Indigo2 IMPACT - PSUs works only outside computer after recap

[swarm]

Hello everyone, I am in the process of repairing the IMPACT power supply. Because of the leaking caps it stopped working like all the others. I cleaned the whole board, checked everything and recapped. The power supply is now able to start up when it is not connected to the computer. With 1 ohm on the 5V rail it only delivers 4.75V. There is a similar voltage drop on the 3.5V rail as well. The Power Good signal is 6V. The 12V rail is ok. If the 5V standby is connected to the power on signal, the PSU will always start.

However, when the PSU is connected to the computer, it does not start at all when the power button is pressed. (No bad tantalum caps or shorts found in the computer). It also does not work to jump start the PSU while it is connected to the computer.

The traces, the caps... everything looks fine. The only chip that seems to be partially affected is shown. It was packaged in SO8, but the label almost disappeared. I don't know what to order as a replacement.

So:
1) Does anyone have a similar experience after recapping a dead IMPACT PSU?
2) Can someone please give me the name of the part I need?

 

[Elf]

I have not done a recap or repair on these, but I could see if I have an open one that is the same. Do you know the SGI part number?

Also one trick to make part numbers more legible is to wipe some heat sink paste over the top of the chip. It fills in the gaps in the lettering and makes it more legible once the bulk of it is wiped off.

 

[weblacky]

I've not done an Indigo2 PSU rebuild yet, but my personal experience with PSU repair has been, once your PSU no longer starts....it now has semiconductor damage and it's NOT just caps. It's caps have no gotten so bad that they have allowed damage to active components. You now have passive AND active component repairs to make.

My general rule of thumb (I'm not an electrical engineer) is, passives are mostly used to filter & oscillate and otherwise "prepare" the flow of electricity to be consumed by active components. Once the passives age or malfunction enough, they basically don't filter much or nearly at all and the once processed electricity "food" turns more poison. The active components were designed to consume yummy, processed electricity...not ripple-filled, raw, fluctuating power...they then get damaged...now more repair is needed. The time to recap a board is either when it still works, or if was put away like 10 years ago working...you don't start it...just recap it immediately before first start...most of the time it works great. The moment you let those filtering elements fail or degrade enough, that's what kills semiconductors.

In terms of advice, along with what you're doing. I'd suggests you likely don't have your standby 5V rail or something is wrong in startup with voltage present or not. You won't run at all correctly if it was your secondary diodes, so it's not your secondary diodes, nor your main FET(s). So I'd look for two things...voltage regulating transistors, and I'd recheck all your larger resistors for out of tolerance measurements as the startup circuit likely relies heavily on one or more of them. Because your standby 5V is only 1A, that tells me it's a separate circuit then your 5V rail when not running.

 

[Elf]

Because your standby 5V is only 1A, that tells me it's a separate circuit then your 5V rail when not running.

The standby voltage is indeed generated separately, while the main forward converter (providing the main 5V output and others) is shut down

 

[epitaxial]

Are these supplies switchable between 120/240? If you're running from 120 measure what's on the main high voltage cap.

 

[ghost180sx]

@swarm did you ever fix your supply?