Anyone using Peltiers on their res to chill coolant?

I think you tend to over complicate things. http://www.bit-tech.net/hardware/cooling/2010/07/20/hailea-hc-500a-water-chiller-review/7

You could run it in the same loop as your Watercooling like they did in that review. Though what I would do is run two loops off the same reservoir. One with my pc components and one with just the hailea chiller. The larger the reservoir the less often it would need to turn on to actually cool down the water.
I'm actually planning my current build around the ability to change my cooling setup on a whim through quick disconnects, so I may buy one of these myself at some point.

SeeThruHead said:

I think you tend to over complicate things. http://www.bit-tech.net/hardware/cooling/2010/07/20/hailea-hc-500a-water-chiller-review/7

You could run it in the same loop as your Watercooling like they did in that review. Though what I would do is run two loops off the same reservoir. One with my pc components and one with just the hailea chiller. The larger the reservoir the less often it would need to turn on to actually cool down the water.
I'm actually planning my current build around the ability to change my cooling setup on a whim through quick disconnects, so I may buy one of these myself at some point.

Click to expand...

The bit-tech test shows the flaw of these things : noise
these things are designed to cool large bodies of water, not to be constantly hammered by 1/4 gallon heating up all the time. Introduce a (very) large reservoir and you've got a thermal buffer. The chiller will still kick in and will run longer, but it will take longer BEFORE it kicks in the FIRST time.
Once equilibrium in 50 gallons is reached, the chiller will kick in almost as often as if it were just 1/4 gallon. But you can let it run to cool the res while you are away (and pc is off).
The question is how many hours of hard gaming or folding are needed to heat up 50 gals above the set or ambient temperature? If you are lucky, it will never kick in while you are in session

In order to avoid issues & contaminated distilled water, you should use two loops and a plateHX. Well you dont really have to if you dont want to, but ONE loop will have drawbacks or issues.

And nobody start about the extra "cost" of a second loop... If you can afford a $ 700,- chiller, you can spare $100 for an extra pump & some tubing

Not being rude or anything - The quote system isn't working for me for the last couple days. The idea of using many gallons of coolant and a big aquarium chiller seems fun but sort of against the point of this setup. Part of the intention is to have a system that can run near/around or below ambient while still being 'mobile' enough that I can move it easily. Given that even 25 gallons of coolant would weight over 200 pounds / 95KG, it means that I would either have to have the rig on some kind of mobility platform or I would have to toss the distilled water in order to convert it to a mobile configuration and then fill it back up again in the new place. Just seems like to much of a pain in the ***. That's really not that big of a deal in an everyday, real life, practicality kind of way since it's not really necessary to move a machine around often but since this will/would be a fun project meant to be a challenge with a certain outcome I can't justify using all that big volume moving parts equipment. For the same or less money I can get a much smaller and easier to deal with system with somewhere around the same outcome, with much less noise. This much of the concept is up for debate, though. I don't know if I can actually make a rig that is smaller than a large microwave that can achieve ambient or sub-ambient without the use of lots of coolant and noisy moving equipment.

Negatron said:

I don't know if I can actually make a rig that is smaller than a large microwave that can achieve ambient or sub-ambient without the use of lots of coolant and noisy moving equipment.

Click to expand...

No, you can't.

Pretty sure that's what we've been trying to tell you...

So you now see the reality. Engineers deal with this every day once you have the knowledge. Some stuff is just crazy and won't work. You're slowly gaining the knowledge, possibly .05% of why it won't work.

Granted, we'd love to see an elegant cheap solution, but you ain't got the background to make it happen.

Welcome to the real world of real physics and cost vs benefit.

Please excuse me if you've already given this information, I speed-read the thread and didn't see it.
I'm feeling vaguely inspired at the moment for some reason and am full of ideas, but first I need to know:

1) How much below ambient are you looking for?
1.1) Would at ambient work for you?
1.2) Are you willing to gunk your motherboard/GPU/GPU/etc. up with vaseline for wetness protection?

2) What all are you hoping to cool? CPU? GPU? RAM? All of the above?
2.1) Exact model numbers are needed (4770k? 4570k? G930? etc.)
2.2) How much OCing are you planning here?

3) What PSU do you have?

4) What's your budget?

5) How much ongoing power consumption are you willing to deal with? (I realize this is a difficult thing to come up with a number for, probably the best way to answer it is in $/month of power. I'll need your kWh costs too if you go that route)

6) How much DIY are you willing to do?
6.1) If limited, how much are you willing to pay someone to do things (like, say, active temperature management for the coolant) for you?

7) What's the weight limit, if there is one?

hokiealumnus said:

No, you can't.

Pretty sure that's what we've been trying to tell you...

Click to expand...

