Author Topic: LED panel driver failed, housing partially melted  (Read 8869 times)

0 Members and 1 Guest are viewing this topic.

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
LED panel driver failed, housing partially melted
« on: October 15, 2014, 05:48:49 am »
Hi everyone,

I'm not really looking to repair this, but am interested to see if anyone can suggest what could have caused it. At work we have about 30 100W LED panels which were put in when the building was renovated last year. After maybe a few months of light use, the panels started failing one by one. I think only about 5 are working now.

Our electrician removed two of the drivers and found that the housings on both had partially melted. He hasn't removed the rest (it's quite a hassle), but we assume they all suffered a similar fate.
I opened one up and found a burst cap and burnt-out diode. The hole in the housing is right above the diode.

Our mains voltage is about 252V. I can't see what the capacitor is rated for, unfortunately.

We contacted the lighting company who blamed the dimmers the electricians installed. The lighting company said we should have used a trailing edge dimmer. We used a Clipsal universal dimmer instead, after the lighting company said it "should be ok". Interestingly the housing says "Triac dimmable" which means leading edge should work anyway.

Can this sort of damage be caused by the wrong dimmer type, or the dimmer not putting out an acceptable waveform?

The only other things I can think of is the possibility that the roof insulation was sitting on the driver, causing it to overheat. They weren't switched on every day though.
Or could the drivers just have been faulty, or not well-designed?

Internal photos here: http://imgur.com/a/OIPGU


Thanks!

 

Offline sunnyhighway

  • Frequent Contributor
  • **
  • Posts: 276
  • Country: nl
Re: LED panel driver failed, housing partially melted
« Reply #1 on: October 15, 2014, 09:22:15 am »
Our mains voltage is about 252V.

Australia uses 230V as the nominal standard with a tolerance of +10%/?6%.
That would translate to 216.2V and 253V.
A measured mains voltage of 252V seems too close for comfort to me.
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #2 on: October 15, 2014, 09:31:00 am »
Our mains voltage is about 252V.

Australia uses 230V as the nominal standard with a tolerance of +10%/?6%.
That would translate to 216.2V and 253V.
A measured mains voltage of 252V seems too close for comfort to me.

It is pretty high... but still within spec I guess. The UPS here used to keep sending me extended high-voltage alerts until I changed the threshold to a higher voltage.

I took one of our spare LED panels and driver and plugged it straight into a wall (so no dimmer) and ran it 24/7 for a week, with the driver sitting on the carpet. The driver only felt warm, as you would expect. I should hook up a dimmer inline and see if it makes a difference.
 

Offline ovnr

  • Frequent Contributor
  • **
  • Posts: 658
  • Country: no
  • Lurker
Re: LED panel driver failed, housing partially melted
« Reply #3 on: October 15, 2014, 09:32:59 am »
Wow, that's somewhat unusual: A film cap that's spilled its guts.

This makes me think that the cap failed badly, which took out the diode. It really shouldn't be due to the somewhat excessive line voltage, but it obviously didn't help.

I haven't looked very much at the circuit, but it strikes me that it may be a resonant converter? And that the cap was woefully underrated for such duty, and thus the fault lies with the supplier of the panels & supplies.
 

Offline bktemp

  • Super Contributor
  • ***
  • Posts: 1623
  • Country: de
Re: LED panel driver failed, housing partially melted
« Reply #4 on: October 15, 2014, 09:45:27 am »
There seems to be no electrolytic input capacitor: Those two large cylinders seem to be hollow, therefore I think those are inductors for handling the voltage from a leading edge dimmer. That spilled cap seems to be the only capacitor after the rectifier. If it fails then there is absolutely no cap in there.
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #5 on: October 15, 2014, 09:46:52 am »
Wow, that's somewhat unusual: A film cap that's spilled its guts.

It does look quite interesting. The guts that spilled out is hard like glue.


Quote
I haven't looked very much at the circuit, but it strikes me that it may be a resonant converter? And that the cap was woefully underrated for such duty, and thus the fault lies with the supplier of the panels & supplies.

