Peltier Cooling

[CycloWizard]

I'm not terribly familiar with Peltier coolers. However, I'm working on an application where they would be perfect if they are applicable. I want to keep a steel block at a constant temperature while it is hanging in an oven at an elevated temperatuer. I can easily feed curent to the Peltier. My question is: is it feasible to remove the heat from the steel block and get it outside the oven using a Peltier? If anyone has any insight and/or can link me to some products that would work for this, I would appreciate it. :beer:

edit: It looks like I'll be moving up to about 20 W, though if that's too high, I can easily lower it.

 

[firewolfsm]

Remember that the peltier will create a heat gradient from the center of the block extending to the surface. It won't be even if that's what you need.

 

[CycloWizard]

Originally posted by: firewolfsm
Remember that the peltier will create a heat gradient from the center of the block extending to the surface. It won't be even if that's what you need.

I'm looking to stick one on the end (or maybe one on each end) to pump the heat out. Th thermal conductivity of my sample is very high, so the internal thermal resistance should be very low. Of course, my experiment will create a very small thermal gradient within the sample no matter what I do, so it's all relative.

 

[bobsmith1492]

You'd want to bolt a peltier onto a heatsink with thermal grease or a pad, just like a processor. Then, probably a thermal pad between the peltier and your block would be your best bet for good contact.

Here is a heat sink pad that should do the trick (a bit pricy though... call up Fujipoly or Laird and they'll send you a free sample sheet )

There are lots of peltiers on Ebay that can take 20W no problem - this one, for instance - a pair of 136W ones for cheap.

Then all you'd need is a high-current DC power supply. Also, you'd have to control it if you wanted to maintain a particular temperature... one could whip up a circuit to interface a thermistor with a variable-output PSU to to the job.

 

[silverpig]

1 VERY important thing about using a peltier: You HAVE to cool the other side with a heatsink and fan or else it won't do anything except heat up your sample even more. If you just run a peltier by itself it'll get cold on one side and hot on the other, then very quickly (a few seconds), the cold side will get hot and the hot side will get hotter.

 

[CycloWizard]

Originally posted by: bobsmith1492
You'd want to bolt a peltier onto a heatsink with thermal grease or a pad, just like a processor. Then, probably a thermal pad between the peltier and your block would be your best bet for good contact.

Here is a heat sink pad that should do the trick (a bit pricy though... call up Fujipoly or Laird and they'll send you a free sample sheet )

There are lots of peltiers on Ebay that can take 20W no problem - this one, for instance - a pair of 136W ones for cheap.

Then all you'd need is a high-current DC power supply. Also, you'd have to control it if you wanted to maintain a particular temperature... one could whip up a circuit to interface a thermistor with a variable-output PSU to to the job.

Controlling it won't be an issue. This is for a long-term research project, so there will be a little money to throw at it. I can whip up a software controller without much trouble.

Originally posted by: silverpig
1 VERY important thing about using a peltier: You HAVE to cool the other side with a heatsink and fan or else it won't do anything except heat up your sample even more. If you just run a peltier by itself it'll get cold on one side and hot on the other, then very quickly (a few seconds), the cold side will get hot and the hot side will get hotter.

Right. So maybe I should have phrased my question, "How far apart can the "sides" of a Peltier be?" Maybe a little more info is needed.

I'll have the steel block hanging from a load cell. I want to measure the change in mass (due to condensation) over time of the steel block while maintaining it at a constant temperature. Thus, I would like to avoid having a traditional heatsink attached to the block itself, hence my interest in the Peltier. Since condensation will be running down the side of the block, whatever I end up using should not be on the sides, though top and bottom should be fine. I can even machine it and put it inside, but I don't think that would really help with a Peltier.

 

[Ticky]

What I'd do is use a chilled water system. Just run water through the steel block, and chill the water, either with a Peltier, or a compressor/evaporator. That lets you move the hot end as far away from the sample as you want. I assume if you can pass current to the steel block, you can pass water to it. Once you got it started, it would be easy to maintain a constant temp, and you could minimize the thermal gradient with proper layout of the passages inside the block.

 

[silverpig]

Originally posted by: CycloWizard

Originally posted by: bobsmith1492
You'd want to bolt a peltier onto a heatsink with thermal grease or a pad, just like a processor. Then, probably a thermal pad between the peltier and your block would be your best bet for good contact.

Here is a heat sink pad that should do the trick (a bit pricy though... call up Fujipoly or Laird and they'll send you a free sample sheet )

There are lots of peltiers on Ebay that can take 20W no problem - this one, for instance - a pair of 136W ones for cheap.

