What the fcuk ??? Looping the hot side to the cool side with TECs means the temps will increase, period.Quote:
Originally Posted by NaeKuh
Printable View
What the fcuk ??? Looping the hot side to the cool side with TECs means the temps will increase, period.Quote:
Originally Posted by NaeKuh
No, you can create a liquid delta ;)
Its done but normally needs much larger systems, I've gotten cold side down a few degrees actually, seems to help temps when the rad can take the heat ;)
If the goal is to achieve and maintain a low temp at the heat source (CPU, GPU, whatever you are cooling) using circulating water to move the heat, then in a steady state model, the water exiting the heat source will be at about the target temp. If that temp is close to ambient, a rad won't do much for you - all it can do is pull out heat inserted after the load (from the pump etc.) . The rad will also keep the temp from going below ambient - it transfers heat both ways. You probably want something like a car thermostat ahead of the rad - route water around it unless the temp is above ambient, to keep the rad from working against you.Quote:
Originally Posted by n00b 0f l337
That's where a post-rad chiller comes in. In order to maintain ambient or even below ambient at the heat source, you obviously need to have the input cooling water below the desired exit temp. If something close to ambient is desired, then even a degree or so is useful.
Martin's original calculation of 395 watt/seconds for 1.5GPM was actually mathematically correct, and as he later stated, the calculation also needs to take into account heat load. But the calculation went south from there...
If we are looking for max efficiency and are using any of the combinations of TECs I have discussed, we will see about 140W transfer. OC'd CPUs can put out more than that, so maybe a higher design center is needed. We can either sacrifice some efficiency, or go with more TECs. But if your average load is less than the TEC transfer, the water will gradually get colder. You can also maintain temp above a lower limit by running water through the rad - that is more simple than varying the TEC power. Another way to control cooling below the desired level is to use the hot side. If you are using air cooling, slow the fans, if you are using WC, slow the pump or shunt the cooling water. Either way, the TEC hot side gets hotter, the resistance rises, the differential rises, and less heat is transfered out. You can achieve very tight temp control with any of those methods.
It is also fairly easy to control TEC input power directly, for low current loads. Essentially running PCM via a big power MOSFET into a low pass filter so the TEC sees a relatively constant voltage.
If I were designing to cool the big heat loads in a state of the art OC'd system (CPU and GPU), I would expect an average heat load of around 250W and upwards of 400W under heavy load. To match cooling without burning power all the time, we could let the power run up if the system starts to get hot, which means using the PCM control.
Let's say we are using the 4 03178 TECs I described earlier. We run them off the 12V line, but with PCM control around output temp at the load, or at the chiller. At idle, PCM sets the TEC voltage at 5 or 6 volts, you pull 140W out, power use is low (65W or so in addition to the heat load), and you are only burning what you need to maintain temp. If the load dumps in more heat, the TEC ramps up as the PCM duty cycle increases, and with the 600W or so transfer available at 12V, I'm sure you will be able to maintain temp if you can supply the current. But 60A or so is serious power, and the PCM system would need to be big too. You'll be burning some serious power at that point though - around 1000W in - so maybe you clamp the upper end at the 50% point, which I think would be around 300W in and 300W transfer (something like 10V, 30A), and let the rad pick up the slack. Your component temp would go up some, but would still be lower than the rad alone.
Here's what I mean:Quote:
Originally Posted by n00b 0f l337
Run a loop on the cold side and another loop on the hot side. Measure the temps in the cold loop vs. TEC voltage. Now run the loops in series. The temp coming off of the cold side (where one would presumably place the water block) will now be wamer than it was at equal volts across the TEC. Therefoe the temps rose.
That's pretty straightforward. In a closed loop system you are pumping heat in to power the TEC, and feeding that heat, plus the transfer heat, back to the rad in the same stream as the cooling water? No matter how you do that, it's hard to see how you get benefit. It is possible to run both hot side and cold side to the same rad, but that also kind of defeats the purpose, since you are asking the TEC to cool its own self generated heat, with a little help from the rad.Quote:
Originally Posted by Clue69Less
I think the sensible thing is just run separate loops - way easier to tune and control.