It's rated at 400V. Poor quality?
The electrolytics are Rubycon YXF. At least that's what the label says.
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #6 on: October 15, 2014, 09:51:15 am »
There seems to be no electrolytic input capacitor: Those two large cylinders seem to be hollow, therefore I think those are inductors for handling the voltage from a leading edge dimmer.

Yup, they look like inductors. They are well heatshrinked, but I can see a bobbin underneath. They're glued down too, presumably to keep them quiet.

Quote
That spilled cap seems to be the only capacitor after the rectifier. If it fails then there is absolutely no cap in there.

What could cause it to fail?

Thanks!
 

Offline ovnr

  • Frequent Contributor
  • **
  • Posts: 658
  • Country: no
  • Lurker
Re: LED panel driver failed, housing partially melted
« Reply #7 on: October 15, 2014, 10:01:22 am »
On the matter of no input electrolytics: The SMPS controller (L6562D) is a PFC controller, and isn't meant to have one.

Of course, that means the poor film cap gets a pretty hard job.
 

Offline bktemp

  • Super Contributor
  • ***
  • Posts: 1623
  • Country: de
Re: LED panel driver failed, housing partially melted
« Reply #8 on: October 15, 2014, 10:08:07 am »
Quote
That spilled cap seems to be the only capacitor after the rectifier. If it fails then there is absolutely no cap in there.

What could cause it to fail?
High supply voltage, high current, bad quality. Hard to say.
The life expectancy of a foil cap depends on the voltage applied: If the rated voltage is applied it can be rather short. Maybe the high input voltage and the repetive current spikes killed the cap. After a fast rise of the input voltage (due to the leading edge dimmer) the LC input circuit could produce a overoltage. Probably the cap was a bit underrated for the circuit both in its voltage and current rating.
My guess is, the cap failed and without any capacitance the voltage spikes burned the snubber diode.
The overall quality looks good. I have seen much worse power supplies.
 

Offline Kjelt

  • Super Contributor
  • ***
  • Posts: 5836
  • Country: nl
Re: LED panel driver failed, housing partially melted
« Reply #9 on: October 15, 2014, 10:34:10 am »
My guess is that they haven't tested this PSU properly under all conditions. If you say this and many others failed within a year while they were not even powered on each day then it should have been found in the testing phase of the product but not many companies continue to do this anymore.
This looks like an OEM bought chinese origin PSU (found it on baba for a few $ see picture) and the practice is often unfortunately that the customer does the real testing  :(
If it was up to me I sent them all back to the manufacturer asking for a decent supply esp. if you paid normal price. If they ship shit let them take the consequences or they will continue to sell it.

BTW: if your mains is really 252 VAC than you operated it outside the max. conditions and that could be taken as an argument from the seller. If this is really the case then the installer is to blame, he should have checked the conditions in the building with the fixtures before installing and should not have installed them in the first place.
 

Offline sunnyhighway

  • Frequent Contributor
  • **
  • Posts: 276
  • Country: nl
Re: LED panel driver failed, housing partially melted
« Reply #10 on: October 15, 2014, 12:43:39 pm »
Our electrician removed two of the drivers and found that the housings on both had partially melted. He hasn't removed the rest (it's quite a hassle), but we assume they all suffered a similar fate.

Sounds like ambient temperature might play a part here, or restricted airflow.

Can you confirm it adheres to the following 2 specs?

 - Thermal Dissipation:Nature
 - Operation Temperature:-30~45 degree
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #11 on: October 15, 2014, 12:45:22 pm »
High supply voltage, high current, bad quality. Hard to say.
The life expectancy of a foil cap depends on the voltage applied: If the rated voltage is applied it can be rather short. Maybe the high input voltage and the repetive current spikes killed the cap. After a fast rise of the input voltage (due to the leading edge dimmer) the LC input circuit could produce a overoltage. Probably the cap was a bit underrated for the circuit both in its voltage and current rating.
My guess is, the cap failed and without any capacitance the voltage spikes burned the snubber diode.
The overall quality looks good. I have seen much worse power supplies.

