FADE TEST Osram LW E6SG surface mount LED

Special thanks to jtr1962 for sending me a few samples of this LED to test. Next to an ordinary 5mm LED, you can see it is very tiny.



You can read more about this in jtr1962's "white LED lumen testing" thread. I really like the neutral white ~5,500K light from it. Of course, in this tiny surface mount form factor, it has a very wide beam and is best suited for strip lights or back lighting (such as the intrument cluster in cars). At over 25 ma I recommend heat sinking as the tiny form factor can't dissipate heat all too well. It is designed to operate at up to 50 ma continuously.

Abuse time! I soldered thin 3" copper wires to the leads so that I could plug it on to my test board. I it was driven at around 39ma, so the LED ran warm. Near the end of the test after taking a measurment, I made a mistake and plugged it on the wrong pin of my LM317 in current regulation mode. I was alerted moments later when the LED started to blink! As it turns out, it was getting 150ma and sent the tiny LM317LZ (in TO-92 package) into thermal shutdown mode. It got about a minute of this abuse. After that, I let it run for another 24 hours at 39ma before taking the last light meter reading at 20ma. Results indicate that is a good LED and it was purchased off of ebay believe it or not.

Yaxis=Light output Xaxis=hours

Odd stuff: The cathode is marked by a bevel in the data sheet, but is in a different position on my samples. Datasheet also claims there is an ESD protection diode, but I did not get any current flow in reverse bias. Are these fakes? I don't know. jtr's tests seems to match the datasheet and I find them to be stable after my abuse.

Those are great results for 39 mA, and thanks for doing these tests! I look forward to seeing how long degradation to 70% takes. Regarding the lack of ESD diode and incorrect pinout, it's entirely possible that this was a batch of Osram's manufacturer rejects. That would explain why they ended up on eBay, and also why they still meet spec. Sometimes batches of parts with incorrect pinout are still sold for distribution via unconventional channels, rather than being destroyed.

Here are my test results in case anyone is interested:



Vf at 20 mA is 3.23 volts and output is 3.15 lumens. That's 48.8 lm/W ( spec sheet says 50 lm/W @ 20 mA, so these meet specs ).

BTW, I'm running these in LED nightlights at ~20 mA since early July, and so far, so good. These are the first LEDs I've found which haven't severely degraded in this application.

The thought of the rejects had crossed my mind, but it seems unlikely they botched both the molding and the ESD diode. It could be that I am not understanding the function of the ESD protection device. Reverse biased at 9 volts and no conduction. Seems like no ESD protection.

Cree XR-Es don't reverse conduct either, but the XP-E packages do conduct and have a visible ESD device next to the die.

I'll have to tinker further with controlled static zaps and higher reverse bias voltage. I think they probably are Osram rejects given the close specs and longevity.

BTW, did you ever get a chance to test the Cree 503C white one your usual test rig? I know you said you tested it in your "coffee can" integrating sphere, but was concerned with accuracy. Thanks jtr!

JohnR66 said:

BTW, did you ever get a chance to test the Cree 503C white one your usual test rig? I know you said you tested it in your "coffee can" integrating sphere, but was concerned with accuracy. Thanks jtr!

Click to expand...

Not yet-not feeling well last 2 weeks, and also my 5mm test jig needs to be repaired.

Y-axis is in.. lumen?

SirJMD said:

Y-axis is in.. lumen?

Click to expand...

No it is frequency in Khz. It is the light level to frequency output of the meter. It is linear and can be used for comparative tests of light intensity.

Total lumens would require an integrating sphere or several measurements at different angles and calculations. jtr1962's tests show the lumens of the LED.

JohnR66 said:

No it is frequency in Khz. It is the light level to frequency output of the meter. It is linear and can be used for comparative tests of light intensity.

Total lumens would require an integrating sphere or several measurements at different angles and calculations. jtr1962's tests show the lumens of the LED.

Click to expand...

So a certian level of light intensity corresponds to a certain frequency?


Meassured flux at a given distance and knowing the spreading of the LED, should be enough to calculate the total output.

No idea if there's a meter for it, or an easier way?:shrug:

SirJMD said:

So a certian level of light intensity corresponds to a certain frequency?


Meassured flux at a given distance and knowing the spreading of the LED, should be enough to calculate the total output.

No idea if there's a meter for it, or an easier way?:shrug:

Click to expand...

Yup. These seem to be designed to interface with a micro controller. With an accurate time base. the pulse train can be easily converted to a light value.

That would be easy if the beam was like a stage spotlight with a well defined edge and a smooth even illumination across the circle. 5mm LEDs have uneven beam shape, rings and other issues that make it hard to measure. jtr1962 seems to have this down pretty well. I'll let him do the total light output measurements. I'd like to calibrate my meter with some high accuracy meter so I can provide true intensity figures instead of relative figures.

For brightness and fade comparisons the frequency will work just fine as a relative figure.