't already...

[Johnny10]

This helped me with my first high current SMPS power Supply repair.


http://micro.rohm.com/en/techweb/knowledge/acdc/acdc_pwm/acdc_pwm01/940/