Hmm, the universal dimmer makes it a bit murky, as there is no indication as to what mode it runs in. It is meant to detect the type of load connected and switch between leading or trailing edge. I'll see if I can get more info tomorrow.

So one possibility is that the leading edge waveform and the resulting spikes killed the cap. That sounds like a driver design problem as it specifically says it's triac-dimmable. Is there any way a malformed waveform could have caused damage?

If the dimmer ran in trailing edge mode it should be OK then?

The quality inside does look decent. I was expecting worse before I opened it, especially after I saw the price on alibaba.


My guess is that they haven't tested this PSU properly under all conditions. If you say this and many others failed within a year while they were not even powered on each day then it should have been found in the testing phase of the product but not many companies continue to do this anymore.

Yup, only a handful remain working. Maybe I should leave one running for a month to see what happens.


Quote
This looks like an OEM bought chinese origin PSU (found it on baba for a few $ see picture) and the practice is often unfortunately that the customer does the real testing  :(

Yeah that looks like the one. The internal build looks pretty decent for that price, unless there's a hidden cost somewhere.

Quote
If it was up to me I sent them all back to the manufacturer asking for a decent supply esp. if you paid normal price. If they ship shit let them take the consequences or they will continue to sell it.

Ahh, that's where the fun begins. The lighting company has shut down. The office management are looking into the next course of action. Meanwhile I'm just trying to find out what could have happened.

Quote
BTW: if your mains is really 252 VAC than you operated it outside the max. conditions and that could be taken as an argument from the seller. If this is really the case then the installer is to blame, he should have checked the conditions in the building with the fixtures before installing and should not have installed them in the first place.

But 252V is still within spec. I would have thought anything sold here would have to be able to handle it.

Hmm, I just checked with the UPS remotely, the voltage is now 245V. I do remember seeing 252V in the past, but unfortunately the software doesn't log the actual voltage, it just logged the extended overvoltage and AVR trim events. I'll bring my meter in tomorrow and double-check.


If thats what I think it is a CRCM flyback, that little SMA diode is part of a RCD snubber and at 100W plus conversion efficency it will see large current spikes. From the datasheet L6561

I'm doing a CRCM flyback AT 30W and Ipk is 1.8A, if you don't accomodate for propagation and switching delays you could easily hit over 2A. For a 100W converter I would guess at least three times that, so that little diode is taking a shit kicking. Should have used something like this
You have to also watch ratings for those film caps not all are equal some have serious derating for temp and frequency for VAC, VDC and Irms.

Just read this after typing the reply above. So you suspect a driver that hasn't been properly designed, with an under-rated diode and cap?

The spare driver I tried seemed to run fine when I left it running for a week, but maybe that's not long enough for problems to crop up. There was no excessive heat or any noises. I didn't test it on a dimmer though. Could the addition of a dimmer cause the issues, or could it just be a bad batch of parts, or just a matter of time regardless of dimmer?


Thanks for the replies everyone!
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #12 on: October 15, 2014, 01:15:54 pm »

Sounds like ambient temperature might play a part here, or restricted airflow.
Can you confirm it adheres to the following 2 specs?
 - Thermal Dissipation:Nature
 - Operation Temperature:-30~45 degree

Not sure what it means by nature. I'm guessing non-active cooling?
30-45 degrees - I'm not sure, I'd have to stick a thermometer there.

Apart from the 100W panels, we also have a large number of 36W panels of the same make elsewhere in the building, with a different brand of drivers. None of those have failed. The LED downlights in the same areas as the failed panels are all still working too. None of the other lights are on dimmers though.


 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 15391
  • Country: za
Re: LED panel driver failed, housing partially melted
« Reply #13 on: October 15, 2014, 05:51:41 pm »
I would say the capacitor is not rated for the pulse current through it, and that poor diode is also very underrated as well for what it is supposed to handle. Probably a combination of poor selection, hot running and being dimmed to low levels. Worst case would be being dimmed to around half power, where every cycle you have a massive spike of close to 350V applied to the input. It does look like those input chokes are damped as each has a large fusible resistor in parallel with it, and there is a MOV which has not been cooked after them as well.
 

