I'm stuck with a power supply that has me completely stumped. I have three of these things, all three have the same problem. The PSUs power a range of TVs. These TVs often get stuck in standby, where turning it on will make the power LED dim. During this condition, the rails of the power supply are running, but at low voltages, around 1-2V, when they should be 24V/12V/5V etc.

The service manual is here: http://www.eserviceinfo.com/download.php?fileid=56543 (page 83-85 has the PSU.)

I ordered a broken PSU accidentally. Thankfully, it only had a busted PFC section so I disabled that, turning it into a simple bridge rectifier and filter cap. Now I can get the PSU to make a slight buzzing sound and output the 1-2V but nothing else. Another PSU I messed up (removed a resistor accidentally - replaced it with a solder bridge) seems to give similar issues.

The STBY_ON_OFF pin controls the switching of the PSU. Turning the TV on causes this line to go low. Normally it floats high. All the time during the time the PSU is attempting to start this is low. I am therefore considering the main board to be okay.

I have a scope here to do some probing, but I can't probe the primary side due to the high voltages there. I also have an accurate multimeter.

I am completely lost and would appreciate some guidance on how to continue.

 

Take a look at Sheet 1 of the supply.

Straight above where it says "SHEET 1"

One side of R822 should be near zero, the other should be VF_MAIN, somewhere around 15v-20v. If VF_MAIN isn't present, find out why. I don't see where it comes from.

Check the output of the LM317; it should be around 12.6v.
If it's less, and VF_MAIN is good, then you have a problem on VCC1 or VCC2. Check to see if the caps are bulging near the 317. If VCC1 isn't up, you won't get the 2525 running.

Is 400v present? If not, the 2525 won't have any power to work with.

Divide and conquer.

 

Take a look at Sheet 1 of the supply.

Straight above where it says "SHEET 1"

One side of R822 should be near zero, the other should be VF_MAIN, somewhere around 15v-20v. If VF_MAIN isn't present, find out why. I don't see where it comes from.

Measuring to a heatsink connected to the (-) terminal of the main filter cap gives me 17.95V. The LM317 has one end connected to the resistor, the other end travels across the board to the standby controller, TNY266, on page 2 upper left.

N.B. Ignore the IRF7314 - that is the PFC controller. They put in the wrong symbol. I believe it is actually a FAN7529.

Check the output of the LM317; it should be around 12.6v.

Measures 13.13V - on the high side, but it is at least present.

If it's less, and VF_MAIN is good, then you have a problem on VCC1 or VCC2. Check to see if the caps are bulging near the 317. If VCC1 isn't up, you won't get the 2525 running.

Is 400v present? If not, the 2525 won't have any power to work with.

Divide and conquer.

Actual voltage is 330V standby, 310V run, seems a bit low but 310V run works out to 220V mains... it's 230V here. I suspect my multimeter is averaging out the ripple. I have bypassed the PFC because it was broken on one board. (Looks like the board was dropped... transformer for PFC broke off... nothing else broken though.) The other board had the same problem (PFC broken) but I did actually get it to run for a few weeks... then it died again. I haven't looked at the third board, but I suspect it to have a similar issue too.

---

A company is selling spare parts for this TV to fix my issue: http://www.sparedparts.com/products/details/17pw20-1_repair_kit_for_standby_and_back_light_faults

Looks like some polypropylene capacitors, some electrolytic capacitors and a 7805/transistor (depending on model apparently.)

I am thinking bad polyprop caps would mean it wouldn't be able to switch enough current into the transformer so would explain the low output voltages? Looks like a half-bridge drive using two MOSFETs and two caps to generate the AC for the transformer...

 

Here is the output AC from the secondary transformer.

Looks really low. There is definitely switching going on, but the output is only about 1Vp-p.

What would/could cause this problem?

The 24V rail measures 1.6V. All other rails measure in the hundreds of mV. The output diodes aren't normal diodes, they are MOSFETs configured as synchronous rectifiers, I think. If the voltage doesn't exceed the gate threshold voltage (about 1.5V), maybe the output MOSFETs will not conduct, explaining the very low readings on all rails but 24V.

