Salvaging LED downlights and back to back electrolytics.

[aargee]

Hi all,

I've been given a quantity of faulty 10W brand name downlights, the MR15 12VAC powered type. These designs use a micro fan to draw air over a diecast heatsink that attaches to the rear of the LED PCB. They all have failed fans, the LEDs still work but I imagine quickly over heat.
I had this great idea to mount four of the LED modules and their drivers onto a length of 6mm thick x 100 wide brushed aluminium channel and mount it as a workbench light to replace my ageing fluoro.
This works quite well, without the need for fans, the channel only getting slightly warm.

During my tests I noticed the 220uF/35V/105C electrolytic cap on each board getting hot, too hot to touch. The ESR is around 0.1, so that's OK. The CRO shows varying voltage from +8 to -8 volts stepping at about a 1 second rate, hardly any high frequency component. I think it's the negative swing that's heating it up and I think the design is the way it is for price and space constraints.

Question is, I'd like to increase the longevity of the drivers and maybe doing back to back electrolytics to make the capacitance bipolar, will this solve the problem? I was just going to solder another 220uF across it with opposite polarity, but this would double the capacitance (I don't think that's a critical design issue).

Thanks in advance...

[AndyC_772]

I don't like the back-to-back approach much. Could you substitute the type of bipolar electrolytic cap that's used in loudspeaker crossovers?

[Experimentonomen]

a bipolar cap is the exact same thing as two back to back lytics.

[peter.mitchell]

I'd probably investigate why there is so much negative swing on the cap.

[Crazy Ape]

Hi all,

I've been given a quantity of faulty 10W brand name downlights, the MR15 12VAC powered type. These designs use a micro fan to draw air over a diecast heatsink that attaches to the rear of the LED PCB. They all have failed fans, the LEDs still work but I imagine quickly over heat.
I had this great idea to mount four of the LED modules and their drivers onto a length of 6mm thick x 100 wide brushed aluminium channel and mount it as a workbench light to replace my ageing fluoro.
This works quite well, without the need for fans, the channel only getting slightly warm.

During my tests I noticed the 220uF/35V/105C electrolytic cap on each board getting hot, too hot to touch. The ESR is around 0.1, so that's OK. The CRO shows varying voltage from +8 to -8 volts stepping at about a 1 second rate, hardly any high frequency component. I think it's the negative swing that's heating it up and I think the design is the way it is for price and space constraints.

Question is, I'd like to increase the longevity of the drivers and maybe doing back to back electrolytics to make the capacitance bipolar, will this solve the problem? I was just going to solder another 220uF across it with opposite polarity, but this would double the capacitance (I don't think that's a critical design issue).

Thanks in advance...

Don't put the caps in reverse parallel like that, you'll just end up with two hot caps instead of one.
What you could do is use two caps of double the capacity in back to back series.


Info:
"If two, same-value, aluminum electrolytic capacitors are
connected in series with the positive terminals or the negative
terminals connected together, the resulting single capacitor is
a non-polar capacitor equal in capacitance to half of the rated
capacitance of either of the original pair. The two capacitors
rectify the applied voltage and act as if they had been partially
bypassed by diodes. However, at all levels of AC voltage the
apparent capacitance is half of the rated capacitance as you
would expect for capacitors connected in series."
http://www.digikey.com.au/Web Export/Supplier Content/CDE_338/PDF/CDE_AEappGuide.pdf

[amyk]

A 1Hz rate from -8 to +8 sounds very odd... what is the driver circuitry like?

[aargee]

The four lights I've dismantled all behave the same way, I'm not even sure of the caps main function here. I didn't really want to reverse engineer the circuit, it's very compact, SMD, double sided. I've attached images of the driver, LEDs and complete MR15 device. The series connected caps sounds like the way to go.

[SeanB]

The cap is most likely a mains reservoir. try replacing with a lower ESR one and see if the voltage problem and heating goes away.