EEVblog #725 - LG Plasma TV Teardown

[DanielS]

OTOH I think he should have tried swapping those caps just to see what happens. Where's the sense of curiosity?

No need to replace them in a (mostly) working device: put a scope probe on the output rails and see how clean they are. If there is only 100mV of ripple on the 5V rail, then you know without having to swap caps that caps on the 5V rail are most likely still good enough regardless of how bad they might physically look and you are better off looking elsewhere for more likely suspects.

I recently repaired a 24" LG LCD. The caps on the 12V rail had leaked and were from the infamous Samwha WB-series, but there was only 100mV of ripple on it. I replaced them anyway with Panasonic FM which have significantly better specs just to spare myself the trouble of potentially having to re-repair it because of them later but on the scope, this made practically no difference despite doubling the total nominal capacitance on top of that. Capacitors on the 24V rail were in far worse shape with 1Vpp of ripple and wild fluctuations (3Vpp if you include switching noise peaks) that went down to a much more steady 200mVpp (2Vpp including switching peaks) after putting new caps in.

My point: if you have a (kind of) working device, measure before blindly replacing components - spare yourself the trouble and cost of replacing parts that most likely have nothing to do with the immediate issues until you have identified repairable high probability suspects. If you do not find any such smoking gun, you may still choose to try blindly replacing caps afterward but at least you do so knowing there is a fair probability it won't help much, if at all.

[tom66]

In the case of a COF sitting on the flat flex for the X address electrodes, it seems to be common to heatsink these chips by using the metal bar covering the X address boards as a big heat sink. These COF are then pressed against this metal bar with thermal paste in between, heatsinking the COF driver ICs.

I don't know the reason why many plasma TVs are designed like this. Wouldn't it be better to have the driver ICs mounted directly on the X address boards (with heat sinks, of course), and just have a plain flat flex cable going to the X electrodes in the plasma display panel? It would certainly be a lot easier to service these TV sets if that was the case, but there might be some design complications involved there which I don't fully understand, perhaps...

The main issue is the  number of connections. A 1080p display will have 5,760 address driver outputs. For a screen about 45" wide (~50" diagonal) that's going to be 0.2mm between each address driver output. Difficult to do that with conventional PCB connectors and layouts. So the PCB is fairly cheap, and the TCPs themselves have the complex layout density. I guess it's cheaper to do that.

The heatsinking that manufacturers use has been reducing year on year as manufacturers pushed for improved efficiency. Modern address drive forms less than 30W of a PDP. Problem is jumping to 4K would double that, and also add ~50% more sustain losses, already a push for 1080p displays with lower brightness trying to meet energy regulations.

[haxtormoogle]

Moment I saw the caps on the low voltage power supply I was saying to myself "YES fix them caps dave! FIX them!!!!" Then the rest of the video was pure torture waiting for him to mention them, By the end of the video I was yelling at my computer! The caps Dave! The CAPS!!! Why didn't he attempt to fix it?!? What if he replaced the caps?!? 

[Pentium100]

They still make Plasma TVs and I bought one (LG 150cm) a three months ago. The reason for buying a plasma instead of LCD was this:
1. The plasma TV was cheaper than equivalent size LCD
2. The black levels are better.
The black levels are not that great, but I read on a forum that it is possible to tweak the voltages to get the black level down - something that I will do after the warranty expires. It also has the white level power limiting, but I guess nothing can be done about that (since that's more likely to be done in software so getting a bigger power supply would not change anything).

The TV actually does not use a lot of power, 150W or about the same as a LCD of the same size. However, I do not really care about the power - as long as it does not trip the breaker I'm fine, especially since I watch that TV probably an hour a week on average. I care a bit more about the power consumption of the devices I keep on all the time (servers, bitcoin miners probably use 1.5kW continuous).