Dell 2209wa - momentarily blinks every two seconds [ Video + Audio + Service Manual ]

[ohren]

Hello,

This is the second time I'm asking for advice in this forum. The first was about troubleshooting a PSU, which resulted in a successful repair. Naturally I want another fix of the Ikea effect.

Today I'm trying to diagnose a Dell 2209wa monitor manufactured 2008, and used 12000 h (according to factory menu). It worked fine all these years up until the other day. You'll find schematics in the attached pdf.

Please excuse possible erroneous terminology below -- I'm not that familiar with the domain.

Problem

Visible:
Screen is working normally if it's been unused for a while. After a few minutes the picture disappears (backlight goes out) for about one 60th - 120th of a second (estimated from video recording, attached in this post). This repeats about every 2 seconds.

It happens regardless of video source, whether there's any video input, or whether video cables are connected.

Audible:
Very faint buzzing coinciding with the blinking, primarily heard when pressing ear up against the monitor. Seems to be the loudest fairly close to the power cable socket.

I could predict that the screen would fail from the louder than normal buzzing that came from the monitor pre-failure. That loud buzzing is something I haven't noticed since it started blinking.


What I've done

• Inspected the power supply board. No visible problems. No bulging caps felt.
• Disassembled it down to the CCFLs. Couldn't inspect their sockets without desoldering them from the inverter board though, so didn't do that.
• Looked through the service manual (attached) and got the impression that the backlight will turn off if there is a problem with one of the lights, but I don't know enough to intuit the timescales of the processes described in the manual, and so I can't determine if it's at all a likely cause.
• Unable to determine if LCD also turns off, due to the very short time that it goes black (1/60th of a second).
• The power LED does not blink, but on the other hand it keeps shining long after unplugging the power so I disregard it as a clue.

What I haven't done

• Inspected the inverter board. It's soldered down to the lamp sockets and covered with a sheet of frosted mylar film held down with double sided sticky tape. The surface mounted electrolytic cap was visible, but I currently don't know how to read the signs of when those go bad.
• Removed the CCFLs. They look fragile. I don't know if they're locked in place or can just be pulled out. I can't see the sockets without desoldering the inverter board from all 12 sockets. The tubes look fine, nothing out of the ordinary.
• Measured voltages. I'm not that experienced so I imagine it will take a lot of time and energy for me to measure safely. I'd prefer to have some kind of plan, and some idea where to start, instead of making arbitrary measurements.

I'm wondering if someone has any intuition as to the cause and where I should concentrate my efforts? I imagine that the frequency of the blinking and the short timescale of the black screen would give some clues to someone with experience.

Can I exclude more parts by doing some additional tests?

Interesting pdf pages (mostly diagrams):

5. Power overview
6. Inverter overview
8. Inverter PWM circuit
16. DC_24V ,DC 12V and DC 5V Output Circuit and Feedback circuit
+ PCB layout pictures at the end

[ohren]

New info

• Lowering the backlight brightness results in backlight blinking with higher frequency

I noticed "You may edit your posts" at the bottom of the page but couldn't find where to do it.

[Per Hansson]

Originally Posted by ohren

I noticed "You may edit your posts" at the bottom of the page but couldn't find where to do it.

You can only edit your post for a short time after it has been posted.
This is mainly to prevent clever spam bots going in and editing in spam after a few weeks when no one will notice...

As for the fault one idea is to set a manual exposure on your camera if you can, and have it well exposed when the monitor is off.
Then power it on with a picture but also have a 500W light behind.
The idea is that if it is just the backlight that goes out you may be able to see the picture.
You can of course do this visually without a camera too...

[ohren]

Tack för the suggestions.

I first decided to treat it as a backlight issue, but after inspecting the tubes I could find nothing really objectionable. After measuring the 24 V going from the power supply to the inverter I concluded the voltage seems to drop in sync with the blinking, with no LCD connected. I'm therefore assuming that the LCD isn't relevant for isolating the problem further.

The challenge for me now is to understand whether the drop in voltage (only measured with a DMM since that's the best I can do) is the cause or the effect of the tubes shutting off. I can't intuit the probability that it would be an effect, so I can't rule it out.

Is there a way to figure out whether it's the cause (I don't have advanced equipment), or is it natural to assume it at this point and just buy caps?

[ohren]

Well, I'm considering replacing C862-C865 (can be found by searching for "C862" in the manual) which seem responsible for 24 V, unless someone has a suggestion for a better course of action. They don't look bad but that's the only idea I have.

