Caps still failing in LCD monitors, but maybe not due to the caps themselves?

[Th3_uN1Qu3]

Hey all, i've serviced a 19" LG monitor a couple weeks ago. Power supply would go into protect mode when connecting the panel, CCFLs would run just fine if plugged in on their own.

Of course it was due to bad caps, i managed to start it up once with its original caps after resoldering some connections, and i got some ripple measurements, then it wouldn't start up at all. I replaced all caps except the one after the AC rectifier which was fine. I can't recall the ripple numbers right now but the most notable change was that it went from "could be trouble" to "insignificant" on the 5v rail. If i remember right, after the recap it was down to less than 5mV p-p. I also added some heatsinks to a couple DIP-8 mosfets that had visibly browned the board, then called the owner to pick it up.

I'm a novice in SMPS work so i was asking some stuff over at DIYAudio forums, specifically it was about the wire size (thickness) per Amp of current, as the book i read (Switching Power Supply Design, Third Edition) asked for a whole lot of wire compared to what i've seen in practice. I wanted to use an EI-33 core from an ATX power supply for 300W output at 70-0-70v, and i was wondering how the heck would all that wire fit inside. As people had obtained 200W+ from that core before, i knew they used thinner wire so i asked for a formula. I got a reply that said:

"I use 7.75A/mm^2 (...) SMPS transformers get away with a lot of current density because of direct airflow and consumer grade product lifetimes."

That instantly reminded me of that monitor. Consumer grade product lifetimes... In the service manual MTBF was rated for 50k hours, and the monitor was exactly 3 years old, manufactured July 2007. That doesn't quite line up, does it? Anyway, before i put the monitor back together i noticed that the switching transformer was getting really really hot (where's the "direct airflow" in a LCD???). And those capacitors are so close to the transformer... Things that aren't said about 105C capacitors is that they are rated for just 2.5k hours at that temperature. I believe you already know that here but most of the people don't, and i believe that some power supply engineers don't know that either. The board in that monitor could have easily accomodated a layout change. Either that, or they are engineered for failure in the first place, "consumer products". Then why the heck do they advertise that impossible MTBF number, and particularly, why in the SERVICE MANUAL? It seems silly to me.

[PlainBill]

Re: Caps still failing in LCD monitors, but maybe not due to the caps themselves?

Hey all, i've serviced a 19" LG monitor a couple weeks ago. Power supply would go into protect mode when connecting the panel, CCFLs would run just fine if plugged in on their own.

Of course it was due to bad caps, i managed to start it up once with its original caps after resoldering some connections, and i got some ripple measurements, then it wouldn't start up at all. I replaced all caps except the one after the AC rectifier which was fine. I can't recall the ripple numbers right now but the most notable change was that it went from "could be trouble" to "insignificant" on the 5v rail. If i remember right, after the recap it was down to less than 5mV p-p. I also added some heatsinks to a couple DIP-8 mosfets that had visibly browned the board, then called the owner to pick it up.

I'm a novice in SMPS work so i was asking some stuff over at DIYAudio forums, specifically it was about the wire size (thickness) per Amp of current, as the book i read (Switching Power Supply Design, Third Edition) asked for a whole lot of wire compared to what i've seen in practice. I wanted to use an EI-33 core from an ATX power supply for 300W output at 70-0-70v, and i was wondering how the heck would all that wire fit inside. As people had obtained 200W+ from that core before, i knew they used thinner wire so i asked for a formula. I got a reply that said:

"I use 7.75A/mm^2 (...) SMPS transformers get away with a lot of current density because of direct airflow and consumer grade product lifetimes."