Offline bktemp

  • Super Contributor
  • ***
  • Posts: 1623
  • Country: de
Re: LED panel driver failed, housing partially melted
« Reply #14 on: October 15, 2014, 07:35:49 pm »
I do not think the snubber diode is largely underrated: The snubber has 3x 100k 1206 in parallel and a 1206 capacitor. The mosfet is 600V rated, therefore the cap has to be at least 250V rated, I doubt it has more than 10nF. The 0.3ohms current shunt gives a theoretical peak current of about 5A, but only a part of it and only during a very short period flows over the diode into the snubber, otherwise the capacitor needs to be much larger to absorb the energy. Even a small 1A diode should be able to handle those peak currents.
If the power supply uses DCM or CrCM, the voltage has already decreased when the mosfets turns on again and the snubber diode should already be in blocking state.
But why did the diode fail? I can only guess:
- The diode has high recovery losses because it is too slow
- The snubber capacitor failed short circuit due to overvoltage stress and therefore the diode had to conduct a lot more current
- Due to a high supply voltage the blocking voltage of the diode was exceeded and it acted as a zener diode, dissipating a lot of power

I think the theory that the power supply was not tested enough during development is quite plausible. Probably the combination of high input voltage + dimming could have caused those component failures. At least the main capacitor seems to be underrated because it has to handle several amps of ripple current.
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #15 on: October 16, 2014, 05:53:50 am »
OK I just hooked up the Clipsal 32E450UDM universal dimmer inline with one of the spare panels and driver. The transformer in the driver buzzes quite loudly across the range, and is loudest at the lowest setting. I'll leave it running for a few days and see what happens.
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #16 on: October 17, 2014, 01:54:24 am »
It has been running for nearly a day now at minimum brightness. The buzz is really annoying, but the case temperature is still normal. I'm going to look for some insulation from the ceiling to cover it as a worst-case scenario.
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 15391
  • Country: za
Re: LED panel driver failed, housing partially melted
« Reply #17 on: October 17, 2014, 07:51:05 pm »
Half brightness will stress it most, and just place it in a fire blanket to simulate the roof insulation with no fire risk.
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #18 on: July 30, 2015, 05:11:05 am »
OK it's been ages so I thought I should post an update. After lots of time wasted talking to Clipsal and the building's project manager, we've decided to just replace the drivers at our cost.

I bought 25 MeanWell HLG-150H-42A drivers. I went for the 150W model just for some headroom, and adjusted it to run the 100W panels at about 80W.

4 have been replaced and they seem to be working fine. Our electrician is coming in tomorrow to do more.

Hopefully the MeanWells last longer!
 

Offline bookaboo

  • Frequent Contributor
  • **
  • Posts: 607
  • Country: ie
Re: LED panel driver failed, housing partially melted
« Reply #19 on: July 30, 2015, 06:45:02 am »
The meanwells are a decent driver alright, how are you doing the dimming though?
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #20 on: July 30, 2015, 06:51:56 am »
The meanwells are a decent driver alright, how are you doing the dimming though?

We've disconnected all the dimmers. I think dimming via a wall dimmer is just a problematic way to go. Next time I'll make sure they run wiring for 0-10V control at the very least. I'm surprised the lighting consultant didn't insist on that!

Actually I need to double-check to make sure all the dimmers have been disconnected. When I tested one of the new drivers with one of the original sockets in the ceiling, the light panel started blinking intermittently even though all the (supposedly disconnected) dimmers were turned to max. The dimmers and switches are installed in a large 20-gang B-style metal plate which is a pain to open up, so I'll just wait till we pull down another panel tomorrow and check the waveform. Fortunately I was able to use a nearby downlight's power socket!


« Last Edit: July 30, 2015, 06:54:36 am by eug »
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #21 on: July 31, 2015, 04:51:47 am »
BTW I'm look at replacing another few drivers for some LED strips in a hall.