It's very likely impossible for you to know that. If you are saying that I can't do it because no one else has done it, no one else has been able to, or no one else has ever tried it then that's all fine and everything but I they are only reasons why it is likely to be difficult. None of those things are reasons why I can't do it.

Conumdrum said:

So you now see the reality. Engineers deal with this every day once you have the knowledge. Some stuff is just crazy and won't work. You're slowly gaining the knowledge, possibly .05% of why it won't work.

Granted, we'd love to see an elegant cheap solution, but you ain't got the background to make it happen.

Welcome to the real world of real physics and cost vs benefit.

Click to expand...

What, then, is the reality? My idea might be crazy for many reasons but that doesn't mean it won't work. So far no one has given me any actual reason why it will not work. Not knowing that it will work or knowing how it wouldn't work is not the same as knowing it won't work. Like I said I'm pretty good with physics; have been ever since I started dreaming about it when I was 5 years old and excuse me, but, how do you know about my background? What I'm saying is, actually, I most certainly do have the background to make it happen. I don't know if you mean to be insulting but I have been in the real world of real physics since I started studying and dreaming about the science when I was in the 3rd grade. Sure, I have no real idea about the cost vs benefit but, again, I already understand and accept the possibility that it will be expensive and won't acheive the same result per dollar spent level of other systems. Then again what prototypes or first attempts at this kind of thing ever do? I mean that much is pretty well a given; obvious to anyone who's ever tried to build something new. At this point I still can't figure out what makes you think I don't have the background or understanding of physics to make it happen........

Negatron said:

What, then, is the reality? My idea might be crazy for many reasons but that doesn't mean it won't work. So far no one has given me any actual reason why it will not work. Not knowing that it will work or knowing how it wouldn't work is not the same as knowing it won't work. Like I said I'm pretty good with physics; have been ever since I started dreaming about it when I was 5 years old and excuse me, but, how do you know about my background? What I'm saying is, actually, I most certainly do have the background to make it happen. I don't know if you mean to be insulting but I have been in the real world of real physics since I started studying and dreaming about the science when I was in the 3rd grade. Sure, I have no real idea about the cost vs benefit but, again, I already understand and accept the possibility that it will be expensive and won't acheive the same result per dollar spent level of other systems. Then again what prototypes or first attempts at this kind of thing ever do? I mean that much is pretty well a given; obvious to anyone who's ever tried to build something new. At this point I still can't figure out what makes you think I don't have the background or understanding of physics to make it happen........

Click to expand...

Fill out the questionnaire above and let's do this.
If you/we do some actual science on the subject there'll be something to point at and say "see?" next time someone asks. Whether it's "See? It works" or "See? It doesn't work" is hard to say. I do know I don't want to pay the power bill.
If you're not the electronics sort I can work up a TEC control box to keep things at ambient more or less. I'll get to work on that nowish.

Bobnova said:

Please excuse me if you've already given this information, I speed-read the thread and didn't see it.
I'm feeling vaguely inspired at the moment for some reason and am full of ideas, but first I need to know:

1) How much below ambient are you looking for?
1.1) Would at ambient work for you?
1.2) Are you willing to gunk your motherboard/GPU/GPU/etc. up with vaseline for wetness protection?

2) What all are you hoping to cool? CPU? GPU? RAM? All of the above?
2.1) Exact model numbers are needed (4770k? 4570k? G930? etc.)
2.2) How much OCing are you planning here?

3) What PSU do you have?

4) What's your budget?

5) How much ongoing power consumption are you willing to deal with? (I realize this is a difficult thing to come up with a number for, probably the best way to answer it is in $/month of power. I'll need your kWh costs too if you go that route)

6) How much DIY are you willing to do?
6.1) If limited, how much are you willing to pay someone to do things (like, say, active temperature management for the coolant) for you?

7) What's the weight limit, if there is one?

Click to expand...

1-1.1 - I'm not all that interested in achieveing temps far below ambient. I think a couple degrees below ambient would be the most I care to achieve and even then I would probably only ever try that for benchmarking. Right around ambient or a little above is how I would want to run it normally.

1.2 - I am willing to use vaseline but I hope to use a more permanent and less messy type of solution after I get it working. I have some ideas but I would have to test them to find out if they are plausible.

2 - I haven't considered cooling more than a CPU so far. I don't know if cooling all of the parts that are normally included in a liquid loop is really possible with the rig I have in mind at least not while still achieving temperatures near ambient.

2.1 - I think if this were going to be used practically I would be using it as an HTPC mostly. The fact that I don't have the time or desire to start on this thing any time soon means I don't think any current CPU models are going to make sense to consider. That being said I don't know which CPU I might use but I guess it will be something on the 'lower end of the upper end', maybe something like an i5 4200.