They don't say low ESR on them. Should I aim for low ESR in the replacements anyway? I seem to recall a similar circuit when repairing a PSU which used low ESR there. Maybe I'm mistaken.

[R_J]

You say the 24v drops but to what? 23v, 22v, 0v? I would check the secondary resistance of each inverter transformer and compare the resistance, If one checks different than the others it is likely bad, also check the transistors that drive each transformer.

[ohren]

I didn't mention the voltage since I don't know if my DMM tells me the minimum voltage reached. It says ~23 V in sync with the blinking. Otherwise it's stable at basically 23.9-24.0.

Thanks for the suggestions, I'll check it out.

Sorry about the short reply. My first reply was thrown away by the forum.

[ohren]

Measured the transformers, but they all have the same values.

I had a bit of trouble figuring out how to measure the mosfets. Looked up how to do it, looked up the part, but must have missed something. I'm working under the assumption that you're referring to Q1-Q4 on page 8. I'll try again tomorrow.

Is it likely that they'd fail in a way that would cause this sort of problem? I really know very little about it, but imagined that if they'd fail it would result in some more constant state of failure.

[ohren]

Not sure if I did it right but I couldn't find an indication of any problem with any of the transistors.

[ohren]

I measured the four 470 uF capacitors on the 24 v output out of circuit with a cheap component tester:

[uF] - [ohm ESR]
430 - 0.06
437 - 0.10
439 - 0.10
437 - 0.09

I don't know... doesn't seem too bad to me. Capacitance is within 10% of spec. Is it still likely they're the problem?

Not sure what's my next step now. I'd try identifying a failing tube except that the platic holding them in place is incredibly brittle and will invariably break if I try to unplug one, and they all look in equally good condition, so I thought it'd be best to exhaust all other options first.

[ohren]

Would the above capacitor values suggest that I need to replace them, or is the likelihood high that I would do better to put them back and concentrate on other parts?

[Per Hansson]

Looks fine to me, you can also confirm the ESR with their datasheets, generally if it is about 100% higher than what it says I would consider them bad.

[ohren]

Thanks for the intuition! I couldn't find a mention of ESR in the datasheet so I decided to put the caps back. They at least seem to be high reliability 125 deg C caps.

The next thing I did was to remove one tube and plug in a standard ~11 W commercial tube used in desk lamps (of a type mentioned in that other thread that recommends using old energy saving bulbs to test for bad tubes -- see photo). I also tried to connect it in parallell with each existing tube to see if I could aviod pulling out the old ones while testing.

At this point I noticed some more gaps in my understanding of what I was looking at (I don't know much about high voltage AC):

1. When I replaced a tube, the replacement test tube (tested, working) didn't seem to light up at all. Did I have the wrong expectations? Why doesn't it visibly light up? It was undoubtedly connected (see #3). It was not connected via the conductive rubber sockets, but directly to the metal contacts.
2. When I connected the test tube in parallel with the old tube the old tube didn't light up (expected), but neither did the test tube (see picture). Is it a bad idea to test for bad tubes this way? With my limited understanding I imagined the good test tube would "take over" if it were in parallel with a tube that is bad enough to make a difference. This time the test tube was connected via the conductive rubber sockets.
3. In both of the above scenarios the backlight did not turn off, which it would have done had any of the tube sockets been empty, suggesting that the replacement tube in #1 actually "worked", without lighting up. Is this to be expected?
4. When, as a test, I disconnected one of the pins of the test tube (while it was in parallel with the old tube) it lit up more than when both its wires were connected. I'm not sure I can explain that. Granted, the test tube had its glass resting on the presumably grounded metal chassis of the backlight assembly. Does that explain it sufficiently? If I were to guess it's basically the same as holding the glass and touching one of the contacts to the transformer output, as described in another thread here, as a means to test inverter transformers.

Is there any point in testing with a test tube that doesn't light up? If so, I'll start replacing the tubes one by one (breaking a lot of plastic).

[ohren]

I isolated the problem.

Turns out the PFC + PWM controller is overheating. Alcohol sizzles on it, and the problem goes away for a while.

Before potentially replacing it the question is whether this is more likely because of internal deterioration, or failure of an external component. I'd be grateful if someone could lend me their intuition:

1. Do these controllers commonly get very hot during normal operation? I'm not sure why that would be, but if so, could that sufficiently explain internal deterioration?

