Troubleshooting LCD monitor LED backlight circuit

[tautech]

Do you mean hot air station? You don't need it and actually it is dangerous to use on phenolic PCB. Often it makes a bubble (expanding moisture) if heated with hot air.

Use mine all the time on my home etched PCB's, just heat gently enough to lift/wipe the IC off and you'll never damage a PCB.
I even reuse SOIC's lifted with hot air, haven't cooked one or damaged a PCB yet..........

[wraper]

Do you mean hot air station? You don't need it and actually it is dangerous to use on phenolic PCB. Often it makes a bubble (expanding moisture) if heated with hot air.

Use mine all the time on my home etched PCB's, just heat gently enough to lift/wipe the IC off and you'll never damage a PCB.
I even reuse SOIC's lifted with hot air, haven't cooked one or damaged a PCB yet..........

Do you etch phenolic PCBs or FR4 may I ask?

[tautech]

Do you mean hot air station? You don't need it and actually it is dangerous to use on phenolic PCB. Often it makes a bubble (expanding moisture) if heated with hot air.

Use mine all the time on my home etched PCB's, just heat gently enough to lift/wipe the IC off and you'll never damage a PCB.
I even reuse SOIC's lifted with hot air, haven't cooked one or damaged a PCB yet..........

Do you etch phenolic PCBs or FR4 may I ask?

phenolic PCBs

[wraper]

phenolic PCBs

Don't know why someone would use such crappy material for their DIY PCBs  . Maybe it is better than average. Made a video specially for you.

[tautech]

phenolic PCBs

Don't know why someone would use such crappy material for their DIY PCBs  . Maybe it is better than average.

As you and I are intelligent men, I'd thought you'd grasp some of the hazards of working with FR4 in a hobbyist environment.
FIBREGLASS DUST.
Need I say more?

[Revenant]

I finally broke down and got a scope since that's something I've been planning to do for a few years anyway, and now seemed like a good time to finally pull the trigger.

I have an OZ9907GN on the way from china but it will probably be another week or two before that arrives. In the meantime, I probed the 0-3.3v vadj pin and v+led pin with respect to board ground, along with one of the 4 LED string channels.

The first 3 pictures show channel 1 as v+ led, channel 2 as vadj. The pwm signal for vadj looks messy to my untrained eye, the built in calibrator signal has a lot less noise as does a test arduino square wave generator. Maybe it's just noise from other components near the probe.

The second set of 3 pictures show the LED channel. This looks messy when brightness is set to < 100%, but as the fifth picture shows it seems fine with negligible ripple at 100% brightness. The flickering persists at 100% brightness, so I will see if any of the other 3 channels shows anything obvious on the waveform that correlates to the flicker problem. I'll also compare waveforms to the good monitor.

[Revenant]

Realized I wasn't in AC coupling mode

At 100% brightness, the ripple varies across all four channels. One of them is particularly bad, 2.4V peak to peak, and one is as low as 1.2V. I can imagine the varying ripple between each channel might cause interference flicker to show up, though I am not familiar with LED backlight circuits and what a normal expected ripple for them are, as well as the tolerance in ripple differences between LED channels. Next step will be to take apart the good LCD monitor again and look at the areas of interest on that with the scope to get a comparison.

[Rasz]

ripple

so did you, or did you not swap 4 electrolytic capacitors whose legs are visible above(below?) diodes on this pic
https://www.eevblog.com/forum/repair/troubleshooting-lcd-monitor-led-backlight-circuit/?action=dlattach;attach=152868;image

[Revenant]

so did you, or did you not swap 4 electrolytic capacitors whose legs are visible above(below?) diodes on this pic
https://www.eevblog.com/forum/repair/troubleshooting-lcd-monitor-led-backlight-circuit/?action=dlattach;attach=152868;image

The area with the 4 capacitor/diode/inductor triplets use 100nF film capacitors. I desoldered those and tested each, all came in the 96-99nF range. I wrote the ESR down somewhere and can't find the paper now but none of them tested significantly out of range of the others. It seemed like they were fine so I soldered them back in after.

EDIT: I checked the 12V rail which the OZ9907GN IC uses, the same ripple exists on there at the same frequency range, waveform looks very similar to the ripple waveform in the above posted pictures of the LED channels. Ripple Vpp is 800-920mv.

[Revenant]

I checked one of the channels on the good monitor (a lot harder to solder up test wires to the board on that one so couldn't do all 4). Ripple was a lot less on that one, about 600-800mvpp. I also checked the 12v rail and its ripple ranged from 150-300mvpp, quite a bit lower than the bad one. I have a 12v SLA battery, is there any safe way to use it in place of the existing 12v line to see if the ripple is causing the flickering? For example, one method I've seen mentioned is cutting the 12v line to the IC and substituting the battery + in its place while connecting - to ground.

[Revenant]

I decided to try and completely bypass the smps and hook up a known good ATX power supply to the 5V and 12V rails in order to see if the flickering would go away. Unfortunately, this didn't work, the monitor would briefly power up for about a second and then turn off (update- see below). I figured the reason might be the soft power on/off signal. I disconnected the ATX supply and powered it up again using its SMPS, and this time it powered up normally. I decided to probe the soft power on/off signal pin. It appears power on is +3.3V, but there is a terrible amount of ripple on this pin as can be seen from the attached picture. Could this badly rippling soft power signal be cause of everything, or is it merely a symptom of the existing ripple?