2.2 - Overclocking depends on how high I have to rev the rig up to keep the temps down. Basically I would want to keep the clock speeds at a level that would still allow me to keep the temps near ambient without using any fans. It might even turn out that I would have to underclock the CPU to get that. I hope not and I don't think it's very likely to turn out that way. In any case, I prrobably wouldn't overclock more than 8% or 10% except for testing and benchmarking.

3 - I don't have a PSU that I plan to use nor do I have a specific model in mind. That's something I'll decide when/if the time comes. If I had to tell you how I think it will work I'd say I will probably use a run of the mill 600-700 watt 80+ bronze modular PSU, as well as an external industrial power supply, and some kind of dev board for control, for the peltiers and any other auxilliary equipment.

4 - I don't have any specific amount of money I am willing to spend or not spend. I don't think it's a project that would cost a huge amount of money. I would think it can be done for $2,000USD on the low end.

5 - There's no getting around the fact that it's going to take a lot of energy to achieve what I have in mind. Again, I can't give you a relevant number because there is so much that I haven't even tried to figure out at this point. I would like to keep it below about 2,000 average watts but who knows how close Ican actually come to that...

6 - The point of doing the thing is to do the thing myself. I would only pay someone to do something for me in the event that I couldn't do it myself. Aside from not owning certain equipement like a cnc laser or a milling machine I can do everything in mind myself. There are always unexpected problems with this kind of thing so I can't say for sure just how much of it I'll need help with.

7 - No practical weight limit. I don't want to use 300 pound copper billets or 40 gallon tanks of water but that much is obvious due to the nature of the rig. What I mean is that the kind of thing I have in mind will naturally come in at an acceptable weight.

@neg : its not that it isn't gonna work... that's the thing, it can/will work... only not as well as you'ld expect. And certainly not very efficient. But as an exercise/project/experience it might (or might not) keep you happily busy.
I'ld say, give it a try, see how it goes.
Note: pix or it didn't happen

Ah well, If we could only come up with a cheap & easy way to turn energy into matter, life would be a lot easier. (1)
Unfortunately, 'till further notice the laws of physics can not be changed in this universe - even if your name is Montgomery Scott.

(1) Incidently, we have managed to turn matter into energy - its called the A-bomb

CPU at/below ambient or water at/below ambient?
I ask because given an i5 2400 (which cannot OC) a decent air cooler will have it running at the same temp a water cooler with ambient water would run.

Bobnova said:

CPU at/below ambient or water at/below ambient?
I ask because given an i5 2400 (which cannot OC) a decent air cooler will have it running at the same temp a water cooler with ambient water would run.

Click to expand...

I wasn't aware of the limitation of the 2400. That's just an example anyway though. I definitely won't be using a CPU that has those kinds of limitations but I might not use an i5 or even an Intel CPU. The motivation of doing the project is not to get at or below ambient CPU temps. I could do that for less trouble. The motivation is just to build something fun and learn from it. The goal would be to get the coldest part of the coolant below ambient. Anything beyond that is bonus. Keeping he rig at or below the size I have in mind and not exceeding the maximum energy consumption that I want excludes keeping the CPU at or below ambient as a possibility. Maybe it will be possible to get the CPU below ambient at idle but operating the machine in a useful scenario while still keeping the CPU below ambient won't be possible I'm afraid. I might under clock the CPU to see just how low of a stable temp I can achieve but I like fast CPUs and so I don't anticipate ever running any of my machines that way for longer than it takes to prove something.

Ok cool, I just wanted to know where the line was. Coolant at or a bit below ambient is the answer to that one.
I seem to have lost my TEC so I'm going to have to get a new one off ebay.

Bobnova said:

Ok cool, I just wanted to know where the line was. Coolant at or a bit below ambient is the answer to that one.
I seem to have lost my TEC so I'm going to have to get a new one off ebay.

Click to expand...

I think the best way to go about this is to use a copper resrvoir, about 2 liters in volume, with one wall being constructed of very heavy gauge plate; maybe even as thick as one inch. In the stream I will have mounted two small CPU heat sinks probably from a pretty old design. They will only need to be of a small size since they will be submerged in liquid but the bigger the cold plate the better and more efficiently it will work. The cold plates/heat pipes of those will be mounted to the inside of the thick wall of the reservoir. On the other side of that wall will be peltiers and you will need to have quite a few of them but just how many I don't know. The important part here is that it's better to have more peltiers and more heat sinks on the hot sides of those. I think that it might be a good idea to mount a big heavy gauge copper plate to the hot sides so that more heat sinks can be used or, maybe if you have the skills, tools, etc, you can cut and bend copper sheet and solder it to that copper plate directly. This is the way that I would do it. It will cost less and work better than mounting a bunch of pre made heat sinks to that plate. Either way the fins of the heat sinks must be oriented parallel to gravity and there must be plenty of air space below and above them so room air can enter from below and move up and float away after it becomes heated. If more thermal conduction is needed, you can add another heat sink to the coolant stream and, theoretically, an unlimited number of peltiers. This is why I think lower capacity peltiers are better for this rig. The actual ideal design, number of heat sinks, volume of coolant, number of peltiers, etc, is something that will have to be hammered out through calculation and trial.

