Post Rad Chiller Concept

[n00b 0f l337]

Gotta get some distilled water, doing some wiring.

[NaeKuh]

Well i gave that air thing a thought. However this is what i know about TEC's.

The colder you keep the hotside the cooler the cold side will be.

Martin made these blocks so they kick buttin being free flowing. He also glass beaded the interior, so theres quite a bit of surface area on both the hotside and cold side.

Needless to say, i dont think martin will ever make another pair. He died making mine.


What i intend on doing with this unit. At this point i made the size and gauged the performance and footprint. This unit is almost definitely going on an external box. The hotside will probably be cooled off a 480GTX.

I may drop in another meanwell, and upgrade the TEC's thats if my poor breaker can handle that amperage.

[Scarlet Infidel]

Well i gave that air thing a thought. However this is what i know about TEC's.

The colder you keep the hotside the cooler the cold side will be.

Very true, though I don't think you can assume that water is automatically going to make the hot side cooler than heatsinks.

In the way you have yours set up, I think water is the right way. You have few, relatively high powered peltiers and are using a beefy water cooling radiator.


However, for discussion sake, consider my lower powered project with 10 peltiers and which is designed to be very quiet. Each peltier will be cooling approx 20 watts of heat and adding at least 10 of their own. Lets say 35 watts of heat on the hot side per peltier to be conservative.

Lets say the very cheap basic heat sinks I'm using have a c/w of 0.4. This is based on 2 14cm yate loons. Then the hot side would be approximately 14 degrees above ambient.

Now, if i water cooled them, I would have the thermal resistance of the hot side water block and the thermal resistance of the radiator. A PA120.2 is what I'm making mine to compete against (in terms of size/number of fans) but ill use the figures of a PA120.3 as that would make more sense to use in a chiller.

As my project is going for quiet and cheap id go for yate loons at 7v. But for this ill use the 12v figures. So I would estimate the hot side c/w is approximately 0.035. This has to cope with 350w giving a a hot side temp of 12.25 above ambient. But then we have to take into account the extra pump which may well be 18w (for a DDC) bringing the hot side temp to about 13c above ambient. This is why I'm toying with the 6w DB-1 - especially on the peltier cold side.

So really, for me, using air cooling gives about the same results as water but is smaller, cheaper, less power hungry and quieter. Of course you are using a different type of chiller with different goals and I think you are right to use water, I just wanted to illustrate a point.

NB all figures are wildly approximate and use all sorts of assumptions (such as that both schemes have the same resistance from the TIM).

[NaeKuh]

well here you run into the problem i ran into.

Did you factor the fact that it can take 350W of energy to change 1C in h2o?

[n00b 0f l337]

Each peltier will be cooling approx 20 watts of heat and adding at least 10 of their own.

Flip that

[Scarlet Infidel]

well here you run into the problem i ran into.

Did you factor the fact that it can take 350W of energy to change 1C in h2o?

That doesn't actually make any sense. 350W for how long to change what volume of water?

Edit: are you saying a system like mine is likely to use 350w to get water only 1c better than normal water cooling?

That's not what I'm saying at all. In the example above is was taking 100-150w to keep the water cooling a 200w load at 3-5c below ambient compared to straight water being 5-10c above ambient with the same size and noise constraints.

The above was just comparing how you cool the hot side of a chiller in my specific circumstances, not talking about the merits of a chiller on the whole which I believe can be done effectively and efficiently.

[Scarlet Infidel]

Flip that

Are you saying that as a pessimist (or a realist from experience) or because that's how you think peltiers work? I don't want to suggest you don't know what you are talking about when you probably do, but for the benefit of those that don't know:


When you lower the voltage across a peltier element from its maximum you increase its coefficient of performance. So at low voltages you can pump more energy across the peltier than it consumes itself.

However doing this also has the affect of lowering the heat differential maintained across the element.