There are 3 strips per side of the hall which take 12V ~4.2A each. I was thinking of replacing the 3 independent drivers per side with 1 high-current power supply.

Is there much benefit in getting one with a lot more overhead? Will they last longer in reality?

I'm deciding between a 22A and 16A one. Three strips will take up about 12.6A, so I'd be running one power supply at ~60% load and the other at ~80% load.

The old drivers were 12V 12.5A per strip. When switched off, the strip would blink a few times before going dark. Would that be because the load is too low?

Thanks!
 

Offline coppice

  • Super Contributor
  • ***
  • Posts: 5525
  • Country: gb
Re: LED panel driver failed, housing partially melted
« Reply #22 on: July 31, 2015, 05:12:22 am »
Is that film capacitor an inlet filter straight across the mains? If so, that supply was never intended for use with a dimmer. The massive higher harmonics you get in the highly distorted waveform coming out of a dimmer puts huge currents through that kind of inlet capacitor, and will usually make it melt. Those inlet cap designs assume a sine wave mains of about the right frequency. A square wave output from a crude UPS can do bad things to them, too.
 

Offline bookaboo

  • Frequent Contributor
  • **
  • Posts: 607
  • Country: ie
Re: LED panel driver failed, housing partially melted
« Reply #23 on: July 31, 2015, 08:29:30 am »
BTW I'm look at replacing another few drivers for some LED strips in a hall.

There are 3 strips per side of the hall which take 12V ~4.2A each. I was thinking of replacing the 3 independent drivers per side with 1 high-current power supply.

Is there much benefit in getting one with a lot more overhead? Will they last longer in reality?

I'm deciding between a 22A and 16A one. Three strips will take up about 12.6A, so I'd be running one power supply at ~60% load and the other at ~80% load.

The old drivers were 12V 12.5A per strip. When switched off, the strip would blink a few times before going dark. Would that be because the load is too low?

Thanks!

It depends on cost etc but I do like to over rate drivers. Of course you need to make sure the drivers are good quality in the first place, poor quality LED drivers are the falsest of false economies.
The other thing to consider is ambient temperature and heat dissipation, this can affect the lifetime of drivers by orders of magnitude.

With regard to triac dimmmers used with LED drivers I think this is match made in hell, I'm currently helping two seperate customers sort out similar messes (constant repeat failures) using this set up. In both cases the kit they installed are quality brands so it's not just the old "wunhunglow" problem (well probably not). Thankfully for me they bought the kit elsewhere and I'm acting as a 3rd party trying to help resolve the mess. In short I'd never ever use such a system.
 

Offline eug

  • Contributor
  • Posts: 21
  • Country: au
Re: LED panel driver failed, housing partially melted
« Reply #24 on: July 31, 2015, 08:36:31 am »
It depends on cost etc but I do like to over rate drivers. Of course you need to make sure the drivers are good quality in the first place, poor quality LED drivers are the falsest of false economies.

Out of the 6 drivers that the original installer put in, 3 have failed. They're generic unbranded ones. That's a pretty poor failure rate!

I'm looking at replacing them with MeanWell HLG-320H-12 or HLG-240H-12 ones. They've got a pretty good rep so should last quite a bit longer. They cost difference is small enough as I only need to get two of them.

My main concern is whether or not having too much headroom would result in 2-3 seconds of flickering when switched off like the original drivers.


Quote
The other thing to consider is ambient temperature and heat dissipation, this can affect the lifetime of drivers by orders of magnitude.

Some of the original drivers I mentioned in the first post here were placed in between the ceiling plasterboard and some insulation! That probably contributed to the short lifespan.


Quote
With regard to triac dimmmers used with LED drivers I think this is match made in hell, I'm currently helping two seperate customers sort out similar messes (constant repeat failures) using this set up. In both cases the kit they installed are quality brands so it's not just the old "wunhunglow" problem (well probably not). Thankfully for me they bought the kit elsewhere and I'm acting as a 3rd party trying to help resolve the mess. In short I'd never ever use such a system.

I think I know how you feel. :)
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf