Dell 1909wf

[squigley]

I have a Dell 1909WF LCD screen I am trying to fix.

Sorry, I'm telling a long story, just to show my diagnostic methods, what I've tried, what I saw, in case that helps for suggestions..

I got it because it didn't turn on, replaced the caps, as this always seems to be the cause/issue, however it still didn't work. I eventually found that one half of the AP9971GD MOSFET chip (just a pair of N channel enhancement mode fets in an 8 pin dip) was failed/shorted, so it was putting too much load on the transformer/SMPS etc, so it was cycling trying to start and restarting etc. I don't have a scope, I'm just using a voltmeter.

Removing the pins of that chip from the board allowed the screen to start up, so I just hacked off the half of the chip that was bung and stuck some random 60V 5A (or higher) rated MOSFET in it's place, and everything worked fine.

A few months later, it started developing issues with flickering, and then the backlights would turn off, requiring power cycling the monitor to get them back on. My presumption was that it must have been the other half of the chip/MOSFET which I didn't replace, had now failed, (although I am now pretty sure dud caps causes that) so I went about removing the other MOSFET, but then it was back to doing the restart cycle thing, even with both MOSFETS removed.

Reading the service manual (later), I find that this may have been caused because it has voltage sensing on the high voltage output of the backlight inverter circuit, which is uses to detect when it needs to be in "striking" mode, or normal operating mode (presumably needs a higher voltage to light the CCFLs up initially) and it shuts down/restarts if it detects the HV didn't come up at all, which it wouldn't when the FETs were missing..

As I mentioned, I now suspect the flickering and shutting off was caused by failed caps, even though I had already replaced them when I was first trying to fix the monitor, so I didn't suspect those, and rather than replacing them again, I started poking around on the high voltage DC side, trying to find if something had failed there, and in the process accidentally shorted the source or drain to the gate on the main MOSFET which drives the transformer for the low voltage DC, rendering it cactus. D'oh.

I tried a couple of random similar FETs I had in my junk as replacements for that FET, but couldn't find one that would work (and tried an IGBT, but that just blew the main AC fuse, oops), so I ordered an exact replacement for the HV MOSFET. It finally arrives from China and I just put that in, but I'm still seeing the cycling instead of starting up.

On the low voltage DC side, I disconnected everything from the transformer windings, (it has 3 outputs, 5V, 12.5V, and 16V) but couldn't measure any voltage from any of the windings. I have since realised I need to test with a diode in between the positive lead, the positive side of the winding, and the negative side of the windings to be able to measure any potential difference, anyway..

Reconnecting the 5V circuitry and powering up the power supply board results in a nice stable 5V, since this part has the opto isolator for the feedback to the gate driver on the HV DC side. When I reconnect the 12.5V audio output, and the 16V backlight inverter parts of the circuit, it starts the cycling again, and I notice that the 5V drops to an unstable 3.3V. The 16V is some unstable 12/13V.

Now I can actually even hear the MOSFET on the HV DC cycling. Ok, so there's something still wrong with the backlight inverter we'd think..

I found the service manual, which is surprisingly useful, explaining all the sections of the power supply, and the startup process etc. In here I find that it describes that the 5V supply is used on the gate driver, along with the 16V output from the transformer via the MOSFETs to power the backlight transformer.

Just for fun, I got 2 variable voltage bench/test power supplies, set one to 5V, and the other to 16V, and after connecting these to the ground of the transformer windings and the 5V circuit and the 16V circuit, after the diodes so the supplies can't backfeed the transformer, I find that the screen does a similar restarting loop, where I see pretty much no current draw on the 5V, about 200mA on the 16V, which then drops down to about 20mA, increases up to 200mA, and cycles. This I understand to be the screen detecting that the HV for the backlights didn't come up, but putting the voltmeter across the output side of the HV transformer for the backlights shows me I don't know how HV since the meter on the 1000V scale doesn't seem to go high enough, and I'm presuming this is because there's no load because the CCFLs aren't plugged in.

They weren't connected because it's awkward trying to connect them while the display board is still mounted in the back chassis and the power supply is upside down etc, so I take the display board out, and set everything up so I can connect the backlights. Applying power again, the screen lights up, and the current draw on the 16V is consistent. Interesting. I connected the display board to the panel, repower it. blue power LED comes on, screen lights up, shows me "no input detected" and the nice little RGB test square moving around the screen.. OK, so it's working. Was perhaps the cycling issue because I didn't have the CCFLs connected when I was trying to power via the main power supply?

