Building a Peltier module

[Rachie5272]

Hello, I want to build a Peltier module from scratch.  I have some ideas I want to test out, and there's no better way than prototyping.

Where would I find the raw N and P semiconductor elements to make this from?  I don't even know what they're called, or how to search for them.  Am I better off salvaging them from a cheap module?

[ironmonkey]

AFAIK they are two different metal sheets, with no semiconductor involved (kind of a inverse termocouple)

Enviado desde mi MotoG3 mediante Tapatalk

[Rachie5272]

Theoretically this should work, but every explanation of Peltier junctions I've seen says they use semiconductors.  Does anyone know why?

[LukeW]

Lots of materials have some Seebeck coefficient (like a thermocouple) but you need a *high* Seebeck coefficient. Doped bismuth telluride is usually the material of choice.

[vealmike]

Do you have to use Bismuth metal, or can you use stuff from any WWII wreck?

[Codebird]

You don't need bismuth. You need bismuth telluride. Good luck getting hold of that. What exactly is it that you want to test?

[CM800]

You don't need bismuth. You need bismuth telluride. Good luck getting hold of that. What exactly is it that you want to test?

Seems his luck is turning!
Here you go:

https://www.alibaba.com/trade/search?IndexArea=product_en&CatId=&fsb=y&SearchText=Bismuth+Telluride+pellet

[PointyOintment]

Theoretically this should work, but every explanation of Peltier junctions I've seen says they use semiconductors.  Does anyone know why?

My understanding is that this is done because it allows you to create a thermopile (many thermocouples in series electrically but in parallel thermally), which is much more effective than a single thermocouple. Also, you need both polarities so that you can put them in series electrically without the plates shorting them out. I forget which, but one of the high-quality TEC manufacturers has a good page explaining this. (They also claim to be able to get good efficiency (something higher than 60% IIRC, maybe up to 95%) while everyone else says TECs don't get more than ~5% efficient.)

[vealmike]

Sorry, I was thinking of Bismark.

[JimRemington]

Buy a roll of thermocouple wire and make yourself a thermopile.

Lots of cool examples here: http://www.douglas-self.com/MUSEUM/POWER/thermoelectric/thermoelectric.htm#rl

[Rachie5272]

I can't find any small quantities.  I guess it's not popular enough.

I want to make some irregularly shaped coolers for a niche application.  If I could find pellets, it would be a simple matter to solder them up on a custom PCB.  I don't care about efficiency at the moment.

[CM800]

Contact some of the Sellers on Alibaba and ask to buy a sample quantity, It's not hard and I doubt it will be too expensive.

[Kleinstein]

The efficiency of metallic peltier elements is very limited. The bismuth telluride based ones reach a much higher efficiency - though still not very high. Something like 10% of the carnot efficiency is about as good as it gets. So it depends on the temperatures and what you look at, what efficiency you get. Conversion of heat flow to electricity might have a hard time to reach 5% efficiency as the maximum temperature is rather limited. The COP ( = heat moved divided by electricity used) can reach high numbers at low temperature difference, but will fast get smaller if a significant temperature difference is needed. So in a cooling application it is very important to get the heat away from the hot side.
 
It takes 2 types of pellets with proper doping (N and P type). The contacts might also need some extra care, to make them solder-able. So the best guess is probably taking apart one of the cheap modules. I am not sure how well the elements will survive soldering.

[NiHaoMike]

You can reverse engineer the connections in a cheap module, then use a glass cutter to cut it into smaller pieces. (You'll probably destroy at least a few modules trying to figure out how to do it.)

[CM800]

Huh... well...


seems I'm gonna be building a Peltier module, the company offered me free samples!

How does one bond the Bismuth onto copper? D:

[SeanB]

Woods metal or other low melting point alloys are what is used to bond to copper.

[Rachie5272]

Woods metal wouldn't be much use in a Peltier module because it melts at 70 °C.  Having your cooler fall apart when things get hot probably isn't the best idea.

For prototyping, conductive epoxy might be the easiest.

[PointyOintment]

According to my limited knowledge, these things are usually assembled using solder. The melting point of the solder is what determines the maximum hot-side temperature—any hotter and it would fall apart.

[zenman]

Are any of you all still interested in this?  I have been working on making both Bismuth Telluride (n-type) and Antimony Telluride (p-type) in my workshop for the last year. I have figured out how to alloy both compounds but I don’t have the measurement tools to figure out their Seebeck coefficients. I am going to attempt to make one of these tools from scratch since the professional ones cost $9000.

Anyway I would LOVE some more brainpower behind this if any of you are interested in helping. I have 100s of unlisted youtube videos of my attempts. I’m slowly organizing them.

I have chunks of both Bi2Te3 and Sb2Te3 that I’ve made. I’m happy to share how I made them or just send some samples.