Then all you'd need is a high-current DC power supply. Also, you'd have to control it if you wanted to maintain a particular temperature... one could whip up a circuit to interface a thermistor with a variable-output PSU to to the job.

Controlling it won't be an issue. This is for a long-term research project, so there will be a little money to throw at it. I can whip up a software controller without much trouble.

Originally posted by: silverpig
1 VERY important thing about using a peltier: You HAVE to cool the other side with a heatsink and fan or else it won't do anything except heat up your sample even more. If you just run a peltier by itself it'll get cold on one side and hot on the other, then very quickly (a few seconds), the cold side will get hot and the hot side will get hotter.

Right. So maybe I should have phrased my question, "How far apart can the "sides" of a Peltier be?" Maybe a little more info is needed.

I'll have the steel block hanging from a load cell. I want to measure the change in mass (due to condensation) over time of the steel block while maintaining it at a constant temperature. Thus, I would like to avoid having a traditional heatsink attached to the block itself, hence my interest in the Peltier. Since condensation will be running down the side of the block, whatever I end up using should not be on the sides, though top and bottom should be fine. I can even machine it and put it inside, but I don't think that would really help with a Peltier.

What do you mean how far apart can they be? Peltiers are generally flat cracker sized things with 2 wires coming out of them. You sandwich the peltier between the heatsink and the block and it helps move heat from the block to the sink.

If you really want to maintain a constant temperature, do what Ticky suggests and run water through it.

 

[Biftheunderstudy]

Yup, chilled water is how the Invar in our laser is kept at a constant temp. And that has to be stable enough to keep the frequency stable.

 

[CycloWizard]

Running water through it was my first idea. Unfortunately, I'd be observing very small masses, so I would need a very high quality pump to keep the noise down due to pumping-induced vibrations, plus the tubing required would add additional weight and bulk to the system. So I thought a Peltier might be an alternative, but it looks like the consensus is that it won't work.

So now I'll ask: does anyone know of a product that is just a steel block cast with a built-in path for water to travel through? I could machine one, but not without adding mass/bulk due to the inclusion of additional tubing and/or gaskets. Something like this, but cast so that it is all one piece (and made from stainless) would be just about right. Something much bigger than this would be even better, though I doubt there is such a thing commercially available since I can't think of a practical application for it.

 

[herm0016]

just take a block and drill 2 angled holes so that they connect in somewhere inside the block, or cross drill it and then weld the sides back up to solid. this does not really sound all that well thought out... the water will be changing the mass all the time, you will have to deal with the force from the water getting pumped through the block etc... is there a way to use a hot block and a cold environment? its much easier to heat things up than it is to cool them. can we get more details on the experiment?

 

[CycloWizard]

Originally posted by: herm0016
just take a block and drill 2 angled holes so that they connect in somewhere inside the block, or cross drill it and then weld the sides back up to solid. this does not really sound all that well thought out... the water will be changing the mass all the time, you will have to deal with the force from the water getting pumped through the block etc... is there a way to use a hot block and a cold environment? its much easier to heat things up than it is to cool them. can we get more details on the experiment?

The idea I had for the experiment was essentially to throw an isothermal block in a humidity chamber and measure the mass of condensate over time. The block has to be colder than its surroundings for the condensation to form. And no, it's not that well thought out. I just had a hare-brained idea yesterday and thought maybe I could test it out without too much trouble.

As far as the angled drilling, that would work if the block is wide enough. I suppose I can make the block wide enough, so that might just work. The force of the water on the block is something that I considered and could fairly easily account for by doing some dry runs (in this case, "dry" meaning "without the condensation" ). However, the oscillations that it might cause could also be enough to disturb the formation of the condensate film, which would render the results pretty much useless. Perhaps if we just built this large wooden badger...

 

[silverpig]

If you want something less accurate, but much easier to do try this:

Make a hole in the bottom of your box. Put an insulated copper tube through this hole. Put a copper funnel at the top of the tube inside the box. Insulate the outside of the funnel and cool it via conduction through the pipe outside the box by having the tube go vertically through a tank of ice water. With the outside of the tube insulated well you should be able to hold the tube+funnel at a constant temperature quite well.

Condensation will form on the inside of the funnel and will run down it into a container on a scale. Make sure the container is well insulated on the outside and chilled to prevent evaporation. There'll be a bit of lag in your signal while the condensation builds up, but you should have a good average after a while.

