Author Topic: Old PC Power Supply (Read 3788 times)

RyanG · « **on:** July 15, 2019, 08:04:29 pm »

I have a LITEON 200W power supply that is pretty old. I'm trying to repair it because it doesn't turn on.

I checked the caps with an ESR meter and looked for burns or other physical damage with none found. I also checked the solder joints on the board, and they all look good. The switches for the mains voltage tested good and I do get voltage at the main switching MOSFET of around 160v.

The only problem I have found is that when I check the voltage on any pin of the UC3844 at the bottom right of the picture I get -160vdc on every pin. And, checking the voltage from the drain of the MOSFET to the any pin of the UC3844 I get 324vdc. Any help would be appreciated. Thank you

bob91343 · « **Reply #1 on:** July 15, 2019, 09:06:12 pm »

First see if the circuit is oscillating properly. If not, troubleshoot the feedback loop. You may have a failed transistor or diode.

If it oscillates, then you should see a nice big wave on the main FET. Verufy transformer winding continuity. Look for shorted diodes.

It's important to understand how the circuit works before you hurt yourself. The voltages get high and at this power level can be very hazardous.

coromonadalix · « **Reply #2 on:** July 15, 2019, 09:52:46 pm »

And use an isolation transformer if you probe with an scope on the primary side.

RyanG · « **Reply #3 on:** July 16, 2019, 12:03:37 am »

I don't have an isolation transformer at the moment. I have one at work for fixing the CRT power supplies I do there, but I can't get it right now.

I never get any voltage on the UC3844's VCC pin, but the OUTPUT pin has some voltage on it, around 0.5v.

The primary side of the main transformer looks good. All of the pins on the secondary side have a resistance of less than 40 ohms between each other.

It has 2 schottky diodes on a shared heatsink. One of them looks normal with around 0.2v drop on each of the diodes in it while the other has around 0.01v drop on each. That is testing them in circuit.

xavier60 · « **Reply #4 on:** July 16, 2019, 12:31:43 am »

Quote from: RyanG on July 16, 2019, 12:03:37 am

I don't have an isolation transformer at the moment. I have one at work for fixing the CRT power supplies I do there, but I can't get it right now.

I never get any voltage on the UC3844's VCC pin, but the OUTPUT pin has some voltage on it, around 0.5v.

The primary side of the main transformer looks good. All of the pins on the secondary side have a resistance of less than 40 ohms between each other.

It has 2 schottky diodes on a shared heatsink. One of them looks normal with around 0.2v drop on each of the diodes in it while the other has around 0.01v drop on each. That is testing them in circuit.

First make sure that the design has a bleed resistor from MOSFETs Gate to ground to stop the Gate drifting high while in the fault state.
Look for high value resistors that supply start up current to the UC3844's VCC pin.

More: The start resistors often serve as bleed resistors for the large capacitors. Check for residual charge before touching it.

RyanG · « **Reply #5 on:** July 16, 2019, 01:10:48 am »

There is around 20M $\$\Omega\$$ between the gate and GND on the UC3844, so I am guessing there is no resistor there.

There is around 500k $\$\Omega\$$ between the VCC and the high voltage rail.

xavier60 · « **Reply #6 on:** July 16, 2019, 01:17:15 am »

Quote from: RyanG on July 16, 2019, 01:10:48 am

There is around 20M $\$\Omega\$$ between the gate and GND on the UC3844, so I am guessing there is no resistor there.

There is around 500k $\$\Omega\$$ between the VCC and the high voltage rail

Check for residual charge on the large capacitors. As a precaution, add a resistor between Gate and Source, 10K to 100K.
The 500K seems too high. Remove and check the high value resistors.

RyanG · « **Reply #7 on:** July 16, 2019, 01:19:00 am »

This PSU has a 115v/230v switch on it, and it appears that on the 115v setting it powers the UC3844 from the negative high voltage, which would explain why I was reading 324v between the MOSFET and the pins of the UC3844

RyanG · « **Reply #8 on:** July 16, 2019, 01:31:41 am »

I added the 10k resistor across the gate and source.

I had the transformer out when I measured the resistances earlier. So, there is about 30k to the positive high voltage rail from the VCC pin on the UC3844

xavier60 · « **Reply #9 on:** July 16, 2019, 01:47:28 am »

Quote from: RyanG on July 16, 2019, 01:31:41 am

I added the 10k resistor across the gate and source.

I had the transformer out when I measured the resistances earlier. So, there is about 30k to the positive high voltage rail from the VCC pin on the UC3844

For the UC3844 to start, its VCC voltage needs to reach 16V from the start resistor then it gets powered from a rectified transformer winding if the PSU starts normally. The voltage on the VCC pins needs to be maintained above 10V.
After replacing the VCC pin's bypass capacitor, it's best to monitor any activity on the pin with an oscilloscope as mains is applied via an isolation transformer.

