LED and constant current supplies

The PT4105 (at least, _that_ PT4105) is definitely not the part on the power supply you indicate above. The TI/PowerTrends PT1405 is a 15 watt DC-DC converter - it will be a fairly large module, not the tiny IC shown in the vendor's picture of the power supply.

Put the LEDs in _series_

Do NOT put them in parallel.

Ed

If the LEDS are going to be left on for more than a few seconds at a time, they also should be mounted on a heatsink if you dont want to fry them.

How much of a heatsink should there be? Most commercial applications I've seen only have the circuit board tracks to sink heat.

seen only have the circuit board tracks to sink heat.

Not a lot :-) If you looked at the OP's link, the LED dies are supplied mounted on what appears to be an aluminium backed circuit board. Each LED is dissipating about 1/2 a watt. They will probably survive for a while albeit running rather warm in free air, but if fully or partially enclosed in a non-metallic enclosure I would expect them to fail prematurely due to overheating. If they are mounted to an aluminium or copper bar or as you say a circuit board with heavy copper tracks and there is a reasonable extra surface area to dissipate the heat they'll do fine. The key thing is to monitor the temperature rise, If they are running more than 30 deg C over ambient they should probably have a bigger heatsink.

e:

I've seen only have the circuit board tracks to sink heat.

Well, I plan to use a salvaged aluminum external hard drive enclosure; that way I get a big heat sink area + a 12VDC supply. And it's free, or nearly so. :-)

If that's not enough, I will add a couple of old CPU heat sinks and some cpu heat sink goop.

seen only have the circuit board tracks to sink heat.

I seem to remember a rule of thumb for LED heatsinking that was something like 30 sq inches of Aluminium surface in free air per watt (rather less with a fan). I'm pretty sure a drive case is big enough. You should be good to go with just a smear of heatsink grease on the back and three screws engaging the notches (dont forget a fiber washer on each under the head so they dont short the LED) or a dab of thermally conductive epoxy to mount them. If you can keep your finger on them without getting burnt, they're cool enough.

seen only have the circuit board tracks to sink heat.

That is a HUGE area, those figures must be wrong.

Yes, that is huge. I do OK with 3 square inches per watt, sometimes less.

In fact, a 1 watt Luxeon "Star" in free air (both sides of the MCPCB exposed, or if the "star" is mounted on a heatsink barely big enough to mount on) *usually* runs adequately cool in a 25 degree C ambient.

===============================================

Area of blackbody radiator (anodized aluminum is close enough) to remove

Area of blackbody radiator to remove

These figures are assuming no convection or conduction - only radiation, but perfect ability to do so.

My experience indicates that heat transfer from a heatsink by convection varies greatly, but normally well exceeds what radiation can do.

Keep in mind that effective area for radiation from a finned heatsink is close to that of a metal box whose overal dimensions are the same as those of the heatsink. But fins do help with convection as long as they are vertical.

- Don Klipstein ( snipped-for-privacy@misty.com)

Thanks for the correction. I forget where I got that from, but obviously someone has slipped a decimal place (probably me :-( ). My first reaction for something like this would usually be see if one of those little twisted fin 30 deg C/watt TO220 heatsinks could handle it.

Sounds reasonably OK:

OK, add that 30 degrees C per watt to the thermal resistance within the LED. That figure is anywhere from 5.5 to 23 degrees C per watt for the various Luxeons, 5.5 to 20 for any of the various blue, green or white Luxeons.

That makes for a total of 35.5 to 50 degrees C per watt for a white Luxeon or as high as 53 for a yellow, orange or red Luxeon, depending on which one.

At 350 mA, power dissipation is usually 1.1 to 1.25 watts, but in a bad case can be as high as 1.4 watts (somewhat less for yellow, orange or red). It is not really good engineering practice to absolutely rely on some of the heat being removed in the form of light as opposed to being removed by the heatsink - the heatsink still has to work if the LED is on the low side in light output, dimmed by temperature rise, dimmed by age, dimmed by dirt on it, or experiences light being reflected back into it.

So depending on which Luxeon, whether its voltage drop is on the low side or on the high side or in between, and your luck, at 350 mA the junction temperature can be anywhere from 39 to 70 degrees C warmer than the ambient - as little as 29 degrees C warmer or a bit less if efficiency at producing light is really good.

Depending on which Luxeon, maximum juntion temperature for 50,000 hour life expectancy is anywhere from 90 to 120 C.

So, a Luxeon I may being operated conservatively on that heatsink at 350 MA but would be at 250 mA. On the other hand, a Luxeon K2 (variant with TFFC) should work nice and fine and conservatively on that heatsink at 450 mA. The Luxeon I will work at least reasonably OK on that heatsink at 350 mA, and the K2/TFFC will work at least reasonably OK on that heatsink at

- Don Klipstein ( snipped-for-privacy@misty.com)