Bad CCFL Backlight

[AaronD]

I have an LCD monitor that works...for about 1 second.  That is:
1. Power on
2. It takes a second or two to boot (I guess that's what it's doing?)
3. It displays the correct image for 1 second
4. Backlight turns off again, but the LCD panel still works if you look at it closely with a flashlight
5. Power-cycle (soft or hard) to repeat from #1

It's supposed to have more dynamic contrast than the panel itself provides, and sure enough, the power supply PCB calls out both an ON/OFF and a DIM signal between the video processor board and the CCFL inverter.  There's also a PID signal, that seems to be a constant 1.8V in all cases, plus some 60kHz noise when the backlight is on, so I have no idea what that does.  Since all the buttons go to the video processor, I figured that the (intentional) inverter control is limited to those 3 signals, so I focused my efforts there to start with.

I really don't care about dynamic contrast for this application, so I'd be okay with just jerry-rigging it "ON" if necessary.  If all of the original functionality can be kept, that's just a bonus.
I've completely replaced a CCFL backlight before, with LED strip lights (and cut the jumper that used to power that inverter, seeing that it was completely independent unlike this one, and wouldn't be missed), but since the tubes are clearly okay in this one, I'd rather not take apart the display module itself to get to them.

Here's a 1min video showing my reverse-engineered schematic, and analog 'scope captures (with my phone on a tripod) of the 3 signals between the video processor and the inverter:


Copy/pasted description:

The backlight comes on for about 1 second when the monitor turns on, then the backlight goes out again.  Power-cycling (soft or hard, it doesn't matter in this case) gets the backlight back for another 1 second, ad-infinitum.  The loud clicks in this video are from the soft power button, amplified by a lot, and the ambient noise removed.

I reverse-engineered the single-sided, through-hole PCB to get the schematic shown here, and then probed the interesting signals.  The high-frequency, low(ish)-amplitude artifacts when the Backlight is ON, are about 60kHz.  I didn't think to capture that explicitly, but pausing on the right frame shows about 3 cycles in 50us.  That's the switching frequency of the CCFL inverter, so I'm guessing it has some parasitic coupling from there.

The 15-pin Controller (CN5), is a surface-mount PCB at 90deg to the main board, using a pin header to connect them as if it were another through-hole component.  I didn't look very closely at that board.

[perieanuo]

usually you find bad electrolytics in main supply filtering, C186/87/25 ... in your schematic
if those are fine and you validated the 'commands' (on/off, dim and pid, that's simple even without oscilloscope, if you want you can push those signals after disconnecting the link from microcontroller and adding some resistors to Vcc or GND depending on command logic), then find over-current loop and see if it's fine
check also that branch ZD801, seems over-voltage protection
in your place i would replace immediately the filtering caps  C186/87/25
one ccfl may be bad also
btw, what is cn5?

[TheMG]

How can you be sure the CCFL are okay? Just because they make light for a second doesn't mean they are good. The inverter driver IC in most monitors will detect a fault condition and shut off after about a second or so, for instance if the voltage across the CCFL is too high, as happens with a CCFL backlight that is reaching end-of-life or one or more defective tubes.

If the monitor has very high hours of usage it's very likely the CCFL tubes are worn out. As they age, the voltage required to keep them lit increases, and they also run less efficiently producing more heat. You can maybe find a way to force the inverter to stay on under such conditions but it could end up causing further damage, I've seen the CCFL ends get so hot they starting melting and smoking the backlight diffusers, resulting in permanent visible defects in the backlight.

[AaronD]

usually you find bad electrolytics in main supply filtering, C186/87/25 ... in your schematic
if those are fine and you validated the 'commands' (on/off, dim and pid, that's simple even without oscilloscope, if you want you can push those signals after disconnecting the link from microcontroller and adding some resistors to Vcc or GND depending on command logic), then find over-current loop and see if it's fine
check also that branch ZD801, seems over-voltage protection
in your place i would replace immediately the filtering caps  C186/87/25
one ccfl may be bad also
btw, what is cn5?

I don't know what that supply is supposed to be, but it looks clean at ~10V on my 'scope.  So I think the caps and zener are okay.  (or at least the fault that the zener is supposed to protect against isn't happening)

CN5 is a small surface-mount board that is mounted to the single-layer through-hole board that I reverse-engineered the schematic of.  The component side is right up next to some filter caps, so I can't get a good picture without desoldering it.  I *can* see that's at least a 2-layer board with a INL816GN CCFL controller in a SOIC-16 package (https://www.google.com/search?q=inl816gn), a 9945N dual N-channel MOFSET in a SOIC-8 package (https://www.google.com/search?q=9945n), a SOT-3 package called D1, and a bunch of passives, but not much else.  The controller is placed towards the control signals, and the FET is towards the transformer, so that makes sense too.