 

[CycloWizard]

Originally posted by: silverpig
If you want something less accurate, but much easier to do try this:

Make a hole in the bottom of your box. Put an insulated copper tube through this hole. Put a copper funnel at the top of the tube inside the box. Insulate the outside of the funnel and cool it via conduction through the pipe outside the box by having the tube go vertically through a tank of ice water. With the outside of the tube insulated well you should be able to hold the tube+funnel at a constant temperature quite well.

Condensation will form on the inside of the funnel and will run down it into a container on a scale. Make sure the container is well insulated on the outside and chilled to prevent evaporation. There'll be a bit of lag in your signal while the condensation builds up, but you should have a good average after a while.

That is a good idea. Unfortunately, I can't go with the funnel because of the theory that I'm working from, which requires a constant cross section. I might be able to re-develop the theory to account for a linear variation in cross section, but I don't think have time to do that by next Friday when this proposal is due.

 

[Ticky]

How about using a water tower? Instead of pumping the water directly into the block, pump it up to a reservoir, and let it gravity feed into the block. If you really want to get fancy, set the pump up so that it constantly overflows the reservoir, so the water pressure due to gravity remains constant.

Does the block have to be stainless steel? Because there are plenty of square-sided commercial water blocks, but they're all copper or aluminum or silver-plated copper.

 

[CycloWizard]

Originally posted by: Ticky
How about using a water tower? Instead of pumping the water directly into the block, pump it up to a reservoir, and let it gravity feed into the block. If you really want to get fancy, set the pump up so that it constantly overflows the reservoir, so the water pressure due to gravity remains constant.

I like the water tower idea. That would alleviate the problems associated with pumping pressure fluctuations pretty nicely.

Does the block have to be stainless steel? Because there are plenty of square-sided commercial water blocks, but they're all copper or aluminum or silver-plated copper.

Yes, unfortunately it does need to be steel. For what I'm doing, I need something that is corrosion-resistant since it will constantly be exposted to saturated atmospheres. I chose steel over other corrosion-resistant materials because it happens to have the highest thermal mass (product of density and heat capacity) at the conditions I'm looking at. That, and it's easy to machine and it's cheap.

 

[tomcat2200]

This sounds strangely like you are trying to make an experiment overly complex.

If you are trying to heat a sample then heat it, if you are trying to cool a sample then turn the heat down.

If you are trying to expose a sample to a given environment but cooling it at the same time, then only expose one side to the environment. You can mount a heat sink on the other side, or better yet machine groves and cool the other side with a fan.

This just doesn't sound like you need to be all that exotic with this.

Of course with the limited information you are providing about what you are trying to do, then little will make sense.

This would at least give you a constant cross section, as you state you require.

 

[CycloWizard]

Originally posted by: tomcat2200
This sounds strangely like you are trying to make an experiment overly complex.

If you are trying to heat a sample then heat it, if you are trying to cool a sample then turn the heat down.

If you are trying to expose a sample to a given environment but cooling it at the same time, then only expose one side to the environment. You can mount a heat sink on the other side, or better yet machine groves and cool the other side with a fan.

This just doesn't sound like you need to be all that exotic with this.

Of course with the limited information you are providing about what you are trying to do, then little will make sense.

This would at least give you a constant cross section, as you state you require.

I need to study steady state condensation, which will occur only if my sample is isothermal. If it's heating up, then it will equilibrate with its surroundings. To test the theory, I need to weigh my sample (or, at least, the condensate that falls off it). It's hard to just collect condensate falling from it because whatever I use to catch the condensate will itself get condensation on it. If you can simplify it and still achieve all of these things, I'm all ears. The theory is fairly complex and was derived by one of these smart guys that lived 100 years ago. Like I said, I might be able to alter the theory to account for a different system, but not by Friday when this proposal is due. Right now, I came up with an alternative method that will give me some of the information, so I suppose if I get the grant, I will have enough time to adapt the theory to my experiments.

 

[PlasmaBomb]

People have looked at the efficiency of film condensation vs droplet condensation. I would maybe look into how they conducted their experiments...

 

[CycloWizard]

Originally posted by: PlasmaBomb
People have looked at the efficiency of film condensation vs droplet condensation. I would maybe look into how they conducted their experiments...

I've read almost 50 papers on that subject in the last week. All of them are looking at industrial usage (i.e. heat exchangers), so it's not difficult for them to maintain constant temperatuers on both sides of the exchangers. I'm looking at perhaps inducing dropwise condensation using new surface materials.