Simple switching power supply repair help please

[Powerspring]

I'm trying to diagnose the PSU board in a hi-fi that is not very well. The output voltage is about 30% of what it should be, even with load disconnected.

I'm a SMPS noob so I'm very grateful for any pointers.

So...

Removed the PSU and connected it up for a probe around with the scope. It seems to oscillate for an extremely short period then rest for a few seconds, and repeat.

Looking at the datasheet for the controller IC (GR8875N), it could be in an over current protect state (even though there is no over-current event). The changes are very quick and I don't have a memory scope so it's hard to take exact measurements. COMP pin looks to hit about 4v.


Datasheet here, page 8, figure 2:  http://www.grenergy-ic.com/attach/product/20130508005510_pic.pdf

PSU is very close to reference schematic in datasheet. It has a +5v rail which also drives the feedback circuit and additional +/-15v rails with no active circuitry.

Thanks in advance

Sam

[capt bullshot]

Check / Replace the electrolytic capacitor connected to Pin 6 of the IC.

[dave_k]

Check for adequate supply voltage to the controller IC (between pin 6 and 4). If the PSU is indeed similar to the reference design, you might find the electrolytic cap associated with the Vcc pin is not healthy.

As a test try powering up the controller IC with an external supply (12v should be fine) and look for oscillation on the output pin.

[Powerspring]

Thank you both for your replies and suggestions 

I've just tried changing the VCC cap. It's still stuck in the cycle described before unfortunately.
Monitoring the VCC voltage shows a slow ramp (a couple of seconds) between 10v and 20v.

Trying to power the IC externally on 15v doesn't produce any output. This could be because the IC has a brown-out detection function so is looking for HV on the HV pin maybe?

[T3sl4co1l]

Check the "aux" supply diode?

And the transformer side should have continuity to GND (low ohms).

Tim

[Armadillo]

Check TL431 and then change the opto-isolator.

[Powerspring]

Thanks for the advice

All transformer windings are sub 1

Which diode do you mean by the "aux" diode? The diode rectifying the winding that feeds VCC tests good.

Check the "aux" supply diode?

And the transformer side should have continuity to GND (low ohms).

Tim

[Powerspring]

Thanks for the suggestion.

I put an external 5v DC supply across that part of the circuit and I got:

4.2v drop across the TL431 and 0.8v across the optoisolator emitter. (Sharp PC817)

If I increase to 6v DC supply, that increases to 0.9v.

The datasheet suggests a typical forward voltage of 1.2v (@20ma) for the emitter.

Sounds a bit low but I'm not familiar with these feedback networks. Am I probably looking at the problem here?

Thanks again!

Check TL431 and then change the opto-isolator.

[Armadillo]

Thanks for the suggestion.

I put an external 5v DC supply across that part of the circuit and I got:

4.2v drop across the TL431 and 0.8v across the optoisolator emitter. (Sharp PC817)

If I increase to 6v DC supply, that increases to 0.9v.

The datasheet suggests a typical forward voltage of 1.2v (@20ma) for the emitter.

Sounds a bit low but I'm not familiar with these feedback networks. Am I probably looking at the problem here?

Thanks again!

Check TL431 and then change the opto-isolator.

I assume you meant you applied the voltage across the +ve and -ve of the secondary. If you want to test this way, you need to apply the secondary "design voltage" to the +ve and -ve because of the fixed voltage divider resistors already installed in the circuit.
You increase the voltage until the design voltage "AND" the TL731 ref pin should read 2.5V.
If you can't get 2.5V after this, you change the TL431.
if you can get 2.5V, then you measure the PC817 Anode and Cathode voltage. If still low, change the PC817.

It is very likely PC817 or TL431.

Even if you can get the above, you still change the PC817 because the transistor side of the PC817 could be defective. There is a way to jumper it, but I don't recommend it. The best way is to change the PC817 to check it.

[Powerspring]

Happened to have a cheap USB charger that used the same voltage reference and opto-isolator... Guess they must be common parts in the PSU world.

Swapped both parts out. No change whatsoever. 

Worth a try

Check TL431 and then change the opto-isolator.

