Overclock.net banner

Would it be beneficial?

  • Yes, indeed!

    Votes: 1 33.3%
  • Maybe a bit.

    Votes: 1 33.3%
  • It wouldn't do anything.

    Votes: 1 33.3%
  • Really? This is a stupid question, you obviously know nothing about thermal science at all. (Which i

    Votes: 0 0.0%

Unusual Usage of Peltier, Would this Work?

1K views 12 replies 7 participants last post by  Puck 
#1 ·
Hey guys,
Honestly I put "unusual" in the title because I've never seen anyone ask about this idea.
Now, I have very little knowledge of the use of Peltiers and TEC solutions, but I figured I'd ask anyways.

I'm doing a new build, and it's an all custom water cooling loop.
I've done the math & the radiators and such are all fine for keeping the temps close to ambient even under load. (5-8C error ratio or so)

Now the question I have is more of a "What if."

I was thinking, what if I took a Peltier, akin to this
And a EK VGA waterblock, or really any waterblock of the right dimensions such as this
And the proceeded to use a thermal adhisive like this one

(And now here comes the part I think is unusual),

and use the thermal adhesive to "glue" the cold side of the peltier to the waterblock.
The idea is that I would then hook up this waterblock to my loop, presumable near the res, and let the water pass through the block, being cooled by the Peltier, if that makes sense.

Of course I would also cool the hot side, but let's leave that aside for now and just assume that's not an issue.
Would the water actually receive any cooling benefit from this setup?
Ignore the thermal capacity to hold the sub-ambient water temps, I'm not interested in finding out if it could be more effectively maintained by an insulated res, etc.

My question is simply, in a water passing through a peltier cooled waterblock situation, would it receive a thermal dissipation that would even be noticeable? I mean, consider the water going through the block at different temperatures as well. I'm well aware that without the proper insulation, etc. sub-ambient water would not hold the temp. What if the water going through it was 10C above ambient for example, would it make any difference?

Keep in mind, this would be more for a "wow" factor than an actual thermal dissipation usage. Imagine having it hooked up to a switch on the case that would turn on the peltier and the cooling element for the hot side, for perhaps an intense gaming session. I just want to know if there would be any benefit at all.

Also, just to make this clear, the waterblock would not be resting on my cpu or any hardware besides the peltier, it would be mounted independently, assume this would be done effectively.

Any/All speculation or facts welcome.
 
See less See more
#2 ·
That would work. A water block doesn't "know" if it's heating or cooling.

The big problem though is that if you cool the water below the ambient temperature, the radiator(s) in the loop will just be heating it up again as they are using ambient temperature air for cooling,
 
  • Rep+
Reactions: Soraniki
#3 ·
Would not work. Peltiers don't really work on a "partial" basis...they can either handle the heat load, or cannot. With that peltier, and a modern CPU, you will not notice any difference in coolant temp since the cold side would be subject to more heat load then the peltier can move, causing it to heat up over time. Eventually, the cold side would no longer be cold...and once it hits room temp, it will do absolutely nothing.

This is besides that fact that a radiator works against a peltier in the same loop.

It it was a much stronger~150w peltier, and a proper TEC waterblock that cools the whole surface instead of focusing on the center, then it would help
smile.gif
.
 
#4 ·
Well the point of is not to cool the temps on it's own. I realize the radiators would be working against it.
I'm saying, in a hypothetical situation the rads would not be able to cool the temps to exactly ambient.
So if the water was 4-10C above ambient, would the peltier cool it closer to ambient.
I'm not trying to make it do most of the work or anything, like I said it's more of a supplementary cooler, i don't feel the need for sub-ambient temps. (yet!)

All I want to know is if it would be able to cool the temps at all. Even if when it's running it moves the temps 1C closer to ambient.
As I stated in the OP, it's for a "wow" factor more than practical application. Even so, I don't want it to just sit there and do nothing.
I'm willing to pay for it if it would do something, even a little.

