Thermoelectric Cooler/Generator

[dilettante]

Has anyone here experimented with using a thermoelectric cooler/generator like the one at http://www.dts-generator.com/ in place of or in addition to a heat sink in a high-watt Luxeon or Halogen application?

I'm certainly no Electrical Engineer and my interest in high performance lights is relatively new, but it seems like you could use this technology to remove heat from an emitter, lamp or heatsink and feed the resulting current back into the light.

 

[Lux Luthor]

This is very interesting!

If you could sandwich it between the emitter and the flashlight body, you might get a large enough temperature gradient to produce a fair amount of electricty, and remove thermal energy at the same time. It says they can adjust the parameters for specific applications too. I just wonder if it's a large enough effect to make it worthwhile.

PS. I wonder if this wouldn't get more attention in the LED mod forum.

 

[Kirk]

The thermoelectric cooler chips I've seen require a pretty hefty power supply. Also, you still have to heatsink the "hot" side of the chip. These chips don't make "cold" as much as move it around.
Regards,
Kirk

 

[B@rt]

Very interesting indeed!
Sort of a reversed Peltier element. I've dropped them an e-mail about pricing and availability. I will post the reply here.

Great find Dilettante

Ps. I agree Lux, this would probably get more attention in the Homemade and Modified lights forum...

 

[Jonathan]

Thermoelectric cooler/generators can do two things.

1) Given a temperature differential (a hot thing and a cold thing), they can generate electricity from the natural flow of heat (that is, the natural flow of heat from the hot thing to the cold thing). They only work if there is a temperature difference, meaning that the hot side stays hot and the cold side stays cold, so you need a heat sink on the cold side, and a continuous supply of heat on the hot side. Because the thermoelectric junction is in the way of the heat flow, the hot side will always be hotter then if it were simply directly connected to the heat sink.

So if you have a Luxeon and a heat sink, putting the thermoelectric junction between them would make the Luxeon hotter, but would generate a bit of electricity.

The problem is that the efficiency of the thermoelectric junctions is quite low, and you might convert 5% of the wasted heat back into electric power. This recovered electricity would be balanced against the lower efficiency of the Luxeon because of the higher junction temperature.

2) Given an external supply of electric power, a thermoelectric junction can pump heat from one location to another, _against_ the natural flow of heat. This means that with an externally powered thermoelectric junction you could have heat flow from a cool region to a warmer region. For example, you could cool the Luxeon down to 0C in a 25C room.

You are consuming electrical power to move this heat around, but this would have to be balanced against the improved efficiency of the Luxeon because of lower operating temperature.

-Jon

 

[B@rt]

Thanks for the explanation Jon, and here I was thinking to get a free ride.

I didn't think about decreased heatsinking capabilities... I'll leave it to the electronics experts to figure out if it is a worthwhile thing to add.

 

[hank]

www.allelectronics.com has surplus Peltier units - operate on 3 to 12 volts DC. 1.18" square for $17, 1.57" square for $26 (both 0.15" thick). "Up to 68 degrees C difference between the two sides ... must have heat sink."

I imagine a heat pipe might be easier to build into a hot LED though.

Why aren't we seeing LED lights that look like the 1930s science fiction ray guns? I remember lots of illustrations of blasters that had big flat disks spaced along the business end that must've been meant as heat sinks/radiators. Any of the copper pipe lights would look cute with a series of big copper washer fins around the lamp head.

 

[Mr. Blue]

OT: Diletante, welcome to CPF!!...Now you will get more stuff for that Claymore kit!

 

[**DONOTDELETE**]

The web site shows that you can get a whopping 25 uA of current at 5V with a temperature gradient of 80 degrees Kelvin. That means you'd be able to supply the quiescent current of a voltage regulator as long as you were operating your 1W LS at 1000mA in the Antarctic. Woo!

Thermal energy conversion is horribly inefficient, mostly due to the nature of semiconducting materials. Finding an efficient thermoelectric device is about as elusive as a room-temperature superconductor, and superconductors have a lot more research money going into them.

What can you do with 25 MICRO-amps anyways?

 

[Saaby]

Originally posted by hank:
www.allelectronics.com has surplus Peltier units - operate on 3 to 12 volts DC. 1.18" square for $17, 1.57" square for $26 (both 0.15" thick). "Up to 68 degrees C difference between the two sides ... must have heat sink."

I imagine a heat pipe might be easier to build into a hot LED though.

Why aren't we seeing LED lights that look like the 1930s science fiction ray guns? I remember lots of illustrations of blasters that had big flat disks spaced along the business end that must've been meant as heat sinks/radiators. Any of the copper pipe lights would look cute with a series of big copper washer fins around the lamp head.

<font size="2" face="Verdana, Arial">Something similar is on my lost of "To-Dos"

 

[lemlux]

It looks like the Peltier junction would draw between 1.75 A and 2.0 A at the 6.5 V to 6.8 V output optimally used by a 5W LS. Thats equal to the current required for 2 to 3 5W LS's.

I have a Peltier cooler of this size that I used between an overclocked AMD K2-300 and a heat sink / fan combination three computers ago. I always ran this computer with open case because of the incremental overall heat generated by the cooler. The hot-side heat has permanently fused the thermal grease or adhesive (I forget which I used)between the cooler and the older heat sink / fan.

It doesn't look like this will be the application that will resurect the usefulness of my Peltier. On the other hand, maybe I could put a 5V fan on the heat sink and drive Peltier, fan copper cooled side plate with multiple 5W LS on the other side & everything connected to multiple parallel 7.2 V NiMH packs.

Nope, even I'm not that crazy.

 

[Evan]

I have one of those peltier coolers for a pentium. The trouble is, if you have more energy to pump out than the device can handle, it isn't as good as a heat sink alone. My motherboard reports CPU temperature, and the pentium 233 ran hotter with the cooler than with an ordinary heatsink. I expect the cooler was desigened for a CPU that dissapates less, say a pentium 120, and the 233 just overpowered it.

I've been thinking about ways to attach it to the side of a soda can to keep my beverage frosty.

 

[lemlux]

Evan:

I suspect a larger more recent heat sink will solve your problem. My current heat sink on an AMD k7 1400 is massive aluminum with a copper disk at the contact area to quickly move heat out to the aluminum fins. These days you look at the total weight of heat sinks to see if they break the CPU with their torque when the tower is being moved. I think AMD certifies a max weight of 300 grams unless the heat sink has outboard mounting supports that mount through special holes in the motherboard.

 

[Saaby]

Originally posted by lemlux:
Evan:

I suspect a larger more recent heat sink will solve your problem. My current heat sink on an AMD k7 1400 is massive aluminum with a copper disk at the contact area to quickly move heat out to the aluminum fins. These days you look at the total weight of heat sinks to see if they break the CPU with their torque when the tower is being moved. I think AMD certifies a max weight of 300 grams unless the heat sink has outboard mounting supports that mount through special holes in the motherboard.

<font size="2" face="Verdana, Arial">Could that be a fine Thermaltake product?? I used a shim when I mounted my (dads) fan just to be a little safer