[Armadillo]

Happened to have a cheap USB charger that used the same voltage reference and opto-isolator... Guess they must be common parts in the PSU world.

Swapped both parts out. No change whatsoever. 

Worth a try

Check TL431 and then change the opto-isolator.

[You sure you swapped with working parts?]

Your comp pin is at 4V and it not correct. The opto is supposed to pull it down. Unless the IC is defective. So, we go drastic steps.
You use 100ohm resistor and pull the comp pin down, if the output voltage did not rise, you have a defective IC. You need to use light bulb to power the circuit up. Because, though unlikely, but the transformer could be shorted. [assuming the IC and the transformer are OK] [**** be careful of lethal voltage and charged cap]

Above steps assume you have already checked the obvious, example secondary rectifying diodes and capacitors are all OK, and you have already verified it. Also checked open resistors etc.

This is a drastic step, You consider it carefully, could be damaging. I do not recommend it. Please do not do it.

[Powerspring]

I assume you meant you applied the voltage across the +ve and -ve of the secondary. If you want to test this way, you need to apply the secondary "design voltage" to the +ve and -ve because of the fixed voltage divider resistors already installed in the circuit.
You increase the voltage until the design voltage "AND" the TL731 ref pin should read 2.5V.
If you can't get 2.5V after this, you change the TL431.
if you can get 2.5V, then you measure the PC817 Anode and Cathode voltage. If still low, change the PC817.

It is very likely PC817 or TL431.

Even if you can get the above, you still change the PC817 because the transistor side of the PC817 could be defective. There is a way to jumper it, but I don't recommend it. The best way is to change the PC817 to check it.

(with the replacement parts fitted)

5V was the design voltage, and this was applied at the to 0v and 5v rails.
The ref pin only read 1.5V
I slowly increased the bench power supply from 5V, the ref pin reached 2.5V when the bench supply was @ 10V.
I went as far as 15V and ref remained at 2.5V

I'm a slow typer so I think I'm one post behind you.

Thanks again for your assistance!

[Armadillo]

I assume you meant you applied the voltage across the +ve and -ve of the secondary. If you want to test this way, you need to apply the secondary "design voltage" to the +ve and -ve because of the fixed voltage divider resistors already installed in the circuit.
You increase the voltage until the design voltage "AND" the TL731 ref pin should read 2.5V.
If you can't get 2.5V after this, you change the TL431.
if you can get 2.5V, then you measure the PC817 Anode and Cathode voltage. If still low, change the PC817.

It is very likely PC817 or TL431.

Even if you can get the above, you still change the PC817 because the transistor side of the PC817 could be defective. There is a way to jumper it, but I don't recommend it. The best way is to change the PC817 to check it.

(with the replacement parts fitted)

5V was the design voltage, and this was applied at the to 0v and 5v rails.
The ref pin only read 1.5V
I slowly increased the bench power supply from 5V, the ref pin reached 2.5V when the bench supply was @ 10V.
I went as far as 15V and ref remained at 2.5V

I'm a slow typer so I think I'm one post behind you.

Thanks again for your assistance!

Then you check the voltage divider first. The resistor ohms should be measured. Also you applied 5V, but does the rail read 5V? Please measure it.
[is your replacement TL431 OK?]

[Powerspring]

Then you check the voltage divider first. The resistor ohms should be measured. Also you applied 5V, but does the rail read 5V? Please measure it.
[is your replacement TL431 OK?]

I applied 5v from a variable bench supply and verified with DVM. The replacement parts both came from a working donor USB charger.

The voltage divider is formed of 27K on the top and 12K paralleled with 360K on the bottom so the previous 1.5v measurement is about right.

The power supply is screen printed with "+5v" and I now think this is wrong. The control board that this powers is printed "9V" for the same rail. I hadn't noticed this before.

Recalculating the divider for 9V gives a theoretical 2.7v on the ref. pin. I've just re-tested it and 9.9v applied from bench PSU gives 2.5V on the ref pin and 1.18v across the photodiode.

