Backlight inverter failure modes

[(In)Sanity]

So I recently acquired a 60 inch Sharp LCD TV with CCFL back lighting for free which has a symptom common to the model of the top half of the screen flickering.  The back light in this area actually flickers.   It takes around 10 hours at the moment of being left on before this occurs.   

The unit has 24 tubes divided in to 4 banks of 6 bulbs each.   So in this case the top 6 bulbs flicker after 10 hours of use.   It has a total of 48 transformers (24 on each side)

Evidence has been found on an interconnect wire of excessive current.   One wire has clear heat damage.  It's one of two wires that feed the same circuit.   

Heat appears to be the trigger for this issue,  however I'm a bit puzzled as to what component would go for 10 hours and then start to act up.  The only capacitors in the circuit path which failed are ceramics that act as a DC block,  as in the transformers are capacitive coupled.   Could one or more of the high voltage transformers be dropping in resistance due to a short or just basic resistance drop due to heat ?  Could it be one or more of the ceramics?    This is not a one off failure,  many have reported the exact same issue with this model.   It's just a normal failure type,  no smoke and flames.

So what's a common failure mode for a back light inverter ?

Thanks,

Jeff

[tom66]

Wow, I'd like a free 60 inch to fix.

I'd investigate if the +24V going to the inverters is stable...

[(In)Sanity]

Wow, I'd like a free 60 inch to fix.

I'd investigate if the +24V going to the inverters is stable...

That's an idea,  but why after 10 hours of use and only for 1 of the 4 banks ?  Unless heat plus extra voltage is just enough to trip that top 6.   I'll check that.   

[SeanB]

Most likely the tubes are nearing EOL, and if you look at them they will have severe end blackening. Try swapping the top and bottom drive inverters and see if the problem moves.

[(In)Sanity]

Most likely the tubes are nearing EOL, and if you look at them they will have severe end blackening. Try swapping the top and bottom drive inverters and see if the problem moves.

The tubes only have about 1 month of run time on them if even that.   The design makes it impossible to swap the inverters.   Also it runs perfect for about 10 hours with no sign of flicker on any of the 6 top bulbs or dimming.   Then it will slowly start to flicker until it becomes unwatchable.  The bulbs never go solid black at any time,  they just cut out for a split second.   Also this is a common failure for this model and it's always the top 6 bulbs.   So I'm still going with an inverter issue.

[mazurov]

The easiest fix seems to be reducing watching TV for less than 10 hours a day. It will actually be good for you.

[(In)Sanity]

The easiest fix seems to be reducing watching TV for less than 10 hours a day. It will actually be good for you.

Haaa,   I actually don't watch TV much at all.   It's not about the 10 hour problem,   it's about having to put a 60 inch TV on the wall only to have it get progressively worse.   

[tom66]

Usually flickering bulbs are bad caps on inverter, bad caps on PSU, failing PFC circuit or failing main bulk capacitor.

[Psi]

In order to prove a bad inverter or a bad CFL you really do need to swap them around and see what happens.


One thing you could try is.. run it with the back panel off and when the problem occurs point a desk fan at the pcbs and see if the problem goes away.

I would also resolder all the joints on the inverter board in question, just in case there's a bad joint.
It only takes a few minutes and when the testing takes 10 hours you want to maximize the change of fixing it.

[aargee]

The wire that has heat damage, is the damage at the end near the connector? I'd be suspicious of the connector(s) and soldering - as it gets warmer, resistance rises, voltage critical circuits fail. We use 12V fed LCD monitors, have had high resistance connector issues cause failing after a few hours of operation, our solution was to bypass offending connectors with direct soldering of wires.

[(In)Sanity]

I found some warrantied inverter boards on ebay for cheap money,  so those are ordered.   Same exact part number, etc.   

Google image search for Sharp RDENC2663TPZZ will show you what I'm dealing with. 

The capacitors are Chemi-Con.   The FET's are Alpha Omega D4184's with OZ9982 dual FET drivers.   I'm almost wondering if the FET's aren't locking up or colliding with each other.   It would explain the burnt wire with no sign of any other stressed component.   

[(In)Sanity]

The wire that has heat damage, is the damage at the end near the connector? I'd be suspicious of the connector(s) and soldering - as it gets warmer, resistance rises, voltage critical circuits fail. We use 12V fed LCD monitors, have had high resistance connector issues cause failing after a few hours of operation, our solution was to bypass offending connectors with direct soldering of wires.

I suspected the same,  so I eliminated the board interconnects and hard wired everything.   While the flicker is now less pronounced it still remains.  I did this over the weekend.   Sorry that I failed to mention it.   

[(In)Sanity]

I just pulled the 11 transformers that make up the top half (22 total not 24).   I found one of them has a secondary inductance 4.5H lower then all the others.  The primary inductance is also lower.   The DC resistance and Q factor however are about the same.   I've got a couple other transformers that are a bit lower then average,  but not like this one. 

[amyk]

I found one of them has a secondary inductance 4.5H lower then all the others.

4.5H?

Either way, do the transformer check again with the unit still warm (but unplugged), immediately after the symptoms appear.

[tom66]

In addition, a cheap way of testing the transformers is to measure their secondary resistance. Significantly different values indicate a defective transformer.

[(In)Sanity]

I have two scopes and a DDS sig gen,  as well as the B&K LCR meter Dave reviewed.   In addition I have a home brew environmental chamber.   So I can either pass a signal through the transformer in question or just use the LCR meter while heating it.   

I look at it this way,  the transformer with the lower inductance is going to draw more current then the rest of them at the same frequency.   The others all come in around 12.5H- 13H @ 10 Khz where this one is just over 8H.    It also has oddly enough the highest DC resistance on the secondary.   If they just wound it short it should have a lower resistance and naturally lower inductance.  So that's a bit questionable as well.

[(In)Sanity]

Thought I would report back that this TV is now fixed.   Here is what I did to "fix" it.

Right side inverter had a burnt wire that interconnects the two halves of what should have been one board.   I found the whole assembly on ebay for $16 shipped.  Replacing this assembly technically did no more to fix the problem then just replacing the burnt wire.   It was improved,  but not 100% fixed.

Using the spare parts I now had I turned my attention to the left side inverter assembly.   I de-soldered the 22 transformers from the spare right side assembly as well as the left side assembly.  Using my B&K 879B LCR meter @ 10 Khz I then hand sorted out the pile of transformers keeping 22 with the highest secondary inductance value.   Doing a nice careful clean solder job I put those hand selected 22 transformers back on the left side inverter.   Cleaned up any flux residue and put everything back together.   

So it's now been running for 20 hours without a single sign of problems.   It's never achieved that before so I'm calling it fixed.   The best it's done in the past is 12 hours.   The ambient temperature is the same.   

I did find that 4 transformers in my pile of transformers were 20%+ out of spec from the average.   Moral of the story,   replace the non conforming transformers.   Based on the high percentage of this model that had the same exact problem I suspect it may have been a quality control issue with the supplier of the transformers.

Jeff