2. Would the failure of some external component commonly be able to make such a controller draw too much power?

Maybe the controller is on its way out, but until then I've stuck a heatsink on it to see what happens. Looks promising so far. Still, I can feel it radiate heat far more than anything else on the board. The heat sink is at about 70 °C according to an IR thermometer.

The circuit can be found on page 75 (U850). The controller is an SG6901ASZ.

[Per Hansson]

It should not get crazy hot, however the datasheet says it dissipates 1W so it will not be exactly cold either.
Do check all the passive components around it, and if nothing bad found then I would replace it.
Or leave it running like you have it now with a heatsink, ghetto mods are fine as long as they work

[ohren]

Thanks for sticking with me and for taking the time to look up the data sheet. I would have attached it if I'd thought ahead. I didn't see your reply until I returned to report my findings but I think I basically did as you're suggesting:

I tried the ghetto mod for a day, but I knew it was stupid to put a bandaid on without knowing where the problem was, so I left the bezel off. Good thing I did, because the chassis got abnormally hot and the problem reappeared.

I opened it up tonight again and measured, and my conclusions were:

1. Fuse resistor FR851 is blown.
2. Diode D856 is shorted.

...which were confirmed by removing them both. They can be found in the circuit diagram by a text search in the manual in the OP. They're close to each other, topologically.

I checked the other fuses and diodes I could find, and I can't seem to find anything else curious about the circuit. I'm debating whether to be satisfied with that the failure of D856 was what caused FR851 to blow. Or could it be the other way around? I'm open to suggestions if it's obvious that the failure of these two also should've taken some other component with them.

Interesting that the monitor continued to be usable for hours despite this. I read up on things as I went along, but I don't have an eye for what role the diode had (in series with an NTC thermistor). Is it inrush current limiting?

Anyway, the blown fuse resistor seems to have cut off the PFC MOSFET's drain, suggesting that... no PFC was happening? That would explain where all the heat came from I suppose (my understanding of AC is limited at best, but that's what I took away from reading about PFC). Am I getting close?

Well, this is more fun than gluing a heat sink to an IC like an idiot

[ohren]

And...

3. The PFC MOSFET Q853 is shorted.

[Per Hansson]

Originally Posted by ohren

Anyway, the blown fuse resistor seems to have cut off the PFC MOSFET's drain, suggesting that... no PFC was happening? That would explain where all the heat came from I suppose (my understanding of AC is limited at best, but that's what I took away from reading about PFC). Am I getting close?

Well, this is more fun than gluing a heat sink to an IC like an idiot

I was going to suggest you to measure the voltage on the main bulk cap.
It should be around 400v when the PFC is working, or around 325v when it is not, so that would have been a tell-tale sign too.
Good job on locating the fault!
P.S: It would be a good idea to desloder the main bulk cap and measuring it on the component tester.
Just be sure to discharge it first because anything over 5v will kill the Atmega processor...

[ohren]

Thank you for the guidance. Getting an understanding of why the cap would be at those two levels with/without PFC sounds worthwhile.

So I replaced the components with stuff lying around. Diode went from 3 A → 12 A, the 0.2 ohm 1 W fusible resistor became a 1 A fuse, and I took a mosfet from an old PSU — only 500 V compared to the original 600 V but maybe that's ok. It's not like the original part survived.

I plugged it in, and for the longest time I couldn't figure out why I lost every single rail. Not even a power light. After more than an hour of troubleshooting I figured I should probably plug the power LED / power button board in and try turning the monitor on ��*♂️ Apparently it saves the power state in memory ��

The bulk cap was at 319 V when turned off, and around 370+ when on. Good call.

Power draw is about half of what it was before. The controller is cooler without the heat sink than the heat sink itself was before. Power factor is around 90-94.

Time to close it up again and let it cook. The bezel stays off for a month I think.

[Per Hansson]

Good work on getting it fixed!

Originally Posted by ohren

Thank you for the guidance. Getting an understanding of why the cap would be at those two levels with/without PFC sounds worthwhile.

I'm not sure if I misunderstood if this was a question but here you go:
With 230VAC RMS line voltage the rectified DC voltage will be: √2 x 230VAC = 325VDC
The boosted (PFC) voltage depends entirely on how the PFC controller works so that can't be calculated without tracing the whole circuit out and reading datasheets.
P.S: The PFC circuit is normally off when the unit is in standby mode.