Don Foreman wrote:
I don't think it'll be a problem. You can run a Star pretty close to
1 watt without more heatsink if it's surrounded with free air. My
3-watt "night burner" lantern just uses a 2" square of .090 aluminum
for a heatsink. It gets warm, but not hot. Using the heatsink
calculations from National's Regulator Applications Handbook (adapted
to use in MathCAD) I figure 3 watts on a 2" square sink would run at
about 130F with 68F ambient. This one doesn't run that warm,
probably because some of that 3 watts is emitted as light. One rule
of thumb is that 1 watt of radiant flux is about 43 lumens, though
that would depend on spectral distribution. (Light Measurement
Handbook by Alex Ryer) The Lux running at 1 watt nominally
produces 25 lumens which would be about .58 watts emitted as light
leaving only .42 watts to dissipate locally as heat.


I see here on the Star data sheet (DS23.pdf) that at peak power,
the device converts power at a rate of 1.19 W. That's 3.4 V times
0.35 A. Think of it as a (admittedly nonlinear) 9.7 ohm resistor.

From other sources, I recall these LEDs are about 3% efficient.
Only about 0.036 W of that 1.19 W gets emitted as light.
So you get to design a heatsink to keep die temperature below 60 C
for *each* LED while losing 1.14 W of heat per LED.

Things get squirrley quick when you have two or more thermal power
sources connected in parallel like this. My little brain boggles
at trying to SWAG a heatsink for 6 or 10 devices!

Some empirical experiments are in order, methinks.

(Happy Thanksgiving)

--Winston