[dcheming]

I'm just curious if anyone has ever played around with using Peltier thermoelectric coolers directly attached to transistors instead of heatsinks? I think this would be interesting to play around with in a SS class A headphone amp.

 

[Dave_M]

It would be possible to cool the transistors to below freezing temperatures with a peltier. But what would be the advantage? Lower thermal noise?

 

[ECM]

But then won't you have to cool the "hot" side of the thermoelectric coupler? the heat has to go somewhere...so you use a heatsink.

 

[dcheming]

Yeah you'd still need a heatsink to dissipate the heat from the Peltier, but I was just wondering if there would be a worthwhile performance gain from the much lower temperatures that would be possible. I figured the noise would be lower and that part life would increase, but would there be any substantial improvement in something else that would justify the added complexity?

 

[amb]

When it comes to noise, it's the input stage of an amp that contributes the lion's share. Active cooling of the output devices would do very little to reduce the actual noise performance of the amp (unless if you're talking about a single-stage amp like a Szekerez). OTOH, MOSFETs are said to sound better when hot.

 

[threepointone]

I'd personally think that it's better to just get a huge heatsink. From what I've been reading with peltiers in cpu cooling, it seems like they're too much trouble than they're worth. I think you have to regulate them to prevent condensation from forming, and although they can move lots of heat, they also dissipate (i think) twice as much heat as they move. And almost always, in cpu cooling, it seems like peltiers are coupled with water cooling systems.

I'd say skip the peltier and get a nice, big heatsink. Or you could add some fans or implement a (silent) water cooling system

 

[dcheming]

Thanks for the feedback guys.

 

[blip]

My understanding is basically in line with everyone else, peltier cooling is a waste except in very specific circumstances where you need to get something very cold and have provisions for water cooling/large heatsinks and just need an ultra-efficient thermal junction. Probably you would be better off looking into heatpipes if you want to accomplish something like that.

But there are other active approaches out there. Look at those amplifiers that they use for radio antenna arrays? To achieve super low distortion/noise they use supercooling. If you really want to, something similar could be accomplished with a phase change system. I imagine it would be hard to get noise free but is not necessarily extremely loud... if what you want is supercooling, it is the way to go.

I'm not sure how much real world benefit it would net you though.

Alternatively, if what you want is simply to not have to use a large heatsink, you might consider fans with thermal controls. If the system is well implemented you can move significant air while producing little to no perceivable noise.

 

[dcheming]

I don't mind large heatsinks really. Actually I like how equipment looks when it has large sinks as side panels. I was just assuming that performance would improve in transistors when cooled to frigid temperatures with a Peltier, but this doesn't seem to be the case. It's very humid where I live and controlling condensation would be a pain anyways. Well this seems like a lost cause; I'll just stick to good old passive cooling.