Are you saying it would just be completely irrelevant?
(not trying to be rude if it came off that way, just want to know)

Also - as I said in the OP, this will NOT be mounted to my CPU. the waterblock simply works as medium so the water flows over the peltier so to speak.
To make this even more clear, there would be a waterblock on my CPU (with no peltier), and then there would be ANOTHER waterblock just for the peltier somewhere else in the loop.
 
#6 ·
This seems theoretically sound and a great idea. Practically, would it be possible to reduce the temperature of fluid moving that quickly through such a small space? Effectively cooling flowing liquid would be difficult, as thermal entropy leans towards heating as opposed to cooling. It would be easier and more effective to cool the slower moving liquid in the reservoir than over the CPU. Like it was already stated, the peltier junction would lose its thermal capability over time and load, decreasing its effectiveness at cooling the liquid. May I suggest a peltier sleeve that encased the res instead?
 
#7 ·
Quote:
Originally Posted by Supersenka4508 View Post

This seems theoretically sound and a great idea. Practically, would it be possible to reduce the temperature of fluid moving that quickly through such a small space? Effectively cooling flowing liquid would be difficult, as thermal entropy leans towards heating as opposed to cooling. It would be easier and more effective to cool the slower moving liquid in the reservoir than over the CPU. Like it was already stated, the peltier junction would lose its thermal capability over time and load, decreasing its effectiveness at cooling the liquid. May I suggest a peltier sleeve that encased the res instead?
I'm hoping by "over the cpu" you're talking about the separate cpu block, because like I said it won't be touching any hardware. I'm assuming that's what you meant.
And yes, thats why I was asking. I was trying to figure out if it would be able to cool it when the water is rushing through so quickly. I think I'm going to try it out anyways,
it won't be for a few months at least, but I will post the results of some stress testing. I'm planning on running a stress test with it off, and seeing what the water temp averages are,
then I'll turn on the peltier and see if it makes any difference.

As for your suggestion, yes I am sure that would be more effective, however I am not interested in attaching it to the res. It's a very cool looking special edition res, so I don't want to
damage it's ascetics.

I will post some results when I get them and then we'll know if it works!
 
#8 ·
Quote:
Originally Posted by Soraniki View Post

I'm hoping by "over the cpu" you're talking about the separate cpu block, because like I said it won't be touching any hardware. I'm assuming that's what you meant.
And yes, thats why I was asking. I was trying to figure out if it would be able to cool it when the water is rushing through so quickly. I think I'm going to try it out anyways,
it won't be for a few months at least, but I will post the results of some stress testing. I'm planning on running a stress test with it off, and seeing what the water temp averages are,
then I'll turn on the peltier and see if it makes any difference.

As for your suggestion, yes I am sure that would be more effective, however I am not interested in attaching it to the res. It's a very cool looking special edition res, so I don't want to
damage it's ascetics.

I will post some results when I get them and then we'll know if it works!
Is there anyway you can run the separate block at less PSI then the rest of the system? If there was a way to increase the internal diameter of the block you wanna use, then decrease the diameter of the outlet pipe, theoretically the liquid should decrease speed through the block, increasing the cooling effectiveness of the peltier. Because water is an uncompressable liquid, having a small inlet through a series of larger diameter piping would decrease flow volume through them, increasing the latency time the fluid takes to travel. This then allows the peltier more time, and time is money in this, to cool the liquid from say 40c to ambient. Maybe after the rad before the CPU would be best, or between the red and the rad. Your choice though, its your money. Maybe an aluminum block with 3/8in inlet, interior channeling at 1/2 or 3/4 through several curve's, back to 3/8 outlet. Cut impressions on the top to drop in as many peltiers as possible with some thermal paste and holes cut for CPU coolers to mount to. Idk, just spitballin ideas.
 
#9 ·
Peltier devices are awesome things. And yes, you can definitely use them to get the coolant in your loop to be lower temperature -- even below ambient if you're willing to put enough peltier devices and power delivery into play.

But, it won't work like you think. Here's how your really need to do it:

CPU Water Block ======= peltier "cold side" water block |||||| peltier "hot side" water block ==== huge radiator loop

The "|||||" in the diagram is the peltier device itself. It has a hot side and a cold side. The "cold side" you circulate coolant past and it will chill the liquid. The "hot side" you need to circulate coolant past, which gets really hot and needs a lot of radiator space to cool.