That instantly reminded me of that monitor. Consumer grade product lifetimes... In the service manual MTBF was rated for 50k hours, and the monitor was exactly 3 years old, manufactured July 2007. That doesn't quite line up, does it? Anyway, before i put the monitor back together i noticed that the switching transformer was getting really really hot (where's the "direct airflow" in a LCD???). And those capacitors are so close to the transformer... Things that aren't said about 105C capacitors is that they are rated for just 2.5k hours at that temperature. I believe you already know that here but most of the people don't, and i believe that some power supply engineers don't know that either. The board in that monitor could have easily accomodated a layout change. Either that, or they are engineered for failure in the first place, "consumer products". Then why the heck do they advertise that impossible MTBF number, and particularly, why in the SERVICE MANUAL? It seems silly to me.

Certainly many monitors could have been designed better. Placing capacitors under heat sinks is foolish. Allowing for a free air flow is also helpful. Maintaining low impedance requires short traces, which requires capacitors must be in close proximity to the rectifiers, therefor to the heat sinks. However, the evidence that certain brands of caps have very high failure rates, often within the first year, while others last 5 years or more would indicate that the quality of the capacitors has a major effect on the reliability of monitors.

PlainBill

[Th3_uN1Qu3]

Re: Caps still failing in LCD monitors, but maybe not due to the caps themselves?

Maintaining low impedance requires short traces, which requires capacitors must be in close proximity to the rectifiers, therefor to the heat sinks.

Or thick traces... And of course, thick traces cost $$$.

However, the evidence that certain brands of caps have very high failure rates, often within the first year, while others last 5 years or more would indicate that the quality of the capacitors has a major effect on the reliability of monitors.

The caps in that monitor were all made by Samxon, which is a brand that this site stocks as well, so it's gotta be a reputable brand. Yet they were busted. More specifically, there was a glaring ommision as opposed to the schematic - a 1000uF/16 volt capacitor instead of a 1000uF/25 volt one. If you'd ask me, i think that's where trouble started.

[PlainBill]

Re: Caps still failing in LCD monitors, but maybe not due to the caps themselves?

Or thick traces... And of course, thick traces cost $$$.



The caps in that monitor were all made by Samxon, which is a brand that this site stocks as well, so it's gotta be a reputable brand. Yet they were busted. More specifically, there was a glaring ommision as opposed to the schematic - a 1000uF/16 volt capacitor instead of a 1000uF/25 volt one. If you'd ask me, i think that's where trouble started.

The quality of a capacitor is not determined by the distributors that stock it, rather the opposite is true. Mouser has begun stocking a known crap brand (Lelon?, Elite?). Also, while it is rare, a crap manufacturer can decide to do it right and start producing quality parts, and even quality manufacturers will produce a defective part now and then.

The parts called out in a schematic do not necessarily reflect what is in the BOM; engineering changes rarely make it into the schematics found in service manuals. And for a 5 volt supply, a 16 volt cap is more than adequate.

Once again, you are falling for the myth that bigger is better. Smarter people know that it's the quality of workmanship that counts.

PlainBill

[Th3_uN1Qu3]

Re: Caps still failing in LCD monitors, but maybe not due to the caps themselves?

The parts called out in a schematic do not necessarily reflect what is in the BOM; engineering changes rarely make it into the schematics found in service manuals. And for a 5 volt supply, a 16 volt cap is more than adequate.

That cap was on the 12v rail which ran at 12.70 all the time. That, along with high temperature operation, made the 16v rating a bit uncomfortable for me. Anyway, we'll see when it'll come back for servicing. Let's hope that it'll be at least another 3 years, shall we?

[PlainBill]

Re: Caps still failing in LCD monitors, but maybe not due to the caps themselves?

That cap was on the 12v rail which ran at 12.70 all the time. That, along with high temperature operation, made the 16v rating a bit uncomfortable for me. Anyway, we'll see when it'll come back for servicing. Let's hope that it'll be at least another 3 years, shall we?

Well, a argument could be made that a 16 volt cap allows enough 'headroom', but I'd have to agree with you that a 25 volt cap would be a better choice. In many cases the 12 volt rail is not regulated, so voltages over what you observed are certainly possible.

PlainBill