AOC LM765 PSU power cycling

[Agent24]

Got this monitor free, was told it had a bad PSU, no other information.

I tested it and it would come on then shut down. I have an AOC LM726 (https://www.badcaps.net/forum/showthread.php?t=7414) that did the same thing and it turned out to be bad caps. This one uses almost the exact same PSU (An 715L1103-1-17 Ver.B - the LM726 has 715L1103-1-C and aside from using radial coils in the inverter as opposed to toroids, they look identical) so I figured it would likely be the same thing. Wrong!


Then I noticed that one of the (boost?) coils - L202 - in the inverter was missing. One lead was still soldered but snapped off in the hole, and the other had been cleanly pulled right out of the solder. I figured that's where I had to start...

The schematic and BOM doesn't say anything useful but I got lucky when I contacted the coil manufacturer and gave them the part number, they told me 120uH. I measured the other coil and it was around 120uH. They also gave me the spec for DCR\current, and I found the closest matching choke at Farnell. I used a standard radial one as my other AOC also has those, although this one has toroidal ones. Figured it shouldn't make a difference.

While I was waiting for the coil and some new capacitors (they were all Hermei trash) I checked the capacitors but found them to be all OK.

When I got my order I put in the new coil and powered up - same problem. Then I changed all the capacitors (just in case) except the 100uF mains capacitor which tested 82uF and 0.5 Ohms and I figure is OK, but still had the same problem. I decided to check the lamps, so I tested them individually but they all came on OK and did not show pink, did not flicker and all seemed OK.

Now, I had previously fixed a power cycling issue in another PSU by replacing the controller and the symptoms on this one were the same - all rails come up to correct voltage every time but just cycle. I found I could get new controllers (SG6841) from eBay cheaply so I bought a lot of 10. When I got those, I replaced the controller but still had the exact same problem.

I was still suspicious about the inverter, so I did a good check through that for any obvious problems. Found some lovely dry joints on the transformers and on some resistors etc that I reflowed but again, no difference. Then I had the clever idea of removing it from the circuit entirely. I desoldered J208 which cuts power to the inverter section. But still the problem persists.



As it stands currently, my theory is that since I get correct voltages (albeit pulsing) from the outputs, the controller, switching transistor, transformer and rectifers are probably OK, so that leaves the feedback circuits. I checked the current sense resistor on the FET and it was OK. I have started checking the secondary side parts too, starting with the through-hole parts.

I checked ZD902 and it seemed in spec, but ZD903 was low at 4.6v - datasheet says it should be 4.9-5.1v
I am not sure if this is due to my tester though - it uses about 100v testing voltage and the datasheet says they are tested at 1.5v, although this doesn't make sense since it's a 5v Zener - how do you test a zener at a test voltage lower than the breakdown voltage? - can anyone explain that?

I would replace it just to see what happens but these diodes are hard to get. Exact number on eBay can be found but for ridiculous prices. Nothing with tight enough specifications is available locally, and I am not so sure the diode is bad anyway.


I've uploaded some photos (ignore the Hermei capacitors, they were taken before I replaced them) and the service manual. Note that the schematic is not 100% exact. There is no CN902 on pin4 of the controller, and the circled parts with IC904 are replaced with a pair of transistors and some resistors. Some resistors are not the same values as on the schematic.
EDIT: Just thought to reverse search PSU model and found another AOC SM with schematic that is correct for my monitor. See attached in the manual for model AOC 171S+

Does anyone have an idea where to look next, and if I should worry about ZD903 or not?

[budm]

Re: AOC LM765 PSU power cycling

So at this point, if you run the power supply board by itself without any other board connected, do you get steady voltage or it is still ticking?

[Agent24]

Re: AOC LM765 PSU power cycling

So at this point, if you run the power supply board by itself without any other board connected, do you get steady voltage or it is still ticking?

Same problem if connected to the rest of the monitor or by itself. Without a load it cycles quicker, though.

[budm]

Re: AOC LM765 PSU power cycling

BTW " and the circled parts with IC904 are replaced with a pair of transistors and some resistors" What do you mean? that IC904 is the ref. Diode

[Agent24]

Re: AOC LM765 PSU power cycling

BTW " and the circled parts with IC904 are replaced with a pair of transistors and some resistors" What do you mean? that IC904 is the ref. Diode

They had two 431s and in my PSU version took one out, changed the circuitry. It doesn't seem to be for feedback, it's got something to do with the circuitry connected to Pin 4, I don't know what it does, the datasheet's reference design doesn't mention it.

I found another service manual that shows the correct schematic for my PSU, that should explain it better. The AOC 171S+ PDF I added to the first post.

[budm]

Re: AOC LM765 PSU power cycling

Regarding to ZD903 5.1V Zener, as you can see, there are two resistors, R924 (for 12V line feedback) and R925 (for 5V line feedback) for maintaining the output voltage, if the output tries to go up above the regulation set point, the more LED inside Opto IC 902 will be driven harder by the Ref diode IC903 so the photo transistor is also turned on harder which will cause pin2 of IC901 to swing closer to ground which will reduce the output voltage.
The ZD902 and ZD903 are used for turning on the OPTO hard on incase the feedback resistor network fail for some reason. You can have the ZD903 out to see if its Break down voltage is too low and causes you power supply to cycle on and off. Most other power supply design does not have these two Zener Diode for second line of protection.