Ferrotec has some nice on-line curves and you can adjust to compensate for hot side temperature. I've been looking at their 9506/031/600 B which is 03160 spec.
The coefficient of performance (COP) is the amount of heat pumping divided by the amount of supplied electrical power. Max CoP for this unit at 30C hot side is 3.3, and occurs at 7.5A and .6V for a 10C differential. For only 4.5W in, it moves about 15W, which is mighty fine efficiency (but not much cooling).
That's a 55mm size, at a nice fat 4.85mm height. Efficiency drops pretty fast as current goes up, with the 50% point (CoP of 1) at 37.5A with 2.3V required, which draws 86.25W to move 86.25W.
Going to the 2.5 CoP point (15A 1V) moves 37.5W across a 10C differential at 30C hot side. So 5 of those on a 5V supply draws 75W and moves 180W. Using 5 of these with 12V PCM control and a 200W cooling design center, we draw 90W on average, and can move 450W at 12V with a total power in of about 480W. That would probably do the job for keeping even a big OC system with dual graphics nice and cool, and with good efficiency (most of the time)... but at max cooling, the hot side cooling needs to unload the power in and the power moved, close to 1000W. Need a big pump and big rad to handle that. But if we are willing to let the cold side rad do some of the cooling, which means we have to let the outlet temp rise a few degrees, we might decide to clamp PCM at the 1.5 CoP, which provides 350W of cooling for 230W in, and a total heat load of under 600W at max load. That setup does not seem totally crazy - it would maintain below ambient for all but the highest loads, and even then would only go a few degrees above ambient.
For the totally freeze crazy, with a big enough cold side setup, you could run 2 sets of 4 of these on a 12V supply. If you can keep the hot side at 40C, you can drive the cold side close to 0C while moving 360W. You will need to unload close to 1500W, but hey, formula 1 is about performance, not cost...
The graphs are fun to play with - for this TEC they are at http://www.ferrotec.com/products/the...ail.php?id=119
Just looked at those graphs, quite interesting. Its a shame none of the models they have match the ones I'm using.
What I'm interested in, is the massive spike in COP at around 1/8 of Imax. Though there's a huge disadvantage of running any lower than this magic point.
Also, at this point you'd only be pumping around 10w per element (on the type of element I'm looking at) so perhaps a slightly higher current is better in that respect (though still at the peak).
For the closest element to mine, for 20w movement at dT of 10c it will require 7w of power. This is a COP of 2.85 with a hot side of around 33c with cheap air cooling.
As I have 10 peltier modules, I'm looking at cooling 200w load with 70w power to ambient (of course this is best case, there are all the thermal interfaces etc). This seems to match nicely to my previous calculations and real world tests.
When do you expect to have your coolerheater assembled and tested? I hope you have the time to run a range of conditions to reveal its capabilities.Quote:
Originally Posted by Scarlet Infidel
I expected to have it done by Easter, but that clearly didn't happen.
I'm stuck at University now and have a rush of work before the end of term, then its exam time. luckily I find exams easy, so i might be able to get home for a couple of days during the 'revision time' between exams.
I would like to stress that I have made a lot of compromises on the design so it will not perform amazingly, it is more of a proof of concept which I will try and improve on.
Every castle starts with a single stone.Quote:
Originally Posted by Scarlet Infidel
I think you said you have 12709's - you can take the 12712 curves at http://www.ferrotec.com/products/the...ail.php?id=115 and derate by 25% to get an idea of where you should go. You can do a quick double check looking at 12708 and 12710 curves, you should be right in the middle but unfortunately the nice ferrotec calculator doesn't have those values and the thermal enterprises curves are harder to read.Quote:
Originally Posted by Scarlet Infidel
If your hot side is 35C, you have more options for low voltage. You can't achieve better than 2.0 CoP with a differential over 15C.