Seems good

Your comp pin is at 4V and it not correct. The opto is supposed to pull it down. Unless the IC is defective. So, we go drastic steps.
You use 100ohm resistor and pull the comp pin down, if the output voltage did not rise, you have a defective IC. You need to use light bulb to power the circuit up. Because, though unlikely, but the transformer could be shorted. [assuming the IC and the transformer are OK] [**** be careful of lethal voltage and charged cap]

The comp pin ramps up to about 4v over about 1.2mS then falls to 0v instantly (where it stays for a couple secs). I'm trying to catch this on a non memory scope by eye so measurements are far from accurate.

I've already got the power supply hooked up to isolation transformer with a series lamp (as I've been poking around with the scope).

I tried pulling down the comp pin with a resistor. The output pin oscillates for a few mS then shuts off. VCC voltage shows a slow ramp (a couple of seconds) between 10v and 20v and repeats.

But wouldn't pulling comp pin low simulate the output being overvoltage anyway?

 

I'd really like to swap out the control IC GR8875N but I can't find one anywhere. An AliExpress seller had some a year ago.

Thanks again!

[Sergej97]

Rectifier diodes may fail, which after T1 transformer. So on high frequency it looks like short circuit.

[Armadillo]

But wouldn't pulling comp pin low simulate the output being overvoltage anyway?.

Thanks again!

Because there is a separate comparator for the comp pin, any voltage above 3.5 volts, shut down the chip as OLP.

So the comp pin is not stayed at 4V.

[Armadillo]

With your isolation transformer and lamp properly setup, it good you have proper setup. 

can you measure the voltage across the mosfet source resistors with your oscilloscope, any overvoltage there?

and next we will check pin 1, what is the voltage there?

[Armadillo]

1.18v across the photodiode.

In this situation, the collector voltage at PC817 is below 3.5v?

I reckon the output voltage of your mystery power supply is 12Vdc, can't be oddball 9.9v power supply.   

[Powerspring]

Rectifier diodes may fail, which after T1 transformer. So on high frequency it looks like short circuit.

All good 

With your isolation transformer and lamp properly setup, it good you have proper setup. 

can you measure the voltage across the mosfet source resistors with your oscilloscope, any overvoltage there?

and next we will check pin 1, what is the voltage there?

Was doing some testing...

Pin 1 is at 2.5v when controller is active

Drop across source resistors is about 0.4v

Then it started working! Oscillating continuously. 9.3v on the output.

I tried powering it off and on again... it didn't work until I probed the source resistors... then it started working again

I tried this several times, and tried pushing the board in this area with an insulated screwdriver. This also made it work.

A simple dry joint? I re-soldered everything in the area.

But now nothing, not even when I push the area of the board that was successful before.

I've spent the last hour metering pcb traces and looking for dry joints but I can't find anything. 

[Armadillo]

Then repeat the exercise, start measurement again.

With the setup and scope, what can you not repair?

This intermittent probing the resistor ON OFF may have killed something.

[Armadillo]

What's Up? Any progress? 

[Powerspring]

What's Up? Any progress? 

Haha thanks for the encouragement

I have family commitments so testing will resume Monday 

[Powerspring]

Over the last couple of days I have established...

The power supply is behaving as it did originally. A few ms of oscillations then a couple of seconds rest. VCC rising to 20v then falling to 10v and repeating every couple of seconds. When the ic is oscillating, the COMP pin rises to 4v then falls to 0v when it stops. Other measurements similar to before also.

Out of interest, I tried temporarily swapping in a different mosfet. No change.

I had a probe around with the mosfet out of circuit. The gate waveform seemed a fair bit cleaner at about 15v peak (about half that before). I wonder if the 8875 is struggling to drive the mosfet properly.

Still unable to find a GR8875N anywhere, I started looking at replacements. I have ordered up 5pcs LD7751 which seems close. As I mentioned before, I'm a smps noob so I'd love any feedback about this choice. The parts will be a few weeks. I go away on holiday around this time so I'll check in again when they're here and I'm back home.

[Armadillo]

If you can, try replace the current sense resistors, let's suspect intermittent. Or temporary bridge it. 

[Powerspring]

If you can, try replace the current sense resistors, let's suspect intermittent. Or temporary bridge it. 

I have already tried bridging them when I was looking for the dry joint. Seem to loose the soft start but no other difference.