How can you be sure the CCFL are okay? Just because they make light for a second doesn't mean they are good. The inverter driver IC in most monitors will detect a fault condition and shut off after about a second or so, for instance if the voltage across the CCFL is too high, as happens with a CCFL backlight that is reaching end-of-life or one or more defective tubes.

If the monitor has very high hours of usage it's very likely the CCFL tubes are worn out. As they age, the voltage required to keep them lit increases, and they also run less efficiently producing more heat. You can maybe find a way to force the inverter to stay on under such conditions but it could end up causing further damage, I've seen the CCFL ends get so hot they starting melting and smoking the backlight diffusers, resulting in permanent visible defects in the backlight.

Hmm...you might be right.  It was my grandma's, and it was often left on all day unattended.  She said that "her computer wouldn't turn on," so I looked at it closely with a flashlight and got her another monitor.  The electronics might be all good, but the controller IC is detecting the tubes' end of life and shutting itself off to prevent a fire.  Maybe I *do* need to put an LED strip in it after all, like I did with my other monitor that had a bad CCFL.

[sean0118]

This guide on badcaps has helped me in the past:

A guide on how to troubleshoot 2 seconds to black
https://www.badcaps.net/forum/showthread.php?t=10419


It's much easier to troubleshoot if you have spare CCFLs to test with.

[AaronD]

Well, I found that I can buy a complete refurbished LED-backlight monitor for about the cost of an LED strip and driver to put in this one.  (and I think I might have damaged the LCD panel too)  Not to mention the difficulty in finding a dense-enough packing (white or RGB) to not see the individual diodes, on a strip that is less than 7.5mm wide, which is the size of the CCFL bracket that came out.

My previous one (barely) succeeded in that, and it also happened to be individually addressable, which wasn't a requirement but whatever.  (it's now discontinued )  It also confirmed my suspicion about the required density: turning off every other LED in a 144/m strip still makes them visible unless I alternate which set is on for each 30Hz update.  (so now I can have 1/2 increments between 0-255, which is really nice in low ambient light)

For that one, I still would have gone with the LED strip and driver because I wanted that extensive control of the new backlight, but I don't need that control for this one.  So I'm buying the refurb instead.

[Koray]

If you have decided to scrap the monitor just take pictures of the boards and list them with board and model numbers on eBay. This way you can recoup some lost money, reduce landfill, and be good for the environment. I list parts like these and put monthly auto-relist on listings and sometimes a part is sold after several years. Doesn't cost anything and not much time either.

K.

[AaronD]

If you have decided to scrap the monitor just take pictures of the boards and list them with board and model numbers on eBay. This way you can recoup some lost money, reduce landfill, and be good for the environment. I list parts like these and put monthly auto-relist on listings and sometimes a part is sold after several years. Doesn't cost anything and not much time either.

K.

I'll test the panel to see if I really did damage it.  If not, I might keep it for a transparent display idea.*  If so, then I might list the boards, or just drop the whole thing in the electronics recycling bin at a place where I volunteer.  I'm part of a crew there that moves donated furniture, and since we make regular trips with the truck anyway, one of our assignments is to take that bin to the recycler.  So I do know where it goes. 

---

*I've seen an amateur puppetry troupe try it with a big TV panel.  I think they were looking for a hologram sort of effect for a Sci-Fi story.  It sorta worked - enough to use it in the show - and also showed how dark the panel really is and how insanely bright the backlight (or stage lighting in their case) needs to be to get through it!

*I've also seen a dirt cheap video projector made from a bare panel on an overhead transparency projector.  It was okay for what it was: a typical transparency projector with a yellow lamp and bad keystone...

[AaronD]

I'll test the panel to see if I really did damage it.  If not, I might keep it for a transparent display idea...

Turns out it's no worse for wear, despite having caught a falling jerry-rigged desk lamp that also got me pretty good.  (it really needs to be a permanent fixture mounted properly, but I don't have that yet)

And despite the connections that this one has between the inverter and the graphics processor, it doesn't seem to care that the CCFL tubes have been removed completely and the transformer is running open-circuit except for the on-board caps.  The panel still does its thing just fine.  If I do use it with a different light source, I'll de-power the inverter, but I'll keep the option for now of giving it new tubes sometime in the future.