If the small sinks you describe are kind of 'OEM Standard' solid aluminum, they probably have .7 to .8 C/W rating. So using your numbers, you would be throwing off 27W per element, and would see around a 20C rise. If the sinks are a little better, like a decent sized copper base aluminum fin, you might get .4 C/W, which gets down to the 10C differential level.
Let's look at the 3.0 and 2.0 CoP points. For your units, the 3.0 point puts you at about 1.8A, which with 10C differential and 35C hot side puts you at about 3.7V per unit, pumping about 20W for 6.6W in. That's similar to what you were thinking. You coulkd do 3 groups of 3 in series off of a 12V supply and hit the voltage pretty well.
If you go to the 2.0 CoP, you need a pretty efficient sink - you can't do that CoP with over 15C differential. You'll need about 2.6A and 5.1V, pumping about 26.5W for 13.25W in. So efficiency starts to roll off but you get more transfer, and you are at a nice standard voltage. You have to unload about 40W per unit. That would raise your hot side to 15C over ambient with a .4 W/C sink.
Taking the 3.0 CoP and using the series power scheme off 12V, 9 units would move 180W with a total heat load of 243W. If that meets your cooling needs (that would drop a 1GPM flow about .6C), you might be fine.
Taking the 2.0 CoP and running off 5V parallel, 10 units would move 265W which is respectable, with a total heat load of 400W. That would drop a 1GPM flow about 1C, and if your heat in is lower than 265W, you should also be able to achieve some nice low temps.
dayam uncle jimbo,
You sure know a ton about TEC's.
Well the original plan was having the hotside and cold side seperated. Never in the design did i plan to have them both on 1 loop.
Originally the cold side was to have a radiator. My idea was to take the coolant as closes to ambient as possible, and then use the TEC's to push me slightly into sub ambient.
Well problem came, radiator is no longer that good thing i was hoping for. The capacity for water to hold that much energy was what screwed my thoery up.
So now its going to be a straight TEC setup, no radiator except hotside. The cold side as i said will be connected to a reservoir, where the water inside will be continuously circulated. The water in the chilled reservoir will then be fed to the system.
Or i can just drop that unit inline on the blocks only and run it solo.
If i was to use a lets say 40mm x 2 radiator, you think this would hurt my performance on the cold side? i dont want to dip too far into sub ambient. I really hate preping things for condensation.
Something like this?
http://www.performance-pcs.com/catal...ducts_id=22275
or even this?
http://www.performance-pcs.com/catal...ducts_id=22597
Since you are sandwiching the TECs between a couple of pretty high performance blocks, you are going to be able to move heat around pretty efficiently. Assuming Martin has given you a clean flat mounting surface, and you use AS ceramique to grease it, you can drive the differential to a very low value with a big enough hot side rad. So the question is how much heat, where do you want to put your operating points?Quote:
Originally Posted by NaeKuh
From Martin's photos it looks like you could put up 3 62mm TECs. If we look at the 12730s we have been fooling with, at 5V they will draw about 6A each, so your total is a 90W load. You will be pulling better than 120W out. Almost any $20 ATX supply will drive the 5V at 90W with no sweat.
You have not talked much about your load, but 120W is not much in today's world. This is a big time chiller and I'm assuming you are planning something fairly big under the hood. With no rad, if the load heat exceeds the TEC transfer, you will run the temps up pretty quickly. Might not be a bad idea to have a small rad in the loop with a bang-bang control at ambient, with maybe a half degree of hysteresis to assure the shunt is on before flow thru the rad is cut off.
Several people have mentioned the '437W TEC' that FrozenCPU has available. That is a 19933 unit. Driving it at 12V with the hot side at 25C puts you at about 15A and a 1.6 CoP, for 180W in, and pumps 280W. And that's just one TEC. Not bad but at these power levels still a lot of juice.
