Do dynamo lights really need heatsinks?

Hi All,
I have found a box full of very shiny NOS Union HS3 halogen headlights and matching tail lights in the bike shed. As some may recall these were very nicely made and worked well (for their era). I have run out of money to spend on expensive dynamo lights but still need a couple more and thought I might try retrofitting the Union's with LED emitters.

Nothing fancy like the back-to-back emitters some techno-wizz types have constructed. Just something super simple. I'm aware of the likely shortcomings re. illumination-point positioning so I'm not expecting miracles. A mate has a nice lathe so I thought I'd just try XR-Es or XP-Gs mounted on a small aluminium 'plug' jammed into the existing HS-3 bulb socket on the light's plastic base and designed to get the emitter as as far back as possible with respect to the curved reflector.

Naturally the question of heatsinking has arisen. Mind you, not one of the LED dynamo-driven lights I have made (from 2 to 6 emitters) has ever felt even warm to the touch after extended use. Most have been mounted on crude alu extrusion but even so, no discernible rise in temperature, even immediately behind the emitters. The 'plug' I have in mind for these Union conversions will be about 9mm in diameter and around 10mm long with a wider 1 X 15mm plate machined on top to take the LED's board. It'll be a tight fit into the housing's plastic base. I'll probably be running 2 white leds in series with one of Martin's simple circuits with a red tail light in the circuit as well (either an XR-E as part of the rectifier or an old B&M Toplight in parallel). So the dynamo's current will be shared across 2-3 emitters.

My question for the hive mind is, will this be adequate with regard to temperature protection? Has anyone actually run any tests - instrumental or otherwise - on temperature changes with dynamo driven leds?

Thanks,
Savvas.

I've spent quite a bit of time fretting over heat sinks for dynamo LEDs. If I understand your 'plug' concept correctly, I think it's similar to a copper plug I made to sit in the base of a Sturmey Archer headlamp's reflector. Personally, i think heat sinking is fairly important if you want to avoid excessive lumen drop as the LED heats up. By my measurements the Cree XML, rated at 3A, gets damn hot running at dynamo currents of about 0.5A. I ran a few tests of the Cree XPG on a small heat sink and was surprised by how quickly it heated up and by the amount of lumen drop. If your plug has a way of conducting the heat away into the lamp housing I think that should be sufficient. I've mounted my LEDs on heat sinks and run 0.5-0.6A through them and have always been surprised by how hot the heat sinks get without a thermal connection to anything else (just on the bench top).

With respect to optics, there are a handful of narrow optics that might have a small enough diameter to fit on your plug. These essentially functionally replace the original reflector.

First things first, connecting emitters in series = current is equal in emitters. Parallel spits current, Series splits voltage. This means that each emitter will have, potentially, the full 500ma running through it.
In practice, not so much. Dynamo lights rarely work at their peak current for long, especially if you don't have a standlight. Second, as the dynamo light's power is more-or-less proportional to speed, you pretty much guarantee airflow, so long as the LED has some kind of path to ambient--either if the aluminum is exposed to the air, or has a good connection to the external housing of the host light.
An LED running at 350ma or 500ma will get very hot--test with with a power supply and DMM if you have one. As long as there is some way for heat to move away from the aluminum socket, you'll be fine, but if its too well sealed, you might run into heat problems then, but its highly unlikely.

Just to confirm what has already been written, some heatsinking is needed. The question is how much.

I retrofitted a reflector from an old Union style headlight...
http://www.flickr.com/photos/kurtsj00/1022652239/in/set-72157601245803377

It was a Luxeon V side-emitter, and I epoxied it to a brass nut on a threaded aluminum shaft. The aluminum shaft had some aluminum plates bolted just aft of the LED. One of the plates served as part of the mounting bracket. Worked nicely enough, despite being ugly!

Maybe you can implement an analogous structure inside your Union headlight, where an aluminum bracket passes through the shell to serve as a thermal path to an aluminum bracket that also acts as a heatsink?? It probably looks better in my head than what you are imagining.

Alternately, you might get by with drilling a few ventilation holes in the shell that would let some air flow by some internal heatsinks?

Steve K.

Thanks folks. lots of good, sound advice. As I said, when I've used my own crudely made lights mounted on aluminium extrusion I've been unable to feel any rise in temperature. Just goes to show that crude bits of hacksawed alloy can be very effective heatsinks if left exposed out in the wind!

You've convinced me that my aluminium 'plug' will need to poke out the back of the plastic molded base of the light - Steve K knows what I mean. I'll probably follow his example and use this protruding bit to mount to an alloy bracket of some sort.

Incidentally Steve, your bike is painted exactly the same as the one this light is intended for! The frame was made for me a few years ago by the same friend with the lathe who's going to do the light stuff for me. He was feeling a bit down so I said 'build me a bike' (he'd built 50 or so previously). Always cheers him up!

Savvas

For reference the IQ cyo has only a small heatsink on top, body is plastic. Its does get warm but not hot.