EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: elepo on February 21, 2022, 02:56:08 pm
-
Hi
I recently tried to repair my computer PSU, and after replacing a couple components it seems to be working fine.
Now I'm wondering what steps I should take to make sure that it is completely healthy and nothing else is broken in it.
Generally what tests are usually done after a power supply repair in order to determine that is is completely repaired?
-
If you are interested here are the specifics of what happened with my PSU:
It died on me while I was gaming a few days ago (circuit breaker tripped). The 5-year guarantee period ended April 2021 and they said they won't repair stuff that are out of guarantee (I offered to pay).
I also have the corona virus and didn't have the energy to go and look for a trustworthy repair place, so I decided to open it up and take a look. (I have to admit that I was delighted when they said they won't repair it, which meant I would have an excuse to do it myself >:D)
Anyway I opened it and I found a shorted diode and a shorted MOSFET (in PFC section).
I replaced the 1N5406 diode with a 1N5408.
The MOSFET was in parallel with another one. The one in parallel was not damaged. So I figured the breaker must have tripped before the damage was able to spread further into the circuit.
I simply removed the shorted MOSFET, figuring as long as I don't draw too much current the remaining one should do. (Replacement is on its way).
I then visually inspected all the elements for any sign of physical damage.
I also inspected the driver IC to see if VCC is shorted to GND which it was not.
I didn't have any tool to check the PFC inductor.
I then packed everything together and plugged the PSU in and it turned on.
I connected a ~100Watt load to it for a few minutes and then touched the replaced diode and the MOSFETs to see if they got hot and they were only warm. (I touched them after unplugging the PSU and discharging the filter caps).
I then connected the PSU to the PC and I'm using it as I'm typing this.
I know one thing I can do is checking the waveforms on the inputs and outputs of all MOSFETs but unfortunately I don't have a scope.
-
Without a scope, load testing it as you did is about as much as you can do.
I would load it up to it's rated output current on each voltage rail simultaneously and use a multimeter to check the voltages are within specification.
-
Without a scope, load testing it as you did is about as much as you can do.
I would load it up to it's rated output current on each voltage rail simultaneously and use a multimeter to check the voltages are within specification.
I see :(
Loading it up to the rated current will be a challenge because it's a 650W supply. The 12V rail for example is rated at 50 Amps.
I have ordered a couple car headlight tungsten bulbs that should draw 150W.
The 100W test I did was actually with a heating element rated for 20 Watts. But I connected a shorter portion of it and put it in cold water. It did draw ~10 Amps but it made scary hissing noises!
-
Yeah, halogen globes and heater elements are good loads for this sort of thing.
Just connect 12V halogen globes etc in parallel until you get the correct current draw. :) A DC clamp meter is useful for making sure you don't add too much load and checking it can deliver the expected current.
-
Halogen bulbs make a cheap load, but also a problematic one especially for a SMPS as their cold resistance is about 1/10 of the operating resistance.
Chances are that you trip the overcurrent protection way before the rated load.
Simple solution: dont parallel connect them all at once, add more step by step
-
Halogen bulbs make a cheap load, but also a problematic one especially for a SMPS as their cold resistance is about 1/10 of the operating resistance.
Chances are that you trip the overcurrent protection way before the rated load.
Simple solution: dont parallel connect them all at once, add more step by step
Oh I didn't know that!
Thanks.
-
It seems you got lucky and no serious damage occured if it still works.
Even if you had a scope, probing PSU primaries is not easy because the negative there is at mains potential for half of the time due to the way full bridge rectification works.
-
A 1N5406 is a general purpose rectifier. Shouldn't this application call for a high speed diode?
-
A 1N5406 is a general purpose rectifier. Shouldn't this application call for a high speed diode?
It's not he main PFC diode I think it's this other one I have highlighted in this schematic:
(https://i.imgur.com/5qBci0E.png)