N.B. It does the same thing with or without outputs plugged in (except the PSU control cable but that needs to be plugged in.)

-----

One thing I have noticed, despite the outputs on the TV being too low for anything to fire up, is that the primary side MOSFETs are quite warm... almost hot to the touch.

 

Can you get yourself an isolation transformer for your oscilloscope (or the TV)?

If you disconnected all outputs and non of the outputs has short circuit, there is no other way than to measure on the primary side.

The oscillograph you posted doesn't even look like a normal switching behaviour.

What's the voltage level at pin10, 15 and 16 of the SG2525?

PS: Is there anything worse than repairing TV-sets?

 

On IC807, the SG2525:

Are you getting ~12v-13v on +VI (pin 15)?
If so, what are you getting on VC, pin 13?

Do you have 5v on VREF, pin 16?

Are you getting anything on OUT_A (pin 11) and OUT_B (pin 14) If so, check the gate on QB14 (low-side MOSFET).

If not, check SHUTD, pin 10. If that's not 5v, then it's being pulled down by Q807; possibly a blown R971.

Find C877 and C878, they should be connected to the trafo primary pin 5. If either of those are blown/shorted you'll have problems.

Make sure you don't get on QB14's drain or anywhere on QB13 or you'll be connecting your probe to the 400v line. It would be somewhat safer to test on both sides of R861.
In any event, you might want to make up a ~10:1 divider using 910k and 100k resistors to probe with just in case QB14's drain is shorted to the gate. Ground the low side of the 100k resistor.

[eta]
Cross-posted with praondevou; we're both in the same area though.

 

Can you get yourself an isolation transformer for your oscilloscope (or the TV)?

It's on my wish list but I don't have one yet. I'd also need a HV probe, probably 1:100.

If you disconnected all outputs and non of the outputs has short circuit, there is no other way than to measure on the primary side.

All except CN803, but that only has standby voltages and a high impedance +33V bus. (For the tuner, I think.) And standby voltages are good, solid 5V & 3.3V.

The oscillograph you posted doesn't even look like a normal switching behaviour.

What would it normally look like? And why are my FETs warm?

What's the voltage level at pin10, 15 and 16 of the SG2525?

relative to (-) lead on primary cap, all measurements DC range, true RMS (does it matter?):
#10: 0.03V
#15: 13.01V
#16: 5.08V

PS: Is there anything worse than repairing TV-sets?

Agreed... but it's like a puzzle. Giving in would be too easy... Plus I spent money on buying these TVs in the hope of repairing them.

On IC807, the SG2525:

Are you getting ~12v-13v on +VI (pin 15)?
If so, what are you getting on VC, pin 13?

See above.

#13: 13.13V

Do you have 5v on VREF, pin 16?

Yes. See above, about 5.08V.

Are you getting anything on OUT_A (pin 11) and OUT_B (pin 14) If so, check the gate on QB14 (low-side MOSFET).

How do I check these without an isolated scope? (I saw a nice portable scope 10 MHz 10 MS/a and battery powered. But only 100Vpk inputs. Could be an addition to my equipment in future, but not yet.)

If not, check SHUTD, pin 10. If that's not 5v, then it's being pulled down by Q807; possibly a blown R971.

I'm getting 0.03V but I think this is good as SHUTD controls an internal transistor; to run, it should be pulled low so the transistor doesn't conduct (conducting causes it to hold the SS pin low, and SS must be more than 0.4-0.7V for the power supply to run.)

R971 measures 0.7ohms, but I don't have a 4-wire meter, so lead resistance probably makes up a big part of that. I could desolder it and pass a current through it, then measure the voltage across it, and apply ohms law, for poor mans 4-wire, but it doesn't visibly look damaged. (It's a big ~3W wirewound resistor.)

Find C877 and C878, they should be connected to the trafo primary pin 5. If either of those are blown/shorted you'll have problems.

Is there a simple in circuit test or do I need to desolder these? I've ordered some new poly caps to replace them in case they are blown. I tried testing them with my multimeter's cap range but it just tries to autorange to the µF range and flashes between that and 0.0pF, so it doesn't like them, or the circuit which they are in.