[budm]

Re: AOC LM765 PSU power cycling

IN that circuit diagram it is used for loading down the running power supply for the IC, it is loaded through the 39 Ohms resistor, the feedback is from the pin2 voltage of IC901 and the opto IC902. I would put it back the way it was.

[Agent24]

Re: AOC LM765 PSU power cycling

IN that circuit diagram it is used for loading down the running power supply for the IC, it is loaded through the 39 Ohms resistor, the feedback is from the pin2 voltage of IC901 and the opto IC902. I would put it back the way it was.

Sorry, I think there's even more confusion now! I didn't change any circuitry myself. The SM for my monitor has the wrong PSU schematic.

Please check the AOC 171S+ SM I added to the first post - it has the correct schematic.

Thanks for the explanation on the Zeners - I will try powering it up with ZD903 removed and see if it works.
EDIT: With ZD903 removed there was no change, still ticking.

[budm]

Re: AOC LM765 PSU power cycling

OK, I am confused, which one are we working on the LM765, or 171S?

[Agent24]

Re: AOC LM765 PSU power cycling

OK, I am confused, which one are we working on the LM765, or 171S?

Sorry. The monitor is in fact an LM765 however the LM765 PSU schematic in the manual is wrong.

But, the manual for the 171S has a PSU schematic that matches my actual board.

[budm]

Re: AOC LM765 PSU power cycling

OK, now I understand.

[Agent24]

Re: AOC LM765 PSU power cycling

Just found something possibly interesting.

While power cycling with no load, I rechecked the outputs and found that the 12v output goes up to a maximum of 12.9v on my meter. I wondered if it was overshooting and causing shutdown. Because it's such a PITA to put this monitor back together for testing, I put a 15 Ohm resistor on the 12v rail for a dummy load. The PSU started up fine although the rail was then only at 11.3v. I didn't leave it for long as it was only a 5 Watt resistor but it didn't power cycle during that time.

I thought I might have it fixed and it was just cycling from no load (It was a while since I worked on it before and I couldn't remember if I had tested it after the most recent component replacements) so I dutifully put it all back together and... no dice.

I get a brief orange flash from the LED when I apply power, then it comes on again green, the speakers pop and you can hear a low background hiss, then it turns off and repeats from green. It's only orange at the initial power on. Gah...

[selldoor]

Re: AOC LM765 PSU power cycling

Hi I have just been repairing a viewsonic (so different boards ) and it had me puzzled for a while. It too just kept switching on and off - even worse I had
repaired it for 2 sec to black perhaps 2 years ago.
Anyway went back to basics and couldnt get a voltage reading on the BIG cap
- kept going to "1". It didnt look bad but perhaps very slight curve on the end.
Swapped it out and total disbelief - it worked so may be worth you checking that.

[Agent24]

Re: AOC LM765 PSU power cycling

couldnt get a voltage reading on the BIG cap
- kept going to "1". It didnt look bad but perhaps very slight curve on the end.
Swapped it out and total disbelief - it worked so may be worth you checking that.

I can't remember if I checked the voltage but it's a 100uF capacitor which tested as 82uF and 0.5 Ohms ESR which is better than what my ESR meter chart says, so I think it's OK? I might try swapping it out later if I can find a spare though.

[Agent24]

Re: AOC LM765 PSU power cycling

I have made some progress, or rather - 1 step forward and 2 steps back



I took apart my LM726 and found the following:

1) The LM726's known working PSU also power cycles with ticking with no load - so that is not abnormal for these PSUs.

2) The LM726 power cycles the same way as the LM765, when I power it with the "bad" PSU.

3) If I disconnect the 12v supply to the inverter, the "bad" PSU powers the LM726 just fine, albeit without backlights.

4) If I leave the 12v supply connected, and disconnect the inverter's power-on signal, the "bad" PSU still works in the LM726, but without backlights.

Conclusion: something is bad in the inverter circuit. I don't know what.

To check if it was my substitute coil (the only thing I had changed), I swapped a known good coil from the LM726's good PSU into the LM765's bad PSU. There was no change. I then tried my substitute coil in the LM726 and the monitor is currently working fine as I type this reply. So, I don't think the coil is the problem.



You might be wondering: did I try the 'Good' PSU in the 'bad' LM765 - well that's where it gets hairy. I did indeed try it, with mixed results...

The first time, it worked fine. The monitor came up and I got a "No signal" message bouncing around the screen. I removed power and went to get a VGA cable to hook it up to something.

I couldn't find a cable easily so I figured I could just use the 'no input' message as a test.

I powered the monitor back on, and it worked. Then, I hit the power button to try turning it off and on, to check the buttons worked etc. It turned off just fine (not immediately after pressing the button, but it did).

