Falling Vf with usage?

[HDS_Systems]

A few years ago I saw references to someone reportedly testing some LEDs and commenting that over an extended period they observed a significant decrease in Vf. I have not been able to locate the thread(s) that discussed the issue. What I would like to know is who ran the tests, how the tests were conducted and what observations were made.

I have performed a similar set of tests at elevated temperatures with two different major brands of high power (1 - 3W) LEDs and have not observed any change in Vf in over 750 hours of continuous operation. I am using a precision bench power supply to drive the LEDs and I can measure Vf accurately to 3 decimal places using a separate volt meter and the kelvin 4 lead method - more than enough to see changes in Vf due to slight variations in temperature. I can also measure the LED's temperature to less than 1 degree C so I can easily recreate a standard environment for taking measurements.

Henry.

 

[evan9162]

http://candlepowerforums.com/vb/showthread.php?t=60217

The Vf drop has only been observed on Luxeon, III, V, and K2 products. It has only been observed with White LEDs, but is probably a characteristic of all InGaN products in those lines. Red/Orange/Amber luxeons don't seem to exhibit this behavior.

It has not been seen with LEDs from Cree or Seoul.

 

[SemiMan]

A few years ago I saw references to someone reportedly testing some LEDs and commenting that over an extended period they observed a significant decrease in Vf. I have not been able to locate the thread(s) that discussed the issue. What I would like to know is who ran the tests, how the tests were conducted and what observations were made.

I have performed a similar set of tests at elevated temperatures with two different major brands of high power (1 - 3W) LEDs and have not observed any change in Vf in over 750 hours of continuous operation. I am using a precision bench power supply to drive the LEDs and I can measure Vf accurately to 3 decimal places using a separate volt meter and the kelvin 4 lead method - more than enough to see changes in Vf due to slight variations in temperature. I can also measure the LED's temperature to less than 1 degree C so I can easily recreate a standard environment for taking measurements.

Henry.

It is right in the Luxeon K2 with TFFC data sheet that I have in front of me. It states that Vf may drop up to 1/2 volt.

Semiman

 

[wasBlinded]

I was able to observe a drop in Vf of just under a couple tenths of a volt over a 24 hour burn in (with a bench power supply) of each of my two U2 heads, FWIW.

 

[VidPro]

Henry, what is the drive current your using? and the quality of the heat removal?
i suspect that it has more to do with how its driven, in total, including any pulses pwm surges, spikes, overdrives, and heat , and if there is break down on items in the die.

my theory would be that in flashlights like what gets used, it happens, because of lesser heat syncing capable, pulses, and even overdrives from battery curcuit led combos.

we do some evil Voltage driving here (instead of current) and the change factor seems to be relative to the damage factor.

 

[HDS_Systems]

Evan9162,

Thanks.

VidPro,

I ran 500mA though LEDs in series so I could test more than one LED at a time. I mounted each LED in a head in normal production style and brought out leads for power and voltage measurement. The free-standing case temperatures stabilized at 44 degrees C after 10 minutes and varied slightly (+/- 1) with variations in the room temperature. By leaning a bag against the cases I was able to modulate the temperature slightly so I could take voltage readings at a consistent case temperature.

The test is still running but so far the Vf has actually risen by a couple of mV but this is still within the tolerance of measurement errors in temperature (mV/°C) and current resolution (1mA).

I have also worked extensively with Nichia LEDs and have never observed a Vf drop during my extended tests.

Henry.

 

[VidPro]

Evan9162,


I have also worked extensively with Nichia LEDs and have never observed a Vf drop during my extended tests.

Henry.

thanks for the info. i am always interested in end results.

the Nichita leds in the 5mm are tough little monkies, dont know about the big ones, i only have 1 lo-moon thing running, with a jupiter in it?
there are also charts and graphs in CPF that show that the nichita 5mm handles overdrive better than the cheap stuff. because they are both in the same encasements (heat), i wonder if it has to do with precision. (i wonder a lot of stuff)

 

[Gryloc]

I just wanted to check, but do you mean a drop in Vf over the entire life of the LED, or the very first 6hrs, 24hrs, etc? I noticed some Vf drop in the K2s in the first few hours of operation. I like to burn in LEDs just so I know what Vf the LED will operate at in the intended operation. I usually just burn in my LEDs at 350mA for a few hours. Of course, even though I use some decent current regulation and I make sure the temperatures stabilize, I haven't been doing this scientifically enough because I have not been writing any of these measurements down.

You say that this is only for Lumileds products, but I thought that I observed something when burning in some Seoul P4 emitters (but I do not have any data to share now for the above reason).

I want to monitor a Rebel 0100 (when I can find another) and one of my new K2 0180 emitters to see what kind of drop I can expect. If the new K2 may have a drop in Vf in the order of 0.5V, that may fix the high Vf problem. With what current level did they noticed that 0.5V reduction in Vf? It is too bad that my method of doing Vf-I curves is a pain in the butt (current regulating driver with adjustable current via external potentiometer). I will try to post results as I take my measurements.

Well, good luck with your research. Thanks for sharing your results (having emitters hooked to a bench power supply for an extended period of time). BTW, I believe that Newbie had a few graphs to show a drop in Vf. I do not remember where (for the K2 emitters?). Just dig around in his old threads.

-Tony

 

[SemiMan]

Can not say I have ever done a "scientific" look at this. I would say in the first 4-8 hours at typical operating currents. I have seen everything from very little to perhaps 0.4V drop. I really should start checking more. I only have about 10-15 minutes on my current K2 160/180s. I may as well go down and turn them on and see what comes about. I am not sure my present heat sink is up to the task, but there is one way to find out!

Semiman

I just wanted to check, but do you mean a drop in Vf over the entire life of the LED, or the very first 6hrs, 24hrs, etc? I noticed some Vf drop in the K2s in the first few hours of operation. I like to burn in LEDs just so I know what Vf the LED will operate at in the intended operation. I usually just burn in my LEDs at 350mA for a few hours. Of course, even though I use some decent current regulation and I make sure the temperatures stabilize, I haven't been doing this scientifically enough because I have not been writing any of these measurements down.

-Tony

 

[evan9162]

I've got a TFFC K2 running at 1.5A right now. Will run it for a few days and see what happens.

 

[evan9162]

So far, no drop at all.

Since we've theorized that heat acellerates the process, I turned off the fan on my heat sink, raising the heat sink temperature to around 60C. At 1.5A, this puts Tj around 100C. Even running this overnight resulted in no permanent drop. The Vf does temporarily drop by about 0.14V due to the increase in Tj, but turning on the fan puts it right back up where it was initially.

 

[evan9162]

3 days now, and still no movement. Going to shut it down - we should have seen something significant happen by now.

 

[SemiMan]

5 hours with the die temperature likely in the 90C range, not a bit of change in the forward voltage. At that temp I am sitting at about 3.3V forward voltage at 700mA.

Semiman