It's possible you could make the plates of the heatsink come through the reservoir and mount the pelts directly onto them. Just need to seal the heatsink base to the res wall maybe with an oring or some rubber. Unless you're going to solder or somehow join the heatsink base the the inside of the res to offer heat transmission between the copper res and the heatsink base.

RnRollie said:

Ah well, If we could only come up with a cheap & easy way to turn energy into matter, life would be a lot easier. (1)
Unfortunately, 'till further notice the laws of physics can not be changed in this universe - even if your name is Montgomery Scott.

Click to expand...

Name dropper.

SeeThruHead said:

It's possible you could make the plates of the heatsink come through the reservoir and mount the pelts directly onto them. Just need to seal the heatsink base to the res wall maybe with an oring or some rubber. Unless you're going to solder or somehow join the heatsink base the the inside of the res to offer heat transmission between the copper res and the heatsink base.

Click to expand...

I did consider the jump in efficiency that direct contact might afford. The extra complexity and number of individual pieces would go way up using this kind of setup though. Really the only practical way that I have thought of to achieve this is to just use a reservoir with 5 sides but this would mean making a whole lot of 90 degree bent heat pipes in order to get the air heat sink in the right orientation relative to gravity since the only side of the reservoir that could be open without all of the coolant falling out is the top side. I think the best way to fix the liquid heat sinks to the reservoir is by brazing. This will eliminate the need for thermal compound and so will eliminate the possibility of thermal decoupling of the heat sink from the reservoir wall.

Of course the other problem with mounting the peltiers directly to the heat sinks is that the surface area of the cold plates of the heat sinks is finite and so would be the number of peltiers that could be mounted to that surface. In order for the system to work quite a few peltiers will need to be used. So either way a conductive material must be used in order to increase the amount of thermally active surface area. I'm not saying the thick copper plate way that I have in mind is the only way to achieve that; it's just the best way I've been able to think of so far.

This is a tec block made by foxrena over on overclock.net (The one on the left) http://www.overclock.net/t/1406591/tec-block-on-4770k/0_100

He's using a peltier sandwiched between a cold plate (CPU side) and a custom waterblock.

Now if you expand that waterblocks size, enough to say mount multiple peltiers to it (with the cold side instead facing the waterblock). Undervolt (can you undervolt a pelt, and if so does it decrease the heat output of the hot side significantly?) them enough that you could dissipate the heat with a large heatsink then I think you would have what you want.

Crude pain cross section:


If you make it square and large enough you could attach it to the side of your case and it would also look pretty neat. Now my main concern about this is that the passive cooling would not be enough to cool the pelts. But the larger you make the block, the bigger the heatsink cooling potential and the more you can undervolt the peltiers.

Negatron said:

Also how come I don't see more people using copper automotive heater cores as radiators for exchanging heat from liquid coolant?

Click to expand...

Cause it doesn't work.

Trying to cool a liquid loop is really hard to do. It'll take two water cooling loops to bring one loop down in temps enough to call it chilling.

TEC # 12715 is about the most powerful TEC you can buy at 40mm or roughly the size of AMD IHS plates. This bad boy you'll have trouble cooling and will take double the rad surface area than you'd expect being capable of producing over 200 watts of heat by itself.

Now lets say you've got a loop chilled with like say 4 or 6 tecs. The only issue with cooling liquid in a loop is that your cooling ALL the other components as well. Meaning your chilling the rads, res pump, tubes and water block at the same time as trying to "chill" or remove the heat from the source IE Cpu. Gotta be the most non effective way I've ever personally tried......

If you do the Math right, anything is possible, but not practical in any way. You'll need a PSU with 20 plus AMPs on a single rail. I can tell you that more than 2 TEC# 12715 will be really hard on any PSU and has the ability to start fires.

Use one TEC right on the heat source and just use the water loop to cool the TEC. Easiest way to go about it and you'll have way less headache. Don't expect to cool any Cpu over 95w effectively. Gotta insulate for condensation just like you would for LN2.

Hope you find TECs as challenging and fun as I do, and don't get discouraged. I've successfully OCed a 6 core FX 6100, http://www.overclockers.com/forums/showthread.php?t=731806 but water cooling and even air would produce the same clock speeds, but Not the same temps and voltages used.