Make sure you don't get on QB14's drain or anywhere on QB13 or you'll be connecting your probe to the 400v line. It would be somewhat safer to test on both sides of R861.
In any event, you might want to make up a ~10:1 divider using 910k and 100k resistors to probe with just in case QB14's drain is shorted to the gate. Ground the low side of the 100k resistor.

I'm contacting an electronics supplier and I'm going to ask if I can get a second hand 100:1 probe and an isolation transformer. I saw they were advertising a 500VA one for £30, which isn't bad, but a good 100:1 probe will set me back *at least* twice that, and working 17MB22.1 PSUs sell for £20-£30 on eBay anyway...

 

Shutdown should be OK, it's HIGH active.
I forgot pin 8, softstart, what's the voltage level?

For the switching pulses I would expect something less... chaotic.

Well if your mosfets are warm, then there is SOME switching going on... In my opinion after visually inspecting everything and test for short circuits on major rectifiers / caps it get's difficult if you can't measure the primary side with a scope...

 

Shutdown should be OK, it's HIGH active.
I forgot pin 8, softstart, what's the voltage level?

For the switching pulses I would expect something less... chaotic.

Well if your mosfets are warm, then there is SOME switching going on... In my opinion after visually inspecting everything and test for short circuits on major rectifiers / caps it get's difficult if you can't measure the primary side with a scope...

Pin 8... this was the problem that I originally fixed. The soft-start FET was holding pin #8 low, preventing the PSU from starting. I disconnected pin #8 from the rest of the circuit (just desoldered it) and wow... it worked... then two weeks later, it died again with a flashing LED. I reversed the fix, then I got the original stuck in standby problem. The new PSU shows stuck in standby; but I'm worried that any fix I do on it will just lead to it failing again in 2-3 weeks. It's like the fix temporarily avoids another problem, but it eventually comes back to bite me.

Pin #8 at the moment measures 0.74V which is good...

 

Pin #8 at the moment measures 0.74V which is good...

Why is 0,74V good? It could be limiting the PWM pulse width. Shouldn't this be a on higher voltage level? That's why when you left it open it worked...

I disconnected pin #8 from the rest of the circuit (just desoldered it) and wow... it worked... then two weeks later, it died again with a flashing LED

This means you now have 2 problems, the original one and the one that caused the current malfunction, right?

If SHUTDOWN is LOW then SOFTSTART should be HIGH, (Q837 not conducting). If it's LOW then something else is holding it down (internally)

I'm not 100% sure but internally the softstart pin is supplied from VREF with 50uA, i.e. with the SHUTDOWN pin on LOW, why is the softstart pin only on 0.7V?

 

Why is 0,74V good? It could be limiting the PWM pulse width. Shouldn't this be a on higher voltage level? That's why when you left it open it worked...

In the datasheet soft-start low voltage is 0.7V max, 0.4V typ. I don't think it limits pulse width, but I could be wrong.

This means you now have 2 problems, the original one and the one that caused the current malfunction, right?

If SHUTDOWN is LOW then SOFTSTART should be HIGH, (Q837 not conducting). If it's LOW then something else is holding it down (internally)

I have a feeling it was just covering up the real problem, whatever that is. Maybe turning off soft start gave a bigger kick to a component and that gave it a few more weeks of life - or something like that. I'm not sure I want to do it again to this new board as I messed up the last one trying to remove my mod after it stopped working.

I have the original stuck in standby problem.

2 weeks ago: fixed stuck in standby - worked perfectly - by disconnecting pin #8
a few days ago: it died with a flashing LED, rails were stuck at around 1/3rd their rated voltage
now: accidentally ordered a broken PSU but I think I was able to fix it (just busted PFC section), this one has the original stuck in standby - rails are stuck at less than 2V.

 

In the datasheet soft-start low voltage is 0.7V max, 0.4V typ. I don't think it limits pulse width, but I could be wrong.