I unhook my 5V and 16V test power supplies, and plug in the AC. We're back to unstable 3.3V and 12/13V. WTF.

Since the low voltage DC side works properly when the correct voltages are there, I'm sure the issue has to be in the HV DC area. What should I be looking at here?

I have just noticed that the service manually is saying that the gate driver IC on the HV DC side is controlled by the 16V output of the aux winding of the transformer (used for the HV side, not the LV, that's a different 16V output), and this 16V output is used to charge a 10uF cap and power the VCC pin of the gate driver. Is it possible that if this cap is defective, the VCC could be unstable, and the chip would then drive the MOSFET out of whack?

I have included the HV DC circuit diagram in case it helps.

Thanks!

Edit: A thought, is it possible that the replacement FETs being a lot higher rated could be an issue? I replaced the blown chip with a pair of IRF1404B FETs, since I found I had a handful of them, although now I am confused because I can't find the same datasheet, which I thought I looked at that showed they had a 200V rating, I am only finding 40V rates on 1404s now..

Service manual is here: https://www.manualslib.com/manual/737481/Dell-1909wf.html

Datasheet for blown fet chip (which I don't think is the issue) is here: https://pdf1.alldatasheet.com/datasheet-pdf/view/248903/A-POWER/AP9971GD.html

[squigley]

As an update..

I picked out a few things to check in the HV DC section, like a couple of caps, and the voltage coming off pin 5 of the transformer which powers VCC on the gate driver.

The caps tested fine, almost perfect for their ratings, so then I just decided to start making my way through the circuit, testing all the components. The next thing I checked was R853, which is supposed to measure 0.2 ohms. It's got red/red something bands on it (don't remember, and it's at home, and I'm at work, but it's certainly not the red, black, silver that I would expect for a 0.2), and was measuring about 0.9 ohms. After wasting a bunch of time trying to find a 0.2 or closer than 0.9 in my junk resistors, the best I found was a 1 ohm, so I added that in parallel for 0.5, in case that made any difference.

Moving on, I powered up the circuit, it's still cycling, and I'm measuring an unstable 12-13V coming out of the transformer on pin 5.

In the service manual, it says " Whenever the Vcc voltage is higher than UVLO (16V),the GATE pin will output signal to drive the power MOSFET(Q850), the high-voltage current source is off and the supply current is provided from the auxiliary winding of the trasformer PIN5."

So I suspect I should be seeing 16V or higher coming out of pin 5 of the transformer, to charge C878 (a 10uF 50V cap), and power VCC, and this unstable 12-13V may be causing U850, the gate driver IC to not startup properly?

Or perhaps it is hitting 16V, which turns the gate driver IC's output on, which turns on the MOSFET, and then it's browning out the output from the transformer, pulling pin 5 below 16V, and turning off the gate driver IC? (since I can hear it or something cycling)

I didn't think to try applying 16V to the positive side of diode D852 in case that stops it cycling.. I'll try that next though.

[squigley]

Losing my mind on this now.

I replaced the .2 ohm resistor with a .22 (closest I could buy, 2 for $2, wtf). Nothing changed.

I tested the diode off pin 5 which charges the cap and provides VCC to gate driver IC. It seemed to be faulty based on what I was measuring with my voltmeter on the diode test.. I thought I had it. I took it out, and tested it with my component tester (that $7 jobby), and it said it was a diode. Looked up the part number, SF10-04, it's a super fast rectifier diode apparently.

I put the transformer side back into the circuit, left the cap/IC out. The voltage is all over the place, from 0 to 40 volts, which seems strange to me, considering it's coming straight off the bridge rectifier (which puts out a stable 138V DC), with the only thing in line being the .2 ohm resistor.

With it out, I tried putting a stable 16V, or 20V, or up to 32V on the track which supplies the cap and the VCC of the gate driver, and this stops the cycling, but I find that it also stops the gate IC from providing and signal/power to the gate of the MOSFET.

I put the diode fully back in circuit, and the cycling continues, along with the fluctuating unstable voltage. I also hooked up the stable voltage, and it doesn't matter where this is set, as low as 6V, or up to 32V, it causes the gate driver IC to not output anything to the FET.

With the stable external voltage disconnected, I see about .25V on the output pin, yet the data sheet for the AP2761 MOSFET shows 2-4V gate thresholds (min/max), so this isn't enough to turn the MOSFET on properly (or at all).

I'm wondering if perhaps I'm looking at a failed gate driver IC now, the LD7575, which I'm wondering if I buggered when I shorted the MOSFET, as this would have put ~138VDC into the output pin.