Putting that TEC at 5V puts you way down on the curve - 7A - for a total power in of 35W and a CoP of 3 at 5C differential. That would move 105W. Now you are in a position to control things pretty nicely - with 105W in, you pump 315W. But you are so low on the curve that any differential will basically shut down the TEC, so cold side capacity is important.
But that also gives you a way to control the cold side to the temp you desire - if you only want a few degrees below ambient to minimize condensation, then letting the hot side rise a little will reduce the heat transfer and maintain your desired temp. There are a lot of ways to reduce the cooling - reduce the flow thru the rad with a shunt would be one way. Proportional valves are expensive and a pain to set up, but here a simple solenoid valve would work, if the shunt does not shift all of the coolant. You should be able to set that to maintain whatever cold side temp you want.
By the way, my solution to small amounts of condensation is to direct a fan at the part where condensation would occur - that usually keeps things pretty dry. Since the bottom of the CPU or GPU is going to be few degrees above the cooling face, you shouldn't have to worry about hidden condensation. Just be sure there is enough airflow around the water block to keep drips from forming. That should be decent airflow - I use 70MM CPU fans mounted sideways but anything with good velocity will work.
The Thermochill PA160 has .02 C/W with a 130CFM fan, so with a heat load of 100W or so it will rise 2C. But that's a noisy fan. Going to something a little quieter in the 50-70 CFM range pushes the thermal resistance to .03 or .04, which gives a 4C rise. That's good cooling but about all we can allow running the TECs at that low voltage.
A smaller rad would cost you big time - you would have to go to a higher and less efficient voltage to maintain cooling.
I know I said earlier that running the hot side hot is more efficient - that's important at high voltage and big differential, but works against you at these low voltages. We are all learning on this project...
mmm..
okey all i intend to put on this loop im guessing somewhere between 200-215W max.
Q6600 + Chipset.
I already have a meanwell S320 to power these tec's. Nol has it right now and is probably playing with it on them.
Martin did give it a really nice flat surface, his craftmanship is top notch, if you look at the previous pages, around 3-6 you'll see the WIP.
He made the surface large enough to cover 5x40mm or 4x52mm TEC's my main concern was on the massive power draw, so i wanted to keep the wattage low, and see if i could tweek efficiency.
As for controling the hotside, thats easily done. I guess i can connect both loops to a kaze master and have inline probes for both hotside and cold side.
The controller can also control the fans on the radiator and i could adjust as needed. :\
I have one of the FrozenCPU 427 units - I will test it at 4V to see what it does. If it can move around 80W at that low voltage, you will get your 215W with a little to spare, putting three in series across the MeanWell. Looking at the photo with the ruler in it, it sure looks like you can take a 62mm - that translates to 2.44 inches. I realize that the TEC will be out beyond the finned area, but at these low power levels I don't think that is any handicap as long as it is on the copper. Martin drilled the bolt holes clear through, I think - but again with the low power you are running, I don't see much of an issue even if some small parts of the TEC are over the bolt ends.Quote:
Originally Posted by NaeKuh
If that works it will give you low power for sure - that puts you right at the absolute max efficiency point. But remember these are semiconductors, and so when you get right at the trip points strange things can happen.
I'll let you know what I find out with a simple test. <EDIT> see the note below - I'm a lot happer with 55mm TECs given your geometry.
If you are not happy with the bigger TECs, here's another option - http://ferrotec.com/products/thermal...ail.php?id=114 which is a 33710 unit which is 55mm square. It has a nominal voltage of 46.4, so your 12V MeanWell will put it well down on the curve.
At 12V, it draws 2.6A, so heat in is 31.2W. You are at the 2.3 CoP point with a 10C differential, so it will move 72 watts. You have room for 4, so you could move 288W at a cost of 125W. If you want to start out with 3 units, you would move 216W at a cost of only 94W. If I were you, I would just put in 4 and use some hot side control - that way any power excursions up to 288W are handled, and at lower power transferred, your power in goes down. You are at a much more stable point than with the '437' TEC also, which makes the performance a lot more predictable. You should be able to maintain a few degrees below ambient easily.
