diagnosing an old ATX PSU help identifying resistors (PowerComputing TCX-20D)

[tony359]

Hi all,

I'm diagnosing an old ATX PSU. Symptoms are:
- PSU turns on regardless of the ON/OFF status for a few seconds
- Voltages don't reach nominal
- PSU then turns off and needs power cycle to power on again

I found two charred resistors, they're across the main capacitors on the primary, I think they are bleeding resistors?

They're both charred, but I don't think it's the issue with the PSU as they read 22kOhm

See pics. Is the black band some custom tolerance then? As 22Kohm should be Red-Red-Orange.

I guess there are 390V in there, at 220KOhm it's almost 7W to dissipate. Could that be just a badly designed bleeding resistor value?

I think I'll try powering on without them, it should work anyways, just the capacitors won't discharge quickly.

Thanks for your help

[R_J]

Re: help identifying resistors

If the capacitors are in series and the resistors across each capacitor the voltage is only around 165v, so 2 watt resistors are ok. You could increase the value to 100kΩ, some use 330kΩ

[tony359]

Re: help identifying resistors

I am reading 240V across each capacitor which does not make sense as they're rated 200V. 490V across the bridge!! Fair enough, I am feeding it 260V, out of my isolation transformer.

PCB photo attached, bridge is at the bottom, caps are marked and resistors are circled.
240V means 2.6W resistors. Even 200V is 1.8V, so those resistors must have run pretty darn hot all their life!

Anyways, besides the weird voltage, that is not the issue as I suspected.

I'll look at the controller IC to see what's stopping it - also what's starting it...

[R_J]

Re: help identifying resistors

The power supply high frequency (switching) is likely effecting your meter reading.
260vac in should give around 367vdc across the bridge and caps or 180vdc across each cap
I suspect the power supply is only rated for 250vac input

[petehall347]

Re: help identifying resistors

are you sure its a 240v supply ?

[tony359]

Re: help identifying resistors

Power supply is indeed rated 240V - my isolation transformer outputs 260V, I need to check if I can lower that.

I've never seen such a high voltage across caps. Meter is a Fluke.

Anyways, I see pin 8 of the IC (ON/OFF pin, IC is FA5304) is going high/low all the time - that way I get like 2/3 of the voltage at the outputs.

Controller IC: https://pdf1.alldatasheet.com/datash.../FA5304AP.html

However if I stop probing, I can see the IC switches off after one try as the voltage drops to zero. If I put my probe back, it starts doing HIGH/LOW/HIGH/LOW again.

VCC is stable at 11.6V.
I don't see anything on the overcurrent pin.

Wondering whether the ON/OFF circuit is faulty as pin 8 does that HIGH/LOW thing regardless of whether the green pin of the ATX connector is shorted to ground or not?

Below pin8 and pin5 which is the main output.

[tony359]

Re: help identifying resistors

I think the ON/OFF circuitry needs attention. See pic below. There is a small vertical PCB which is getting AUX AC, the green "power on" wire input, GND and outputs 5VSB and the signal going to the ON/OFF pin of the controller IC.

5VSB works. I can see 5V/0V on the green wire when I short the wire to ground.

However I see 1.3V voltage drop across the optocoupler, regardless of the position of the ON/OFF switch.

It's a small bunch of passive components, I'll check them. It seems a simple circuit.

Would you agree?

[petehall347]

Re: help identifying resistors

just remembered a supply i had years ago . funny thing that was .. would work fine until a power cut then i had to unplug it and switch it off when power was back on plugged it back on and then switched it on ..

[tony359]

Re: help identifying resistors

That would work for me - if it stayed on

I'll work on it later. Meanwhile, can I replace those resistors with something of higher value? 2W - 56KOhm maybe? That would give 0.7W to dissipate which feels more manageable.
Clearly the current 22KOhm can't be right.

[tony359]

Re: diagnosing an old ATX PSU help identifying resistors

Ok, I am a bit stuck on this PSU.

First, I don't understand that extreme voltage at the bridge output. I don't see a PFC but something is boosting the voltage.

With 242VAC in, I get 479V out of the bridge which is 239.5V per capacitor and each is only rated 200V. The switching is not really working so I doubt this is affecting my reading. I have also checked with the scope and I am seeing the same values.

Maybe because of that, those 22kOhm resistors get to 100C after 45 seconds.

I did manage to fix the ON/OFF line, now the PSU only starts when I have the PSU_ON line grounded as expected. It was a broken trace.

Still, the PSU behaves weirdly
- If I am probing the ON/OFF pin, it cycles ON/OFF all the time.
- If I am NOT probing the ON/OFF pin, it only tries once, then turns itself off.
- If I feed it 240V, it's completely dead. The above only happens with 247VAC or more.
- I don't see activity on the OCP line.

The topology seems pretty basic. It looks like a cheap PSU
I am a bit stuck! Any pointer is appreciated.

[stj]

Re: diagnosing an old ATX PSU help identifying resistors

where are those 2 black wires going?
that looks like a switchable design - are you sure it's not configured for boost-mode for 120v input?

[stj]

Re: diagnosing an old ATX PSU help identifying resistors

yes, checked the trackside pic.
it's switchable - make sure the switch is off or desolder the 2 black wires - if they are linked it's in boost-mode

[CapLeaker]

Re: diagnosing an old ATX PSU help identifying resistors

Is there a 110v / 220v switch on the back of the psu? There is no way you get 480vdc out of the bridge rectifier while feeding 240vac. Unless this thing is not made for 240vac to begin with.

[tony359]

Re: diagnosing an old ATX PSU help identifying resistors

Thank you!

That's exactly why I am puzzled.

Yes, PSU is rated 110V and 240V. The switch at the back is set to 240V. I also tested it on 120V (feeding 120VAC of course) and I get 396VDC out.

I tested the switch in continuity mode and it works.

Those black wires go to what I think is the Aux transformer for the 5VSB, which works fine, I do get 5V out of the 5VSB line.

[stj]

Re: diagnosing an old ATX PSU help identifying resistors

no, the black wires link one of the incoming AC lines to the common point of the 200v electrolytics.

[stj]

Re: diagnosing an old ATX PSU help identifying resistors

remove the switch and it's wires,
and check the bridge rectifier for any shorts

[tony359]

Re: diagnosing an old ATX PSU help identifying resistors

no, the black wires link one of the incoming AC lines to the common point of the 200v electrolytics.

One of the black wires connects the common of the capacitors with the AUX PSU.

The other wire, which has the 240V/115V selector switch on it, links one side of the AC mains to the very same AUX PSU point where the other wire goes.

The bridge tests ok, no shorts.

[tony359]

Re: diagnosing an old ATX PSU help identifying resistors

My apologies you are right @stj.

What I said is also true but since the wires end up in the same place on the AUX PSU, by selecting 115V I end up connecting one of the mains lines to the common leg of the capacitors.

Forgive me for my ignorance, how is that even possible? Connect mains to an electrolytic capacitor?

[R_J]

Re: diagnosing an old ATX PSU help identifying resistors

Here are many atx smps schematics, I'm sure one will be close, and many are 110/240v switchable https://danyk.cz/s_atx_en.html
Any chance of getting the make and model of this power supply?

[stj]

Re: diagnosing an old ATX PSU help identifying resistors

you have AC on one side of the cap, the other side still gets dc from the rectifier.
when the polarity changes the cap boosts up the voltage.