LED light teardown

Thread Starter

ErnieM

Joined Apr 24, 2011
8,377
FAILURE!

After just 4 or 5 days of continuous run time (120 hours max) I noticed one string of LEDs out yesterday. Funny, the night before I discovered if I had a bowl of water in the sink I could see all the LEDs lit up. Last night I saw one string of 3 was out.

120 hours is a bit short of the spec of 20,000 hours by about 99.4%

Brought it into work today to probe for the root cause, but didn't need to probe as the top of one of the LEDs had fallen off! The white ceramic and yellow mask were gone, all that was left was a thin substrate, no chip. I could make out where the wire bonds once were.

So it's a hard to call failure mode: was the LED weak from a manufacturing defect of was it just the first to go from overstress?

I could continue the life test, but at $9 a piece I'm going to exchange it and start a fresh life test.
 

iONic

Joined Nov 16, 2007
1,662
FAILURE!
So it's a hard to call failure mode: was the LED weak from a manufacturing defect of was it just the first to go from overstress?
Were you running the LEDs as designed or "over-stressing" them as you mentioned.

I'd like to place a temp prob on my array and check its front and back steady state temperature, then repeat the experiment with some form of heat sink.
Wonder what the safe operating temperature is of these little guys and what sort of heat-sinking or current reduction it would take to keep them in spec.
There is no datasheet I'm sure, but a general temp would be good to go by.
 

iONic

Joined Nov 16, 2007
1,662
They look vaguely similar to these:
Not the best datasheets but might give a general idea.
http://www.rapidonline.com/Electronic-Components/PLCC-2-High-brightness-surface-mount-LEDs-77736
Yes they do, very much so. Under the "Absolute Max. Ratings the high temp would be +85 C. That's 185 F.... heading towards the boiling temp! It would be interesting to see how close the array comes to that. Of course there are 24 of them packed practically right on top of each other. I could almost bet that the temperature is higher.

So given that the MAX is 85 C, what do you suspect they should be for an optimal life. ...as cool as possible, but would you suspect that say 55 C - 65 C
is reasonable for them to live long and prosper?

DataSheet
 

Thread Starter

ErnieM

Joined Apr 24, 2011
8,377
Were you running the LEDs as designed or "over-stressing" them as you mentioned.
They we being used in their as advertised application: as a replacement for an existing halogen light. This particular light points down, but the glass face was removed. So air is free to circulate, but first the hot air has to flow down.

For a voltage overstress I would expect one of the current regulators to go first, they are not part of the heat sink paths, and any voltage above what the LEDs require is dropped across them.

Lowes offered me a complete refund. For now I'm going to take my money and walk, see what EBAY brings me first.
 

KJ6EAD

Joined Apr 30, 2011
1,581
I can imagine the customer service people at Lowe's grumbling about the guy who returns stuff because it fails to deliver the required luminous flux and suffers catastrophic failure at 0.6% specified lifespan. :D
 

Thread Starter

ErnieM

Joined Apr 24, 2011
8,377
<grin@KJ6EAD>

Actually the longest part of the discussion was if I wanted the return put back on my debit card or just to get cash. My gut call was to say dump it back on the card, makes me whole again, but then I asked the cashier if it was better for them to just give me cash back as I'm sure my bank not only charged them to to initially withdraw the funds but then would charge them again to put it back.

She just shrugged and paid it to my card. Had she been MY employee she would have know what was best for her company and had the answer for the customer.

Lowes and Home Depot accept returns on anything. I've even bought at one and returned to the other, without receipts of course, for store credit. That was back when I was doing some semi-major plumbing and had bought all sorts of stuff I barely understood and of course bought extras and bought for aborted plans too. I still probably have $100 of left over fiddly bits in my shed.

(Did you know that a waste pipe when going from vertical to horizontal needs a different shape elbow then when going from horizontal to vertical? I do but I forget which case gets the "sweep.")
 

iONic

Joined Nov 16, 2007
1,662
I'm trying to find out what to use to monitor the array temperature with. Something simple and somewhat accurate.
 

Thread Starter

ErnieM

Joined Apr 24, 2011
8,377
I have a excellent Omega thermocouple meter here with a microscopically small bimetal tip probe, I wrapped the tip on one of my pins.