Ideally you also need a temperature sensor on the cold side and a feedback circuit that measures the temperature and adjusts current to the peltier devices so that they keep your coolant "cool" but above the dew point. Going below the dew point is bad - because you get condensation. Also in the feedback circuit should be an alarm to alert you if the coolant gets too hot on the cold side, e.g. if the peltier hot side becomes saturated or the device fails, etc.

There are commercial TEC powered chillers that do exactly what you're asking. Not cheap.

Greg
 
#10 ·
If you used a 150w~ peltier then you would get some measurable results. It would need its own loop though to cool down the hot side wattage plus the CPU wattage.
Quote:
Originally Posted by hammong View Post

Peltier devices are awesome things. And yes, you can definitely use them to get the coolant in your loop to be lower temperature -- even below ambient if you're willing to put enough peltier devices and power delivery into play.

But, it won't work like you think. Here's how your really need to do it:

CPU Water Block ======= peltier "cold side" water block |||||| peltier "hot side" water block ==== huge radiator loop

The "|||||" in the diagram is the peltier device itself. It has a hot side and a cold side. The "cold side" you circulate coolant past and it will chill the liquid. The "hot side" you need to circulate coolant past, which gets really hot and needs a lot of radiator space to cool.

Ideally you also need a temperature sensor on the cold side and a feedback circuit that measures the temperature and adjusts current to the peltier devices so that they keep your coolant "cool" but above the dew point. Going below the dew point is bad - because you get condensation. Also in the feedback circuit should be an alarm to alert you if the coolant gets too hot on the cold side, e.g. if the peltier hot side becomes saturated or the device fails, etc.

There are commercial TEC powered chillers that do exactly what you're asking. Not cheap.

Greg
This
smile.gif


Also keep in mind it will need to be significantly larger then that 80w unit if you are overclocking. I would start at 150w. Then keep in mind that your hot side loop needs to be able to dissipate the CPU plus the peltiers worth of heat, so we're talking at least a triple rad for just a CPU and single TEC.
 
#11 ·
Yes, listen to Puck. He knows what he is talking about ! A second rad would help your Delta even more, also look at the type of cpu block that you are going to use. If you are going to try a lower flow in that loop I may be inclined to use a pin type block for max cooling surface area.
 
#12 ·
Why is it when there's an amazing coincidence, it never involves me getting free money?

In this case, the coincidence is that not two hours ago, I ordered the parts to do the exact thing you're talking about!

I bought a cheap waterblock and this righteous mondo heat sink:



It's TEN INCHES WIDE(!)



Those fins are 5" tall.



Why so big? Because I have a 420 watt, 24v, 60mm industrial peltier. I had to buy a 24v power supply.



I'm going to attach the TEC to the asskick heatsink, and put the waterblock on the cold side, just like you said. Then I'll divert my loop through it.

The 5" x 9" fins will have a strong desk fan blowing on them. None of this 12v rinky-dink crap!

The flow will be: CPU --> 3x Radiator --> 2x radiator --> pump --> through the TEC to chill out --> CPU

Everything is all set up like that already and working fine. I just need to add the TEC to give the water one more cooling stage before pumping it into the CPU. Since there's a chance the water may go below 0ÂşC, I'm using premixed car antifreeze as coolant, with a silver coil as a biocide.

The pump is on the cool side of both radiators because last time, I tried putting the TEC on top of the CPU, and the near-boiling water from the hot side of the 420-watt TEC melted the impeller of my previous pump. (I have an 8-core 9590 (220 watt TDP) that I run at 5GHz always.) This time, the water goes through the rads before it gets pumped. That part works great right now.

Will the TEC fight the rads? No way! I don't know why you guys say that. Are you putting the TEC after the CPU but immediately before the rads? I would certainly hope not!

The water coming out of the CPU will be blistering and go right into five 120mm rads before it gets to the pump and TEC again. The thing I like about this setup is that if the TEC forms a block of ice or snow, I don't care. It is nowhere near the motherboard. And if the ice melts, it will melt on the floor.

