From: "R. Lewis"
Subject: Re: luxeon star leds
Date: Sat, 23 Nov 2002 12:47:47 -0000
Organization: Nextra UK
NNTP-Posting-Date: Sat, 23 Nov 2002 12:46:35 +0000 (UTC)
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
"John Larkin" wrote in
> On Sat, 23 Nov 2002 01:11:28 +0000 (UTC),
> >In sci.electronics.design R. Lewis wrote:
> >> The problem is that where the device is modulated the average power may
> >> virtually meaningless.
> >> For example if the led is 'flashed' for 200mS every 3 seconds how can
> >> tell what is happening to the die?
> >> Luxeon appear to promote the fact that the die temp change may be
> >> by monitoring the forward voltage.
> >Calibrate one with an eternal temperature source at 100ua (or whatever).
> >Now, turn LED off, apply 100ua current, read voltage.
> This trick is commonly used to measure mosfet die temps. Apply power,
> then very quickly switch the power off and apply a modest drain
> current such as to foreward-bias the substrate diode. That
> instantaneous voltage drop (pre-calibrated as you describe) shows
> junction temp. You can even watch the junction cool off over time and
> learn a lot about the thermal time constants.
I am not sure how to apply what you suggest.
My problem is that, on increasing the power input, the heatsource (junction)
appears to increase in temp less than the heatsink by some quite marked
amount if the Luxeon junction tempco data is to be believed - or I have done