I don't have another one in my junk, or at least not in the same form factor anyway, since I have stripped a few failed ATX PSUs and I would expect them to have one, so if I can find something similar then I might try doing some janky replacement with some wires coming off the board to the different form factor chip, at least to test if that's the problem.

I'm still really confused by the unstable voltage off the transformer, especially when that pin is not otherwise loaded, I would expect it to be stable, as the input to the transformer, from the output of the bridge rectifier, is perfectly stable (genius).

[fzabkar]

The LD7575 has hysteresis on its Vcc pin. It waits until the Vcc capacitor charges to 16V before it starts driving the MOSFET (switching from start to run mode), and then it relies on the regenerated supply from the transformer to keep the capacitor charged. If there is a short on the secondary side of the supply, or if the Vcc capacitor is dry or high ESR, then the Vcc voltage decays until it hits 10V, at which point the IC switches from run mode back to start mode. Then the cycle repeats. That's called "hiccupping".

C878 is a common culprit in similar PSU topologies.

[squigley]

Hmm, thanks for the reply. It certainly sounds like what's happening.

Last night after I posted I disconnected the low voltage DC side of the transformer for the backlight inverter and the audio/speakers part. This gave the nice stable 5V on the low voltage DC side again. In that state, I stuck a random diode into the 16V backlight output winding with the other end flying, to just measure the voltage, and I would see about 8V, which would then drop to nothing. That was strange as I would expected to see a stable 16V or even higher since it was unloaded. Maybe it was the diode I was using, since it was a 5xxx, rather than a 400x etc.

I also tried removing the MOSFET on the HVDC side, and just measured the voltage on the output pin of LD7575, which was all over the place, between almost nothing and a max of about 1.55V. Again, I'm just testing with a voltmeter since I don't have a scope, which may provide more useful info. I would have expected to see higher voltage since the FET wants 2-4V on the gate.

I have tested C878, the 10uF between the 30V aux winding of the transformer and VCC, twice, but I'll try replacing it regardless and see if that makes any difference, thanks.

[squigley]

I swapped out C878, but it made no difference. The one in there tested at 10uF with ESR of 1.3 ohms. The one I installed tested with 10uF and 2 ohms.

The 8v and then nothing that I was seeing on the backlight inverter output seemed to be due to the wrong diode I had used. I swapped the original diode around so I could measure the voltage, and I thought that I still saw wrong/low voltage, until I moved the diode to reconnect it, at which point the voltage went up to a stable 18V.

OK, maybe the cap replacement worked? No. I hooked up the panel and the video board, put the diodes for the sound in inverter back in circuit, and it was back to cycling again. Argh. I then wondered if there was something wrong with the sound part, or the video board.

I tried some combinations of having the video board attached, the panel attached, the sound/inverter board in circuit, the CCFLs connected etc, but still had the cycling problem, in some configurations. I noticed it especially when putting the sound part in/out of circuit, the cycling would stop/start. At one point I saw the CCFL (only connected 1/2 due to the orientation of the boards/screen) flicker and come on, but that was only when the video board wasn't attached.

I am beginning to think it's an issue with the current draw on the low voltage side, somehow causing the gate driver IC to brown itself out.

I recently got a similar monitor, a dell 20", so I took the back off it to see if it had the same/similar power supply board, and/or gate driver IC that I could have swapped over.. but of course it's totally different.

I also realised that connecting the external 30V supply to the VCC line of the gate driver IC should have mitigated any issue caused by C878..

[squigley]

I ordered some replacement LD7575 chips, thinking I must have damaged the one on board, which is why the output pin to the MOSFET was getting pulled down to ~1.3V, but after replacing the chip, the behaviour is identical.

I think I've pretty much replaced everything on the high voltage DC side now.

[shakalnokturn]

Sorry if I flew across some of your posts and skipped things.

Flickery CCFL's can be caused by aged CCFL's themselves, it's nice to have a spare inverter hanging around to test them individually when stuck troubleshooting.
On a couple of occasions I've seen the HV wires insulation breakdown (go resistive) more often there are leakage problems on the insulation over the ends of the CCFL electrodes sometimes causing arcs to the LCD frame.
CCFL transformers also have degraded insulation sometimes, listen carefully and look out for a faint purple glow in a dark room.

On the inverter itself, I'd check (if you haven't already done):
C2, C8 decoupling, C9, C10.
Replace C13, C14, C19 as these can be challenging to test correctly in low voltage conditions.