You don't have that meter and you don't have pins. Sucks to be you. :mad:

Other then an IR remote meter (yeah I got one of those too but it's a breadboard demo) anything you stick on it and tape down will be in the thermal path. I once knew an engineer who could just touch it and tell you the temp to a few degrees.

Got any of those nice nat semi temp IC's? Could just crazy glue one with a power supply and a DVM to get some information.

Or wait till my slow boat from China arrives and I'll measure mine for ya. :D
 

iONic

Joined Nov 16, 2007
1,662
I have a excellent Omega thermocouple meter here with a microscopically small bimetal tip probe, I wrapped the tip on one of my pins.

You don't have that meter and you don't have pins. Sucks to be you. :mad:

Other then an IR remote meter (yeah I got one of those too but it's a breadboard demo) anything you stick on it and tape down will be in the thermal path. I once knew an engineer who could just touch it and tell you the temp to a few degrees.

Got any of those nice nat semi temp IC's? Could just crazy glue one with a power supply and a DVM to get some information.

Or wait till my slow boat from China arrives and I'll measure mine for ya. :D
Oh, your just too funny ErnieM.

I do have a couple of LM35's!!! I can tape one (right...as if the tape would stay) to the front and one to the back....oh wait....i'll need 2 multimeter. I suppose you have that too!:eek:

...and this is only Gov't work here, work worry too much about the "thermal path". The objective, in the end, is to see what it would take to get one running in a way that will insure the desired long life.
 
Last edited:

Thread Starter

ErnieM

Joined Apr 24, 2011
8,377
Actually three. Handheld Wavetek, a nice HP 3468A and a good Keithley 179 bench meters. Plus 5 bench power supplies, one precision, one power (40V/25A). Oldie but goodie HP digital scope, DC electronic load, microscope, hot air and standard tip soldering stations. My Tektronics docking station just died so I lost my function generator & counter/timer plug in units till I get that fixed.

And yes, we do government work here too.
 

iONic

Joined Nov 16, 2007
1,662
ErnieM said:
Actually three. Handheld Wavetek, a nice HP 3468A and a good Keithley 179 bench meters. Plus 5 bench power supplies, one precision, one power (40V/25A). Oldie but goodie HP digital scope, DC electronic load, microscope, hot air and standard tip soldering stations. My Tektronics docking station just died so I lost my function generator & counter/timer plug in units till I get that fixed.
O.K. Tell me one more time so I can write this down... what was your address? When were you planning a vacation?...:D



I have the LM24(F) sensor, better at at 70 Deg F the DVM reading is .7V @ 185 Deg F the voltage will be 1.85V. Going to try to set it up so that I can just lay the LED array on top of one LM34 and then just lay another on top, or on the back-side of the PCB.

I'll take a reading every minute or so for the first 10 min, then every 5 min for the next 50 min. or until stable.............it it remains stable that is. If not I will disconnect at no more than 200 Deg F. Not trying to see when they will fail, but what it will take to make them reliable.
 

Thread Starter

ErnieM

Joined Apr 24, 2011
8,377
I set mine up and ignored it for an hour. Then made sure the thermocouple was still in pace and took a few readings.
 

Thread Starter

ErnieM

Joined Apr 24, 2011
8,377
Well how about this: after only an 11 day wait the slow boat from china got here and I now have in hand 5 pieces of the EBay G4 24-3528 SMD LED Pure white Light Bulb iONic mentioned. 3 are already lighting up my kitchen in the long term burn-in test. I have 2 more here with me at work.

iONic gave a good reverse engineered schematic so I'm not going to bother with that. I took mine, connected them to a handy Radio Shack 12VAC transformer and lit em up, with these results:

Condition: LED down, still air, suspended above bench. Thermocouple wrapped around device with tip between two LEDs center of board.

13.2 VAC, 262mAAC, 125 °C steady state, for a dissipation of 3.4W and a 28°C/W thermal impedance part to ambient.

Repeating for DC:
13.2 VDC, 173mADC, 95 °C steady state, for a dissipation of 2.2W and a 32°C/W thermal impedance part to ambient.

Interesting, I don't know why the DC read higher then the AC (I would expect the reverse).

Curiously, using a calibrated finger tip the bridge gets hotter then the LEDs.