The following is in ideal conditions of no load on the cold side (not possible) but the theory is sound:

So my peltiers at the full 15.2v can move 89w of heat (for 0 dT) for 137w power usage. this is a COP of 0.65. Here the max dT would be 67 degrees with no load on the cold side.

If I run them at a third of that (~5v) then the power draw is 13w but the element can pump up to about 40w at a dT of 0 which is a COP of 3. Here with no load the dT is more like 30c.

Once you add in load on the cold side it changes all the figures, but the principle remains the same.

[n00b 0f l337]

No peltier element will always use more then they move. If you drop a 137W usage peltier from 15.2V to 5; You reduce from 9amps to 3amps rougly. That now means 15W consumption. Now your going to be closer to around 14-15W movement, however I've never seen a peltier that will begin moving more then they consume.

[Clue69Less]

No peltier element will always use more then they move. If you drop a 137W usage peltier from 15.2V to 5; You reduce from 9amps to 3amps rougly. That now means 15W consumption. Now your going to be closer to around 14-15W movement, however I've never seen a peltier that will begin moving more then they consume.

A couple of years ago, my boss had me measure efficiency curves for a wide range of TECs. The results were dismal. I ran them over their functional range in terms of voltage and hot side temp. We had excellent lab facilities so the heat flow measurements were solid. All of the advertised efficiencies were totally bogus. Some measured as low as 10%, the good ones around 50% at their best - so basically agreeing with your numbers above. At high load, 40% is good.

[NaeKuh]

That doesn't actually make any sense. 350W for how long to change what volume of water?

Edit: are you saying a system like mine is likely to use 350w to get water only 1c better than normal water cooling?

Im sorry i should of finished it.

It takes 350W @ water traveling 1.5gpm for it to change 1C.

So it takes a signficant amount of wattage to pull this thing off. :T

[Liquid.]

Gah more pics!

I've been lurking for a few months and i like this project the best!

[Scarlet Infidel]

And I was worried id steal focus in this thread...

Im sorry i should of finished it.

It takes 350W @ water traveling 1.5gpm for it to change 1C.

So it takes a signficant amount of wattage to pull this thing off. :T

Still don't agree with what you are saying here, there are too many other variables.

A couple of years ago, my boss had me measure efficiency curves for a wide range of TECs. The results were dismal.

Its concerns me that you seem to have done proper testing that shows such poor results.

What do you do as a job?

No peltier element will always use more then they move.

Despite all this, I still think it is possible. Cathar had a good attempt at it a few years back and I've seen a few other had limited success.


I recently set up a small, very crude, testing setup I had one of these peltiers ata little over 8w cooled by a basic heat sink. On the other side I had a poor water block and a pump rated at 27w but that i measured to draw more like 20w.

I know not all 20w of the pumps heat went into the water, but people generally reckon 60-90% will, and when I put the pump alone in a small loop with a reservoir the temp rose at a rate to suggest it was dumping around 16w.

After all this the water leveled off at 1-2c under ambient. Now, I may well have missed something and all you concerns worry me, but how can I have managed results like this in such a poorly optimised test platform?

[THE JEW (RaVeN)]

I don't claim to know the technical details about TEC's, but my experience with them has shown that compressor-based chilling is just more efficient.

TEC's and their inefficiency seems to be a common thought 'round the 'net:

http://www.coolchips.gi/technology/ccalc.shtml

For comparison purposes, a typical thermoelectric (Peltier) cooler has a Carnot efficiency of about 5-8%. A typical domestic refrigerator compressor has a maximum Carnot efficiency of about 45%.

Pelt Basics:
http://www.overclockers.com/tips45/

http://en.wikipedia.org/wiki/Thermoelectric_cooling

Thermoelectric junctions are generally only around 5–10% as efficient as the ideal refrigerator (Carnot cycle), compared with 40–60% achieved by conventional compression cycle systems (reverse Rankine systems like a compressor). Due to the relatively low efficiency, thermoelectric cooling is generally only used in environments where the solid state nature (no moving parts, maintenance-free) outweighs pure efficiency.