EDIT: Turns out I ran into the minimum load required ATX power on issue. Once I had a suitable load on the PS it stayed powered up. With this substitute power supply, the monitor power LED comes on, but the backlight does not. It doesn't appear the LCD is being powered either, as a flashlight to the LCD revealed no image.

[Revenant]

It appears that the LED+ rail is not being completely boosted by the OZ9907GN. I followed the LED+ PCB trace back and it ends at a transformer with two coils. A scope probe reveals a square wave output from this transformer. There was some noise on this, I'll try and get a screenshot. It appears the other coil supplies the 5V rail. The two sides of the board are mostly electrically isolated with this transformer, capacitor (looks like a MOV but is labeled C), and an optocoupler bridging the gap. On the mains side of the PCB I found another IC, OZ531GN. As expected, zero datasheets for this and the only thing I could really find was a japanese blog with a guy who has the same monitor as I do running into issues with the backlight: http://blogs.yahoo.co.jp/little_crown36/13854669.html. Google translate I think reveals that he somehow replaced this with a different chip and got it working again? it is hard to piece together the translation.

It looks like mains comes in, passes through two inductors to a bridge rectifier, after which is a filter cap and then this IC, which connects to a heat sinked transistor, and then to one of the transformer coils.

[Revenant]

I had some more time to try and further investigate the noise/ripple issues on the power rails. I've been able to confirm the noise/ripple bursts on the different rails (V+LED, 5V, 3.3V, 3.3v power_on signal) have the same period of ~2.25us. I traced the 5V rail back to the transformer, and was able to trigger off the 2.25us burst there too. That might just be caused from the issue downstream in the circuit, however, I'm not sure. I need to see if I can find this noise burst happening on the mains side next, but I'm limited to using differential measurements there because of the common earth ground issue.

The attached pictures show the 3.3V rail on ch1 and the output of the transformer for the 5V rail on ch2, triggered off the ripple bursts. The transformer waveform is quite glitchy as can be seen. Last picture shows DC coupled waveforms.

[Revenant]

It appears the 2.25us period ripple/burst is not present on the mains side, but the differential measurement was coarse and noisy so it was hard to tell definitively. I also found out that when the backlight IC is not powered on (3.3v power on signal is low), the ripple goes away and the 5V and 12V rail are very smooth. I attempted to see if the ripple would appear if I put a load on 5V rail, when I tried a couple hundred milliamps none appeared. When I tried this with the 12V rail a small amount appeared, but not at the same period as the ripple appearing when the backlight was on. I then substituted in a ripple free 3.3V signal for the power on to see if the ripple on that was causing the issue, and ruled that out, ripple still occurs.

I found out that the IC in the scorched looking area referenced earlier in this thread is almost certainly a 5V to 12V boost converter. So, it appears the problem might be in either this boost converter or in the backlight IC itself. As is par for the course on this powerboard, I could find no datasheet for that boost converter IC. I will see if I can make and post a schematic of it, perhaps there is a pin compatible IC which can be substituted?

The replacement backlight IC arrived the other day. I am holding off on replacing it until I get a few more measurements. Not sure it will even work, might be bad/defective, so it would be ideal if I can remove the IC without cutting the pins. I'll see if I can try the solder bridge/chipquik methods perhaps.

[Revenant]

I was able to successfully remove the backlight IC without having to cut the pins or use chip quik. It should even be reusable if I need it for any reason. I used the enameled wire method to accomplish this. I soldered in the replacement IC, inspected each pin visually and ensured continuity/no shorts between adjacent pins. Then powered up the monitor again and................ display STILL flickers, ripple STILL present 

Ripple period remains ~2.25us. Decided to probe the +VLED backlight rail with the display soft powered off, the 2.25us ripple goes away, however, a 41.8us ripple burst takes its place. There is a beefy through hole diode that separates this rail from the other coil on the flyback transformer. On the other side of this diode (right at the transformer), the ripple is not present. Maybe the problem lies on the +VLED rail somewhere. There are two small capacitors directly after the diode, look like they're in the pf range, I'll test those next. There is also an unlabeled inductor (no uH value listed) with what looks to be about a 5mm core and about 8 turns of what looks like 16 ga copper wire. Not sure how I can determine the proper value for this. In circuit it tests at .8uH.

I should probably at least see if the backlight cutting out problem has been resolved or not with the IC replacement.

[Revenant]

I have an update on this in case anyone's still interested. I ordered the ebay backlight replacement board which took a long time to arrive. It finally arrived the other day and with it I've been able to pinpoint that the flickering problem originates on the logic board and not the SMPS one. The same exact flicker behavior is exhibited with the replacement backlight circuit. In order to rule out ripple issues on the signal/power lines, I fed the replacement backlight circuit with 5V and 12V from separate isolated power rails that are free of the ripple. I pulled the dim signal to ground (brightness 100%) and enable signal to 5V. Flicker persists even with this.

At this point it looks like I'll need a replacement logic board for the monitor, and no surprise that there are none to be found anywhere. I know from searches that the LG panel in this board has been used in some Acer/Dell monitors, but am unable to find a definitive list of model numbers it has been used in. If I could, I might be able to find a non hp board to replace it with.

it is unfortunate that this has turned into such a time consuming, frustrating mess that's reached a dead end.