It's weird. I had the UC2825 in my mind, where the softstart can very well limit the pulse width, depending on where the regulation is working.

I see, it says in the datasheet, softstart voltage 0.7V, but does that mean it starts to give out PWM pulses above max 0.7V at the SS pin or does it mean that the voltage at the SS pin cannot be higher than 0.7V?

Still strange because of the internal current source to VREF and the internal (not conducting) SD transistor.

Just an idea, since the IC is not shut down and therefore the MOSFET at the SS-pin is not conducting, you could still make a test disconnecting Pin 8. There would be no difference between a non-conducting MOSFET and a not connected MOSFET....

 

It's weird. I had the UC2825 in my mind, where the softstart can very well limit the pulse width, depending on where the regulation is working.

I see, it says in the datasheet, softstart voltage 0.7V, but does that mean it starts to give out PWM pulses above max 0.7V at the SS pin or does it mean that the voltage at the SS pin cannot be higher than 0.7V?

0.7V is approx one Vbe, perhaps this is due to how the 50uA current source works? It might be limited to ~0.7V out.

Still strange because of the internal current source to VREF and the internal (not conducting) SD transistor.

Just an idea, since the IC is not shut down and therefore the MOSFET at the SS-pin is not conducting, you could still make a test disconnecting Pin 8. There would be no difference between a non-conducting MOSFET and a not connected MOSFET....

Disconnecting pin 8 is pretty tricky - I broke the last board doing it. If I have to, I can do it, I guess, but I think pin 8 is good.

 

Then there seems to be no other way than to measure the pulse width at the output with a scope ...
Maybe measuring DC voltage on the output with a multimeter can give you an idea about the duty cycle. Output HIGH level is about 18V (could verify this too with peak voltage measurement.

I don't know how well a digital MM performs in averaging a high freq PWM voltage, though.

Referring to the original problem, maybe there are other reasons why the mainboard would keep the power supply on standby? Wrong PFC output voltage etc?

edit: I don't want to insist on the softstart, but in the datasheet it says: "Soft Start Low Level - max 0.7V"
So what's the softstart High level?

 

One could also take out the gate resistor R970 (of Q837) and connect the gate via another resistor to +5Vref / GND to see what's the voltage difference at the SS pin in these two states...

 

One could also take out the gate resistor R970 (of Q837) and connect the gate via another resistor to +5Vref / GND to see what's the voltage difference at the SS pin in these two states...

That would involve me having three hands but I'll give it a go.

Would it be simpler to observe whether or not the TV switches on properly or if the rails rise? Probing pins on the bottom of the board will be difficult while holding a wire/resistor. If this pin isn't the issue, nothing should happen of note.

 

When you give the command to turn on the TV this should already change the voltage on the SD pin from HIGH to LOW, right? You would only need to measure the difference in voltage on the SS pin in these two situations.

Does this thing turn on without the PFC controller? If so, yes you could try to see if the TV turns on. However the voltage on pin 8 would still be interesting, for future projects.

 

When you give the command to turn on the TV this should already change the voltage on the SD pin from HIGH to LOW, right? You would only need to measure the difference in voltage on the SS pin in these two situations.

Does this thing turn on without the PFC controller? If so, yes you could try to see if the TV turns on. However the voltage on pin 8 would still be interesting, for future projects.

I could add an SG2525A to my Digikey order and play around with it on a breadboard, although I'm not sure if I could replicate the circuit in question. Perhaps consulting ST engineering support would be helpful.

I am going to measure the gate voltage tomorrow (it is late here.) Assuming the fet is not faulty I will see it low when trying to turn on. That would be good enough for me.

I am assuming the PWM is good. The main reason is, the MOSFETs are hot. If the duty cycle were small, I think the MOSFETs would be cooler. The weird switching waveform could just be a matter of the feedback mechanism being a bit cranky at low voltages.

So far, I think it's either the poly caps or an electrolytic cap in the primary side that has gone. The MOSFETs are switching. I hope they are switching enough. They are switching enough to create a slight buzzing sound from the PSU.

Another possibility is a dead opto or dead feedback on the secondary side.