How you got those results, I don't know. Due to the Laws of Thermodynamics, I do know that 8 watts of heat movement + 12-18 watts of heat source != more than 19 watts of heat movement total.

Either the temperature measurement was off, the TEC was improperly rated, or there was something else helping cool the water.

[Scarlet Infidel]

With all due respect

common thought 'round the 'net:

is something that greatly annoys me, it holds back innovation with little real reason. Those efficiency figures are all true, when you run a peltier at its full ratings. If you use its characteristics carefully you can get some good results.

TEC's can easily be made to move twice as much heat as they consume.

Jorlen has just showed a system where 63w of power to the peltiers have given him sub ambient water temperatures with a 100w load.

Ive seen it for myself, Ive run the maths which supports it. Its not witchcraft, you just have to be open minded.

[Liquid.]

Jeez 63w moving 100w?

You gotta link me to something that shows us how to do that.

[Scarlet Infidel]

Right on our doorstep.

[Liquid.]

Right on our doorstep.

I'm still having trouble understanding how he got to move 100w of heat with 63 w pelts.

[n00b 0f l337]

Okay havent gotten a huge amount of time, and away this weekend, but for some reason I've got a leak in a barb. Must not have put it in all the way or something, draining loop, pulling out barb, and adding some teflon tape.

[THE JEW (RaVeN)]

I'm still having trouble understanding how he got to move 100w of heat with 63 w pelts.

Further research shows I had forgotten about COP

However, the linked thread with a near $1000 cost on cheaply attained parts doesn't exactly help TEC's case

If we're all going to go on a COP/efficiency binge, perhaps we should start designing systems ahead of time with a given TEC and heatsink/waterblock. Using the previously linked software below, one could potentially make a decent cooler at a decent cost:

http://www.kryotherm.ru/soft.zip
http://www.kryotherm.ru/?tid=84

I don't think it would be as good/bang for buck as phase, but it would be an interesting project.


Sorry for off topic--looking forward to NoL tinkering

[Liquid.]

Further research shows I had forgotten about COP

However, the linked thread with a near $1000 cost on cheaply attained parts doesn't exactly help TEC's case

If we're all going to go on a COP/efficiency binge, perhaps we should start designing systems ahead of time with a given TEC and heatsink/waterblock. Using the previously linked software below, one could potentially make a decent cooler at a decent cost:

http://www.kryotherm.ru/soft.zip
http://www.kryotherm.ru/?tid=84

I don't think it would be as good/bang for buck as phase, but it would be an interesting project.


Sorry for off topic--looking forward to NoL tinkering

Replace the bolded link with this
http://www.kryotherm.ru/dir2attz/soft.exe

[Scarlet Infidel]

However, the linked thread with a near $1000 cost on cheaply attained parts doesn't exactly help TEC's case

I'm building mine much more cheaply, only time will tell how good it is though.

[Blauhung]

Im sorry i should of finished it.

It takes 350W @ water traveling 1.5gpm for it to change 1C.

So it takes a signficant amount of wattage to pull this thing off. :T

also keep in mind, you are not cooling the water, you are removing heat from the water in a steady state system.

At any reasonable water cooling flow rates, the temperature of the water throughout the entire loop can be considered constant if you've had your system on long enough to reach steady state. Your calculations should not even take into account the temperature change of the water as it moves through individual components on the loop. The pelts will only be removing heat from the cold side that the processor and whatever else you have in the loop creates, and not changing the temperature of the water.. Thus your pelts only have to be rated to be able to handle the heat generated by the components.

The only time the specific heat of the water really enters into any of the modeling that really needs to be done here is during start up if you wanted to calculate the time it would take to reach steady state.

[Clue69Less]

Its concerns me that you seem to have done proper testing that shows such poor results.

What do you do as a job?