It's also expandable. If it does work, the idea is to put a second TEC and waterblock on the cubic-foot heatsink. And maybe a third. My goal is to feed subzero icewater into my CPU waterblock, then see how far past 5GHz I can run it.

Your idea (our idea) is a great one, because it decouples the TEC thermal rating from the CPU TDP. Even a tiny little TEC would make the water colder. A big one, or adding more TECs, will just make the water colder still.

It's also fail-safe in that if the TEC or 24v power supply fails, the system will run hotter, but it will still be watercooled.

My final enhancement will be to put the rads inside a cheap window A/C unit I bought for that purpose. I can't try it yet because I don't have enough tubing. I ordered more though!

-- faye the experimenter,

who, in just five years, killed two motherboards, three TECs, and two CPUS f%#king around with things God never intended man to f%#k with
 
#13 ·
Quote:
Originally Posted by FayeKane View Post

Will the TEC fight the rads? No way! I don't know why you guys say that. Are you putting the TEC after the CPU but immediately before the rads? I would certainly hope not!

The water coming out of the CPU will be blistering and go right into five 120mm rads before it gets to the pump and TEC again. The thing I like about this setup is that if the TEC forms a block of ice or snow, I don't care. It is nowhere near the motherboard. And if the ice melts, it will melt on the floor.

It's also expandable. If it does work, the idea is to put a second TEC and waterblock on the cubic-foot heatsink. And maybe a third. My goal is to feed subzero icewater into my CPU waterblock, then see how far past 5GHz I can run it.

Your idea (our idea) is a great one, because it decouples the TEC thermal rating from the CPU TDP. Even a tiny little TEC would make the water colder. A big one, or adding more TECs, will just make the water colder still.

It's also fail-safe in that if the TEC or 24v power supply fails, the system will run hotter, but it will still be watercooled.

My final enhancement will be to put the rads inside a cheap window A/C unit I bought for that purpose. I can't try it yet because I don't have enough tubing. I ordered more though!

-- faye the experimenter,
who, in just five years, killed two motherboards, three TECs, and two CPUS f%#king around with things God never intended man to f%#k with
That is one serious heatsink, lol.

Unfortunately though, TEC's do not work like that. The bolded+italicized part in question is the issue.

It is very hard to explain TECs without many pages, so this is a simplified version...

Coolant in a loop cannot be seperated into "sections". It may have some temperature differential for the first 30 seconds or so until a full lap of coolant has been made, but after that the water leaving the CPU will be the same temp as the water leaving the radiator since the whole loops temp will balance out and settle.

Now, when you add a TEC to the loop, it moves its rated qmax of heat from one side to another. It can never move more then that amount of heat no matter what. This means that even though your TEC hot side is completely isolated from your loop, the cold side is still only moving the wattage it is rated for...so if you have an 80w TEC installed in your proposed loop, and are running a 120w OC'd processor, you will end up with HOTTER coolant then without the TEC. It will move its 80w of heat, have nowhere to put the extra 40w, which will slowly heat up the TEC itself. This will warm the cold side, and where there is no longer a delta between the hot and cold side since it cannot physically move anymore heat (called "stalling" the TEC), you start having your cold side hotter then your coolant. Even if the cold side is still cooler then the hot side, it is warmer then your coolant so now,you are heating your loop instead of cooling it!

With YOUR loop though, and your powerful TEC, you will not have the stalling issue since it can move the wattage the CPU will put out.

Undersized TEC's do not work becuase they are not made just for "helping" - they have to totally handle a heat load. An 80w TEC on an 80w CPU means you accomplished nothing - it will balance out and have no delta, and be the same as watercooling. Anything below that TEC rating will slowly heat up and stall, and anything above that will actually cool it down. This is why we recommend doubling a CPU TDP for your TEC Qmax - with a double TDP TEC the first half of its Qmax is literally just to do nothing, and the second half actually cools it. Not the most efficient, lol!
smile.gif


None of this is dealing with the hot side, which also has to be overbuilt to hand the heat of the CPU plus the heat of all the TECs combined with some headroom on top while maintaining an adequate delta vs ambient...
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top