RyanG · « **Reply #10 on:** July 16, 2019, 02:07:13 am »

I replaced the VCC cap, but the voltage only goes up to about 1.8v on the VCC pin.

fzabkar · « **Reply #11 on:** July 16, 2019, 02:20:06 am »

I agree that 500K seems too high for the start resistor. I would have expected something like 100K - 150K.

xavier60 · « **Reply #12 on:** July 16, 2019, 02:27:10 am »

Quote from: RyanG on July 16, 2019, 02:07:13 am

I replaced the VCC cap, but the voltage only goes up to about 1.8v on the VCC pin.

There might not be enough start current or leakage path to ground.
Have the high value resistors been checked?
Add a 1k resistor from VCC to ground and measure the voltage drop to calculate the start current.
There could be protection circuitry that is locking down VCC after some delay period or after the first failed start attempt.

RyanG · « **Reply #13 on:** July 16, 2019, 02:31:49 am »

I can see that the VCC pin is connected to the cap then through a low value resistor then a diode to the transformer, which is what will hold the voltage after it starts.

Following the traces away from there it goes to a 1k resistor and a 2N4403 and then through another 1k resistor to a 2N4401, but I haven't followed those transistors yet.

xavier60 · « **Reply #14 on:** July 16, 2019, 02:48:06 am »

Quote from: RyanG on July 16, 2019, 02:31:49 am

I can see that the VCC pin is connected to the cap then through a low value resistor then a diode to the transformer, which is what will hold the voltage after it starts.

Following the traces away from there it goes to a 1k resistor and a 2N4403 and then through another 1k resistor to a 2N4401, but I haven't followed those transistors yet.

The complications on the VCC pin will need to be traced out and made sense of.
Have the output diodes been fully checked?
Is there any brief sound from the PSU when mains is first applied?

RyanG · « **Reply #15 on:** July 16, 2019, 03:51:24 am »

I don't hear any sound. I replaced the output diode that I thought was a problem with another and I had the same problem.

I guess I should have posted the model before, but it's a LITEON PA-4201-9A

xavier60 · « **Reply #16 on:** July 16, 2019, 04:03:49 am »

There appears to be yellow glue on the PCB. Where is has turned brown, it can cause problems by direct leakage or by damaging components that it's in contact with.
After is is removed, remaining invisible residue also needs to be removed with a solvent.

xavier60 · « **Reply #17 on:** July 16, 2019, 04:49:29 am »

It is common for small glass diodes to be damaged by the glue's corrosive state.
To properly clean the PCB, some components will need to be removed.
If the transformer and large capacitors are removed, the PCB can then be finally washed with hot water and dried.

fzabkar · « **Reply #18 on:** July 16, 2019, 05:20:34 am »

@RyanG, I notice that you have been asked 3 times to check the value of the startup resistor, but AFAICT you have not.

See the application circuit on page 11 of the UC3844 datasheet:

https://www.onsemi.com/pub/Collateral/UC3844-D.PDF

Note that the circuit is for a 115VAC supply. Your circuit will have a voltage doubler, so the 56K resistor in the application circuit will need to be increased to 120K or thereabouts in your case.

xavier60 · « **Reply #19 on:** July 16, 2019, 05:37:47 am »

When some of the larger components are removed, it will also be a good opportunity to remove and check the start resistors.
If that's really decomposed glue at each side of the MOSFET's heatsink, it must be clean up before powering the PSU again.
Care needs to be taken so as to not damage and obscure the value of resistors covered with glue.
I would suggest de-soldering any vunerable resistors to measure and note their values before being disturbed.
Also try to note their markings at the earliest chance.

RyanG · « **Reply #20 on:** July 16, 2019, 05:49:14 am »

I have checked for the resistors and the highest resistor I can find in the area around the UC3844 is 12k. See attached picture. I have cleaned the glue to the point where it doesn't touch anything but the heatsink.

I'm up to 1.9v on the VCC pin now.

xavier60 · « **Reply #21 on:** July 16, 2019, 06:01:48 am »

Quote from: RyanG on July 16, 2019, 05:49:14 am

I have checked for the resistors and the highest resistor I can find in the area around the UC3844 is 12k. See attached picture. I have cleaned the glue to the point where it doesn't touch anything but the heatsink.

I'm up to 1.9v on the VCC pin now.

The decomposed glue needs to be cleaned from all areas where it can cause problems. There could be more hidden by the heatsink where I suspect that the opto-coupler/s is/are. How many are there?
The start resistors will be physically larger. They could be near the large HVDC capacitors. What is that smaller electrolytic doing there?

RyanG · « **Reply #22 on:** July 16, 2019, 06:10:46 am »

There are two 100k resistors, one across each of the big capacitors.

There are two low value resistors going to the source of the MOSFET.

There is one big 5W 22k resistor above the heatsink.

Those along with the picture in my previous post are all the resistors on the high voltage side of the board.

I pulled the heatsink and cleaned all the glue

xavier60 · « **Reply #23 on:** July 16, 2019, 06:13:27 am »

What about the smaller capacitor near the large HVDC capacitors? How is it wired?

