What is junction temperature for UV LED forward voltage measurement?

[treez]

Hello,
Page 5 of the attached datasheet gives forward voltage measurements for the UV LEDs at 350mA , and says it's done at an ambient temperature of 25 degC...however, do you know what was the junction temperature for these measurements?.

I mean the measurment might have been done by simply pulsing the leds with 350mA for 1millisecond every 5 seconds, and thus the led junction would be at about 25 degrees C, although the measurement might have been done with 350mA flowing continuously in the LED, which would have meant the junction temperature being much higher. This has an obvious effect on the actual forward voltage since as you know its variable with temperature.

 

[betwixt]

I would guess they assume an infinite heatsink so it would be done for a relatively short time, in the order of 100mS rather than 1mS. You really need their test conditions infomation to be sure. It is only the die and is for direct bonding to a (presumably) heatsink substrate so maintaining temperature is down to you. The data sheet is probably confidential by the way so you may want to remove it!

Brian.

 

[treez]

thanks, though there are a great many similar uv leds on the market, so it wont be confidential. Sorry but I don't think its just the die as on the 4th page it shows the backside metallisation, this is surely the full led?...its not just the die?
We need to assess the maximum forward voltage at 440mA to enable us to correctly design the led driver.

 

[D.A.(Tony)Stewart]

They most likely test everything in 0.1mS to prevent zero Tj thermal rise.
100ms is too long for this low mass 1sq mm part. When they say Ta, they also mean Tj.


Thus 10 repetitive measurements could be made with 0.1% duty cycle in 1 second. But it is up to them if they want to measure 1, 2 or 10 or more times.

I just read that the Illuminating Engineering Society: IES in North America called IESNA are proposing 0.05ms pulses to be adopted and 1% duty cycle max in a short burst.

 

[FvM]

I agree that a pulsed measurement is the only plausible way to specify Vf versus If without giving an exact heatsink configuration.

But in this case, I would expect a Tj instead of Ta specification.

 

[treez]

thanks, so in reality, we will never see 3.8V leds?....because in reality the junction will never be as low as 25degC.....this takes it down to 3.6V max in my book, not 3.8v.

But in this case, I would expect a Tj instead of Ta specification

Yes I agree, the datasheet is unusual on that

 

[D.A.(Tony)Stewart]

Correct if you design for 85C max at max Ta
Double-edge issue.. Too high is bad, too low from Temp is bad.


again I believe but you verify,, -200mV/60'C rise anything more is poor cooling in my books, we never allowed any hotspot>85C for commercial design

I guess you have already considered dumping your PWM and just buy a CC PSU for LEDs

Test Engineers had originally used Ta=Tc=25'C in semiconductor spec sheets, with massive controlled temp heat-sinks , then simply Ta=25 for pulse mode tests, (which you can't do for a uC.) How they do is up to them.

But we rarely know the Rjc of the LED package. A good clue is the Mfg Spec or Power vs Ta slope which is effectively the Rjc of the part, which is upwards of >500 deg C/W in 5mm Leds due to epoxy being a thermal insulator.

on another tangent... Sinkpad thermal solutions are pretty good for MCPCB.