I'm trained as a chemist. In grad school, I spent my first two years building interfaces and writing software for computer control of instrumention. Yes, that was a long time ago. In my first real job, I got involved in thermal analysis and helped modernize calorimeters. Fast forward over twenty years and I was building mass spectrometers for measuring airborne pollutants when my boss found out that I'd worked in thermal analysis. So he gave me a pet project of his where he'd been trying to cool a detector cell with TECs. The problem was that the aircraft did not have "air conditioning" until it got up to ~14,000' and because of that, the cooling of the hot side of the TECs was a problem at startup and also when dropping low to do altitude profiles. The detector needed very stable temps. I finally bought a small industrial air conditioner to cool the hot side and the only down side to that was jealousy from other teams on the airplane.

[Uncle Jimbo]

I'm trained as a chemist. In grad school, I spent my first two years building interfaces and writing software for computer control of instrumention. Yes, that was a long time ago. In my first real job, I got involved in thermal analysis and helped modernize calorimeters. Fast forward over twenty years and I was building mass spectrometers for measuring airborne pollutants when my boss found out that I'd worked in thermal analysis. So he gave me a pet project of his where he'd been trying to cool a detector cell with TECs. The problem was that the aircraft did not have "air conditioning" until it got up to ~14,000' and because of that, the cooling of the hot side of the TECs was a problem at startup and also when dropping low to do altitude profiles. The detector needed very stable temps. I finally bought a small industrial air conditioner to cool the hot side and the only down side to that was jealousy from other teams on the airplane.

I have similar experience - but I was maintaining constant temperature for sensors on oil rigs. The gear could end up on a north sea platform in the dead of winter, or in Africa in high summer. We needed to maintain 25C plus or minus 2 degrees for the electronics package to assure the stability of readings.

One thing for sure - in a TEC setup, you will not see better than 40% efficiency and probably a lot less. For high efficiency cooling of water, I believe the best way is to just use standard air conditioning or refrigerator components.

The advantage TECs provide is the ability to provide tight temp control, and also to provide cooling with a small physical size and better packaging options. Going way sub-ambient is not an option with water and messy with phase change, but fairly easy to do with TECs - attached is a shot of a cooler I helped my nephew build which delivered air at around 5C to a 100W heat load with hot side air cooling. But it used nearly 300W to do it, so the air cooler had to unload around 400W. The Blue Orb is the cold side - it provided air at 5C to a Vanessa cooler on the CPU and generally cooled the air in the whole case. The hot side has a giant Melcor sink which has .08 C/W with 120CFM. Not shown is the drip pan <G>. An interesting project, but for the small improvement in temp at the CPU, done a whole lot quieter and with a tenth the power using a small phase change unit and WC...

[Clue69Less]

I have similar experience - but I was maintaining constant temperature for sensors on oil rigs. The gear could end up on a north sea platform in the dead of winter, or in Africa in high summer. We needed to maintain 25C plus or minus 2 degrees for the electronics package to assure the stability of readings.

One thing for sure - in a TEC setup, you will not see better than 40% efficiency and probably a lot less. For high efficiency cooling of water, I believe the best way is to just use standard air conditioning or refrigerator components.

The advantage TECs provide is the ability to provide tight temp control, and also to provide cooling with a small physical size and better packaging options. Going way sub-ambient is not an option with water and messy with phase change, but fairly easy to do with TECs - attached is a shot of a cooler I helped my nephew build which delivered air at around 5C to a 100W heat load with hot side air cooling. But it used nearly 300W to do it, so the air cooler had to unload around 400W. The Blue Orb is the cold side - it provided air at 5C to a Vanessa cooler on the CPU and generally cooled the air in the whole case. The hot side has a giant Melcor sink which has .08 C/W with 120CFM. Not shown is the drip pan <G>. An interesting project, but for the small improvement in temp at the CPU, done a whole lot quieter and with a tenth the power using a small phase change unit and WC...

Word.