RyanG · « **Reply #24 on:** July 16, 2019, 06:21:45 am »

The negative side of the small capacitor goes to the positive side of one of the large capacitors and one winding of the transformer. The positive side is fed from the drain of the FET through a diode. The big 5W resistor is in parallel with the small capacitor

fzabkar · « **Reply #25 on:** July 16, 2019, 06:26:47 am »

AISI, the Vcc pin must be connected to the positive terminal of one of the bulk capacitors in the voltage doubler via a resistor.

xavier60 · « **Reply #26 on:** July 16, 2019, 06:30:58 am »

Quote from: RyanG on July 16, 2019, 06:21:45 am

The negative side of the small capacitor goes to the positive side of one of the large capacitors and one winding of the transformer. The positive side is fed from the drain of the FET through a diode. The big 5W resistor is in parallel with the small capacitor

Ok, that's a snubber circuit.
You will need to trace where the start current comes from either by tracing away from the UC3844's VCC or finding the source which can be one of the mains phases, the junction of the HVDC capacitors or from the HVDC rail. A common trick is to use the bottom bleed resistor as the start resistor also.

RyanG · « **Reply #27 on:** July 16, 2019, 07:03:35 am »

This is what I have traced out so far. Will do more tomorrow.

xavier60 · « **Reply #28 on:** July 16, 2019, 07:13:57 am »

It uses a charge pump from one of the mains phases. The X2 capacitor is now suspect.

xavier60 · « **Reply #29 on:** July 16, 2019, 08:25:44 am »

Now is a good time to note the values of the MOSFET's Source resistors in case of a future accident.
The current through the lower bleeder resistor could be diverted to the VCC rail for testing the PSU if a new X2 capacitor isn't available.
This has to be done properly, allowing safe clearances.

More: the grounded end of the lower resistor just needs to be moved to the adjacent empty hole that's on the VCC track.

RyanG · « **Reply #30 on:** July 16, 2019, 05:10:19 pm »

I moved the resistor to the empty hole and not much changed. The voltage on the VCC pin was still under 2.5v

xavier60 · « **Reply #31 on:** July 16, 2019, 05:26:13 pm »

Is the resistor good?

RyanG · « **Reply #32 on:** July 16, 2019, 05:27:07 pm »

Yes, the resistor is good

xavier60 · « **Reply #33 on:** July 16, 2019, 05:33:54 pm »

What voltages are there across the 1K resistors drawn on the photo in Reply #27?

RyanG · « **Reply #34 on:** July 16, 2019, 05:38:03 pm »

xavier60 · « **Reply #35 on:** July 16, 2019, 05:48:16 pm »

Time to suspect leakage to ground. Try disconnecting things on the VCC track in turn.
Start with the 39Ω resistor. Double check the polarity of the VCC capacitor or remove it.
As a precaution, solder a jumper wire between the MOSFET's Gate and Source.

Added: Allow enough time for the HVDC capacitors to discharge before doing any work to it.

RyanG · « **Reply #36 on:** July 16, 2019, 06:17:24 pm »

Before removing anything from the VCC circuit the resistance across the rail to ground is 570 ohms.

I removed(one at a time and replaced before testing next):
39 ohm resistor
Both transistors
The first 1k resistor
The diode from the charge pump
The resistor that goes to the optocoupler

The resistance stayed at 570 ohms

Should I remove the UC3844?

xavier60 · « **Reply #37 on:** July 16, 2019, 06:25:43 pm »

That's not leaving much. Try de-soldering pin 7 pin the UC3844. The MOSFET should be safe with the Gate connected to Source.
Also look for any carbon tracking on the PCB, the MOSFET's heatsink is live with HVDC plus 600V peak chopper waveform.

RyanG · « **Reply #38 on:** July 16, 2019, 06:34:04 pm »

With pin 7 desoldered, I am reading >20Meg, slowly counting down, from VCC to GND

xavier60 · « **Reply #39 on:** July 16, 2019, 06:39:54 pm »

The UC3844 was likely damaged by leakage from the heatsink through the glue.
The 100K resistor might as well be put back to where is came from.

RyanG · « **Reply #40 on:** July 16, 2019, 08:34:11 pm »

Looks like that was the problem. I had an UC3845 in my spare parts bin. I put that in and it worked. I have all the voltage rails working.

I did actually swap that out already for the same one I put in it now, but it was before I cleaned the glue. It didn't work then, but it does now.

I should have known because I think I remember Dave making a video about that exact thing.

xavier60 · « **Reply #41 on:** July 16, 2019, 10:54:31 pm »

It's lucky that the MOSFET survived it all, a big problem because of the extra damage that occurs and often not being able to see the values of damaged Source resistors.
I would have a think about replacing the 15Ω Gate drive resistor, it looks corroded,
And the snubber capacitor being so close to the heatsink and resistor.

RyanG · « **Reply #42 on:** July 16, 2019, 11:12:41 pm »

I will check everything out and replace the parts you listed if they seem corroded. Thanks for all the help.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Old PC Power Supply (Read 3788 times)

Share me