In normal kitchen operation they have a good color, but seem not quite as bright as the $9 a piece Lowes units. At 1/4th the price that is quite acceptable.
 

tom66

Joined May 9, 2009
2,595
13.2VAC = 18.6Vp-p.

LEDs are not resistive loads, and AC power != DC power is to be expected.

The difference in currents is 1.51x, which is a little over the scaling factor of 1.41x.
 

iONic

Joined Nov 16, 2007
1,662
Well how about this: after only an 11 day wait the slow boat from china got here and I now have in hand 5 pieces of the EBay G4 24-3528 SMD LED Pure white Light Bulb iONic mentioned. 3 are already lighting up my kitchen in the long term burn-in test. I have 2 more here with me at work.

iONic gave a good reverse engineered schematic so I'm not going to bother with that. I took mine, connected them to a handy Radio Shack 12VAC transformer and lit em up, with these results:

Condition: LED down, still air, suspended above bench. Thermocouple wrapped around device with tip between two LEDs center of board.

13.2 VAC, 262mAAC, 125 °C steady state, for a dissipation of 3.4W and a 28°C/W thermal impedance part to ambient.

Repeating for DC:
13.2 VDC, 173mADC, 95 °C steady state, for a dissipation of 2.2W and a 32°C/W thermal impedance part to ambient.

Interesting, I don't know why the DC read higher then the AC (I would expect the reverse).

Curiously, using a calibrated finger tip the bridge gets hotter then the LEDs.

In normal kitchen operation they have a good color, but seem not quite as bright as the $9 a piece Lowes units. At 1/4th the price that is quite acceptable.
Well, you beat me to the temperature test! It will be interested to see what I get for temps. With your work setup, is it possible to reduce your voltage to 12.7VDC for a temperature test and then again to 12VDC for another. Since a datasheet on such LEDs (perhaps not identical) listed a max temp of 85C. I think we would want to keep it under that...maybe 75C or even 70C for a much longer life.

How log did it take for the arrays to reach steady state?

As tom66 said 13.2V, in a full wave rectification scheme, is above 18V and the temperature reflected this. This certainly compromised the life of the LED's. I think you would want 12VAC as the ebay listing mentions, but even that would be high (about 17V). I think 18V with half rectification would work.

It odd that you seem to think the ebay array is not as bright as the Lowes array, especially when the ebay list 2W and the Lowes array lists 1W. If this really is the case, the Lowes array operating cost would be half. There's some sort of real efficiency gain in them bulbs!

Now I gotta get going on my temperature test!





I made myself a pair of tweezers out of two LM34 Temperature sensors and clamped it to the top and bottom of the LED array. I then applies 12.0VDC to the array and LM34's. After 45min I was not getting much more that 100 Deg F in this completely open air setup. The difference between the top and bottom was negligible. I did notice that the LM34 attached to the LED side was not exactly flush at all times and held them tight with my fingers. didn't seem to make to much difference, maybe 10 Deg F. I did however on occasion have spikes that reaches near 140 Deg F. but still well shy of the 185 Deg F max spec on a similar LED. All-in-all, At the proper voltage 12.0V they performed well. I suspect that with a metal enclosure with a clear class lens, and some method of introducing air flow (separated from the top surface by 1/4 inch spacers and several drilled holes in the enclosure) the array might perform well. I mentioned metal housing as it too would act as a crude heat sink, absorbing heat from the inside air and dissipating it to the outside air.

ErnieM, I'm inclined to trust your temperature measurements more than mine and would be grateful if you could repeat your work test with DC voltages of 12.0V and 12.7V. The 13.2V DC test exceeded both the recommended voltage and temperature spec of the LED's, and this with no inclosure involved.
 
Last edited:

Thread Starter

ErnieM

Joined Apr 24, 2011
8,377
I can just see myself up underneath my kitchen cabinets trying to probe out what voltage is actually going to the bulbs. The lighting system is based on IKEAs 12V lighting products, with NAN lights driven by another of their systems transformers. "Other" transformers were intended to also drive the same 12V 10W lights, just more of them, and with terminations suitable for use with wiring "behind the walls." Hence I got a 12VAC transformer to test these at work as opposed to tearing out what I have at home.