The problem is now that it won't come back on again since then.



I checked the power regulators (1.8v and 3.3v) in the logic board as per the troubleshooting instructions and they are OK.

I then checked the crystal as suggested and get a 3.8v pp sine wave with approx. 70ns period which works out to be close to the 14.318 MHz crystal frequency, so I assume that's OK.

Troubleshooting chart at this point says change the main IC.




So at this point, I seem to have an unknown inverter fault, and now some unknown logic board fault as well.

Anyone got ideas on how to proceed with either one?


EDIT: Logic board is fine. Controller IC appears to have been being confused by a SHORTED(!) button on the front panel. Read 33 Ohms when not pressed!

[selldoor]

Re: AOC LM765 PSU power cycling

Cliffhanger !! eh - cant wait for the next episode.

I have been tinkering with an lm919 that seemed to work sometimes. I have two and so could compare and switch parts narrowed it down to the separate inverter board which I had recapped a couple of years back. Couldnt find anything wrong and checked it out with a magnifying glass as it went out when warm. Still couldnt spot anything so just reflowed all the larger components and
......... no cliffhanger here ....... it works great. Until the next fault develops

[Agent24]

Re: AOC LM765 PSU power cycling

Haha... a cliffhanger indeed! Don't worry, more fun to come later today (hopefully).


I'm operating now on the theory that something is shorted\leaky or being mis-controlled and acting as shorted, since the inverter doesn't even seem to try to come up. The lamps certainly don't do anything.

Almost as soon as the inverter gets the power-on signal from the logic board, the whole PSU shuts down. The power-on signal controls 2 transistors which switch Vcc power to the inverter controller IC.

Since the rest of the inverter circuit can be connected to 12v as long as that power-on signal is not present (and thus the controller IC isn't getting power) the PSU doesn't shut down. So my first suspect is the controller itself (or its decoupling capacitor).


I haven't done troubleshooting on an inverter circuit before, so I'm not sure if there is some common procedure (like measuring waveforms, injecting an external drive signal etc) but I expect that wouldn't help here so I guess I'll start with checking the controller IC for a short and also the transformers\drive transistors etc. Might have to get around to making one of those ring testers... hmm...

[Agent24]

Re: AOC LM765 PSU power cycling

Tonight, on the season finale of Agent24 vs the AOC LM765...


I checked the controller IC on both boards and got the same readings on both between pins so I figured the IC is OK.

Then I moved on and started checking (with diode test) the other semiconductors in the inverter circuits, with everything identical on both boards until I got to channel 1 on the 'bad' board. Q206\208 tested OK but the buck FET Q204 read lower than the other 3. D202 also read shorted, as did Q211. (2SC5706)

I removed Q211 as it was the easiest to do, and found it was shorted base-emitter. I figured Q212 (another 2SC5706) was thus highly likely to be bad too and so I removed that also. It turned out to be shorted base-collector. I checked C214 just in case (as mentioned in other threads) but it was OK at 244nF.

With the shorted transistors removed, I checked the FET and diode again, both now had the same reading as the others so were likely OK.

The 2SC5706 was available at Farnell and RS, but both sell them only in SMD versions, so I ordered some on eBay instead. I decided in the meantime to test the inverter by swapping two transistors from the good board. I also put the substitute inductor back in. Fired up first time and it's still going over an hour later as I write this. Looks like it's a winner!


I think I even have an explanation for why the transistors shorted, too.

You may recall from the first post that when I originally got the monitor, it had a missing a buck inductor, L202. You may also have noticed where I said “One lead was still soldered but snapped off in the hole, and the other had been cleanly pulled right out of the solder.”

I think this is the key. L202 must have had a dry joint on one lead. I suspect this intermittent joint upset the circuit and caused Q211\212 to fail, either due to them going out of correct oscillation, or perhaps the controller IC was driving the buck converter badly to compensate for the intermittent inductor, resulting in voltage spikes etc. I’m sure someone else knows what is most likely to have gone wrong but in any case, I’m happy with the result.

(Note, the monitor doesn’t have stuck pixels, that’s my camera )

[selldoor]

Re: AOC LM765 PSU power cycling

Great!! Lots of useful stuff in here- was the button board ok after all?

[Agent24]

Re: AOC LM765 PSU power cycling

The button board itself seems OK, but the button (+\move right) itself is stuffed. After removing the button the monitor had no issues.

Strangely enough the button did not read 33 Ohms after desoldering, but when I put it back today (thinking it might have been something that was solved by heating during desoldering), I got the non-responsive buttons problem again, and when checked, a 33 Ohm reading again.

I replaced it with another button (the red one in the photo) and it's been fine since. It is rather strange indeed. They also use buttons with a grounded top. I don't know what that is for (ESD protection perhaps?). I didn't have one like that so I used a standard one. I will probably order one of these http://nz.element14.com/omron-electr...0gf/dp/1960949 to replace it next time I order from Farnell, for completeness sake...