MC-E Vf vs temperature?

[marcopolo]

There is obviously a clear linear relationship between luminous flux and tempereature, as per the datasheet

But does Vf vary with temperature? There is no mention of this in the datsheet.

If I have to measue a batch of MC-E's for Vf at varying currents is temperature an issue?

Marco.

 

[space]

But does Vf vary with temperature?

Yes. Very much so.

space

 

[2xTrinity]

Yes. Very much so.

space

This is why it is important to use a driver that regulates the current through the diode, rather than the voltage across the diode. Vf actually drops as temperature increases, so in an extreme case a process called thermal runaway can happen -- At a fixed voltage, the diode will draw some current and start to heat up. This will cause the Vf to drop, which will cause the current to increase even more, causing the temp to increase even further in a cycle.

Not only must your driver regulate current, ideally, you should have a separate current regulator for each diode (or string of diodes) in parallel:

[ driver ] -+----+
. . . . . .[LED].[LED]
. . . . . . . .| . . . |
. . . . . . . .[Ground]

is BAD

[ driver ] -+[LED]- [Ground]
[ driver ] -+[LED]- [Ground]

is good

In the first case, one diode can heat up more than the other, causing it to take a larger fraction of the current (say 75/25 instead of 50/50). If you have many diodes in parallel, driven by a current source, you can burn one out due to thermal runaway. Or worse, if for some reason one of your two LEDs in parallel (driven by the same current source) becomes an open circuit due to bad wiring or something, the current source will overdrive the other one by double, as it's only trying to control total current.

The one exception where running multiple LEDs in parallel is acceptable is with multi-die LED, such as the P7 or the MC-E. In that case, the dice are all on the same substrate so they stay the same temperature as each other. Also, the dice are fabricated at the same time, so their characteristics should be very well matched. The problem is putting separate LEDs (which can have different temps than each other, and different properties than each other) in parallel.

 

[ifor powell]

The datasheet has the tipical 'Tempreture coefficent of voltage' listed as -4mv per degree Celcius. This is per die so series wired it will be 16. So for mesurment and comparison terms you do need to take it into acount. Personaly I just measure the vf the same way each time so as to try and get some rough consistancy.

Ifor

 

[2xTrinity]

The datasheet has the tipical 'Tempreture coefficent of voltage' listed as -4mv per degree Celcius. This is per die so series wired it will be 16. So for mesurment and comparison terms you do need to take it into acount. Personaly I just measure the vf the same way each time so as to try and get some rough consistancy.

Ifor

Interestingly, this property may also be used to measure the temperature of the LED. Test the Vf at ambient with a small test current (not so much as to heat up the die). Then run it at a high current for a long time until the heatsink reaches its steady state temperature. At that point, ramp back down to your previous test current, and check the Vf in this case. Depending on how much lower it is, you can judge the temperature at the die.

Comparing that to the temperature of the heatsink, it's possible to determine the total thermal resistance of your system -- from the LED die, to the skin of the heatsink.