Forgot to mention the leads of the EBay LEDs do not match the halogen bulbs exactly. They are slightly not as thick, they are square vs round, and they are too widely spaced, 0.2" vs about .16" But with a little wiggling and a firm shove they can be forced into the other lamp bases, though the original bulbs may now not work if I ever want to go back.

OK iONic, I ran some more temperature measurements at the other voltages:
Rich (BB code):
Volt    mAmps    Power     Temp°C    Temp°F

12.0    103    1.2W    65.5    151

12.7    142    1.8W    75    167
So either voltage is good for the max temp spec.

I'm considering making an AC 2 DC converter to drive mine, as I seem to be exceptionally sensitive to flashing LEDs. I can see which newer cars have blinky tail lights whenever I driving and I scan across traffic and some primitive portion of my brain flashes "MOVEMENT DANGER" where there is neither movement nor danger. It's very disconcerting as my top brain then has to find the source of the primitive message to thus ignore it.

In the kitchen I get the same message when working at the sink and I pick up on the trails of my hand in the flashes. I hate that.

I will add that my blinky LED product (it simulates the same intensity of an incandescent bulb it replaces for a 10-28V input) is running a PWM frequency of some 200Hz and I do not see the blinky effect on it. That wasn't by design, that was the slowest I could run the PWM off that hardware configuration (and had no room for any other parts).
 

colinb

Joined Jun 15, 2011
351
I'm considering making an AC 2 DC converter to drive mine, as I seem to be exceptionally sensitive to flashing LEDs. I can see which newer cars have blinky tail lights whenever I driving and I scan across traffic and some primitive portion of my brain flashes "MOVEMENT DANGER" where there is neither movement nor danger. It's very disconcerting as my top brain then has to find the source of the primitive message to thus ignore it.
I absolutely agree about the annoying PWM'ed LED lights on many newer cars: Cadillac and Nissan are two that specifically come to mind, but there are others. I also see the issue with the city buses here. It is similar to the color wheel rainbow effect with DLP projectors, in that the persistence of vision effect fails.

Why do only some people notice this irritating problem with LED taillights? I recall hearing some technical name for the condition, but I believe some people have eyes that are just more sensitive to such rapid events as these quick LED pulses. Or, maybe most people don't notice it because they aren't looking for it.

I am curious at what frequency these LED tail lights are modulated. I would guess around 100 Hz or so. I have thought about using a photodiode and a frequency counter to measure it, but I would have to walk right up to the car while it's running and since I don't own one, that's hard.

There is one easy method that I would like to use to measure the pulse frequency: Using a camera with manual exposure time control (e.g., a DSLR), take a photograph of the scene including the tail lights of the subject car. (The car should be relatively small in the frame to leave lots of space for panning; all you care about are seeing the red dots of the tail lights.) Before pressing the shutter release button to take the shot, begin panning the camera at a constant speed. Press the shutter release button and continue panning until the exposure is complete and the shutter closes. Make sure the tail lights of the subject car are in the frame the entire time. Then examine the captured image and count the number of tail light impressions captured. Divide this by the exposure time and you will have the modulation frequency of the tail lights.
 

iONic

Joined Nov 16, 2007
1,662
OK iONic, I ran some more temperature measurements at the other voltages:
Rich (BB code):
Volt    mAmps    Power     Temp°C    Temp°F

12.0    103    1.2W    65.5    151

12.7    142    1.8W    75    167
So either voltage is good for the max temp spec.
Nice! Thanks for the quick response too!
I re-ran my test again just on the LED side of the PCB and made darn sure the LM34 was flush against the LED's, it covered about 2 of them.



I got slightly higher temperatures, maxing out at 123 Deg F at 12V - .75V(diode Vdrop) or 11.25V
I guess we have to remember that the rectifier, mine at least, drops .75V.
That's why I asked for the 12.7V test, to see what the LED's would do if I removed the rectifier and applied a regulated 12V.

I also cranked up the voltage on my source to as high as I could get it (12.53V) and ran into a steady state temp of 133 Deg F.

My guess is that they would likely get too hot run at 12V in an inclosure unless there was a means to allow airflow. This leads me to believe dropping the effective voltage (without diode Vdrop) to 11.0V to 11.5V in an enclosure would work OK.

It will be interesting to overlay a plot of the LED array in an inclosure with the in open air.
 
Last edited:
Top