LED driver - lamp flickering

[vinicius.jlantunes]

Hey folks, wondering if you have any suggestions of what I should I try next.

I have this osram led lamp which started flickering.

I opened it up and replaced the electrolytic caps just in case, didn't help.

I saw Daves video where he had bad joints in his LED drivers and just in case also touched up most joints, also no success.

I did not try replacing the other caps since I don't have replacements at hand. I did a quick search for similar drivers using the descriptions on the board but no luck (I wouldn't mind buying new drivers off eBay or similar). I don't know the nominal output voltage and current.

Any hints?






Sent from my iPad using Tapatalk

[shakalnokturn]

High value resistors then the PWM itself whatever that is.
I have several Osram E27 LED bulbs, they are the worst around, even less reliable than the cheap PRC stuff, I haven't investigated the fault (flickering and other intermittent failures) as they are potted. Possibly Osram has used a poor or badly implemented PWM IC over several products.

A bad LED in the string can cause flickering also, try ruling that out with a bench supply set at similar current to normal use.

[vinicius.jlantunes]

Thanks! I checked the resistors and they seem all fine.

I also tried switching one other cap right after the rectifier also to no avail.

I can measure the voltage at the output of such rectifier oscillating quite a bit around 160 V (AC input is 127 V).

I have another lamp like this one so I took it out to compare and the good one does not show the same behavior so there is something fishy around the input I believe.

Will get back to it later.

By the way, it is a 17W driver, output on the good one is 42 V at ~400 mA.
The output voltage at the bad one seems solid at ~35 V (at least on a multimeter, have not scoped it) but current is limited at ~40 mA.



Sent from my iPad using Tapatalk

[magic]

I don't think much can go wrong with such mains input
Can you post a closeup of the middle part of the board, with U1 and all that tiny stuff?
I see no obvious current regulation here and if it's a constant voltage regulator, minor drift in output can easily cause significant change in brightness or blinking if the converter enters burst mode.

[vinicius.jlantunes]

Here it goes, hope this picture is better, tried an angle that makes the markings readable

I think it reads:
3670-00
R29GG
C81320


Sent from my iPad using Tapatalk

[magic]

I don't know what it is but it looks like constant voltage regulators used in USB chargers.
Right side top to bottom: VCC, MOSFET drive, current sense, ground
Left side bottom to top: some clock/delay generator, startup supply, 2x voltage feedback from AUX transformer winding (these pins are shorted together, right?)

It would be the first time I see a chip capable of regulating secondary current solely by sensing AUX winding, so perhaps it really is just a constant voltage regulator? Then the voltage is set by R6/R7. Check if they are identical on the two boards, check if voltage across C2 is identical.

I can measure the voltage at the output of such rectifier oscillating quite a bit around 160 V (AC input is 127 V).
I have another lamp like this one so I took it out to compare and the good one does not show the same behavior

So what's the voltage on the other PSU? Lower or higher?

The output voltage at the bad one seems solid at ~35 V (at least on a multimeter, have not scoped it) but current is limited at ~40 mA.

Is it 40mA with the normal LED load or 40mA even if you connect more load to it?

[shakalnokturn]

There is no secondary side current sense. But you can assume the LED string has a relatively constant voltage, the primary side auxiliary winding can reflect the secondary voltage close enough and be used to limit it in the case of an open string.
Once you know the secondary voltage and approximate efficiency you can achieve something close enough to constant current using primary voltage, duty-cycle and primary current sense.
Is RS3 the right value? I have doubts on it's condition in the photo.

[mikerj]

I had a couple of LED lamps start flickering due to a bad LED.

[vinicius.jlantunes]

Thanks guys!

So the other day I measured a voltage of ~42 V on the good lamp - it is a 17 W lamp so the current is about 400 mA.

That equals about 100 ohms of load at such voltage, so I put such resistor across the bad driver to see what happens.
It puts out only 0,3 V then.

Interestingly this same bad driver puts out 35 V when plugged into the led strip.

I then put a 330 ohms resistor I had laying around and it puts out 0,9 V.

I measured about 2,6 mA with 100 ohms and 3,3 mA at 330 ohms. It's like it's trying to reach the 400 mA but can't at such loads?

Finally I did 1k ohm - it manages 10 mA / to 10 V.

I don't think the problem is any of the LEDs on the strip because it works fine when plugged into the other good driver.

I scoped the output with the 1k ohm Load:


Then with the LED Strip:


Doesn't show the flicker at least voltage wise! But it's very noticeable, not sure if I did something wrong.

I tried putting a 2 ohm resistor in series and take a waveform of the current but it seemed like a big mess... Can't be right.


Sent from my iPad using Tapatalk

[vinicius.jlantunes]

I will take out the other lamp maybe tomorrow to do some more comparison. It is installed in my kitchen so it's a bit annoying to remove it on weekdays after work...


Sent from my iPad using Tapatalk

[vinicius.jlantunes]

There is no secondary side current sense. But you can assume the LED string has a relatively constant voltage, the primary side auxiliary winding can reflect the secondary voltage close enough and be used to limit it in the case of an open string.
Once you know the secondary voltage and approximate efficiency you can achieve something close enough to constant current using primary voltage, duty-cycle and primary current sense.
Is RS3 the right value? I have doubts on it's condition in the photo.

RS3 reads 8R0 something, really hard to see with the little magnification I have (and it seems to be chipped).

I measured it as 1,2 ohms in circuit (I know I should remove but don't have time right now :-(


Sent from my iPad using Tapatalk

[shakalnokturn]

I'd be curious to the power the good driver will accept to throw into your load resistors.
Maybe something like 36-40V worth of zener voltage followed by a power transistor and resistor would make a more appropriate dummy load.
Anyway at this point I'd probably be swapping the PWM IC's over between good and bad driver boards.

[magic]

RS3 reads 8R0 something, really hard to see with the little magnification I have (and it seems to be chipped).

I measured it as 1,2 ohms in circuit (I know I should remove but don't have time right now :-(

That's OK, it seems paralleled with the two 2R7 resistors, which gives 1,16Ω indeed.

Beware that a normal 100Ω resistor would get fried pretty fast by 17W of power.

A 1kΩ (2W) or 10kΩ (0.25W) may be worth trying on the good PSU for comparison, to rule out the possibility that this is some very smart and sophisticated LED driver which detects that something is wrong with the LEDs themselves or that it's only a resistor. I doubt it, though. Probably what happens is the LEDs just draw almost no current until enough voltage is reached, which allows the faulty PSU to reach 35V.
Nevermind, I missed that the same LEDs work with the good PSU so this PSU is bad for certain. Yeah, I think it's either input voltage collapsing under load for mysterious reasons or bad controller chip as somebody suggested.

[David_AVD]

I couldn't see a clear answer to whether or not the LED string had been tested seperately?

[shakalnokturn]

I couldn't see a clear answer to whether or not the LED string had been tested seperately?

....

I don't think the problem is any of the LEDs on the strip because it works fine when plugged into the other good driver.

....

[magic]

I also missed that the LEDs are good.

It looks like C7 may be cracked. This would certainly be a problem. Check R10 too.

With resistors as load, test voltage across the big electrolytic capacitor and C2. Make sure the electrolytic you installed is low ESR or put back the original one. Maybe there is some problem here, maybe you were right that input voltage collapses under load for some reason.