If you keep the differential around 5C, you will get even better efficiency - that puts the CoP over 3.
That would be a truly awesome setup, and would revise the thinking of a lot of people about TEC efficiency (me included!)
<EDIT> I discovered that you can plug heatsink performance into the ferrotec calculator so i plugged in the PA160 with 130CFM. At 12V, with .02 C/W cooling, that TEC moves 120W! so 4 of them would pump 480W with 125W in. That is serious cooling.
can you link me to the tec's you recomend.Quote:
Originally Posted by Uncle Jimbo
i'll buy them.
Also i was planning on using a MCR320 for the hotside. :T However i could easily get a 480GTX if needed.
The unit is a Ferrotec High Power Thermoelectric Module - Peltier Cooler Model 9500/337/100 B.Quote:
Originally Posted by NaeKuh
Their power TEC page is http://ferrotec.com/products/thermal.../highPower.php
This unit is at
http://ferrotec.com/products/thermal...ail.php?id=114
They have an inquiry form at http://ferrotec.com/products/thermal.../teInquiry.php
You should be able to figure out where there is a distributor or reseller who can supply you with the parts from those links.
The MCR320 has .02 C/W for flow rates 1.5GMP and above. That is the value I used in my model - it should do fine for the hot side. I think it is also less restrictive than the 480GTX.
Okay, ran into a problem, my other pump is leaking like a civ. Got another pump I could borrow Naekuh? Else I think I'm at end of line :(
Gained 4-5C though at the waterblock by running in a single loop, however rad heat output was much higher of course. And actually temps got better with a res then inline, saying that quantity of liquid would help that situation.
This is exactly what i was guessing too. :DQuote:
Originally Posted by n00b 0f l337
Which is why i wanted to paralell a res where one line goes in a loop just to chiller and back, maybe at a slow flow rate, while the other path is high pressure high head going straight to the cpu.
Do you get my logic nol?
I have a DB-1 i can mail out on monday, i can buy a DDC-3.2 and have John at Jab-tech to mail it to you while i mail you the spare XSPC top i have. Or i can send you another D5 in about 2 weeks as soon as dave mails me my sammy chips.
Take your pick.
Another D5 would be good for testing if you can spare the time.
Parallel res though as I tested though really did not work out, not sure why, probably because you lost your temperature differential at the resevoir.
mmm... i was hoping to keep the res colder then the return on the cpu. :TQuote:
Originally Posted by n00b 0f l337
Dumps the heat back rather directly. Also this system seems to work best with LOW FLOW, vs HIGH FLOW.
In your diagram, the res basically acts as a 'capacitor' - it stores energy and stabilizes the system. If it is insulated, and your TEC capacity is greater than the heat load, you can certainly drive the temp below the CPU return.Quote:
Originally Posted by NaeKuh
I did a quick experiment using a setup like this, but with a dummy heat load.
- 1 gallon res, insulated (a Styrofoam portable cooler)
- heat load variable from 0 to 200W
- 100W to 300W TEC cooling in the second loop
- 1.5 GPM in both loops
1. With no heat load, I ran the TEC loop at 200W. Water temp dropped from 25C to 15C in 15 minutes.
2. I set the heat load to 140W. Res temp continued to drop, but much more slowly, reaching 12C in 15 minutes.
3. I set the heat load to 200W. Temp began to rise very slowly, reaching 13C in 15 minutes. That probably reflected pump heat in.
I did the same test, but with the TEC set at 100W. Results were:
1 - water temp dropped from 25C to 20C in 15 minutes with no load.
2 - with 140W heat load, temp started moving up - reaching 23C in 15 minutes.
3 - ran with no heat load until res was at 15C. Ran 200W heat load. temp rose to 20C in 15 minutes.
So the res acts as I would expect - it dampens the rate of change. For a system where you have relatively short spikes of high power, your dual loop design should maintain the average temp for quite a while.
This also opens up an easy way to control the temp - use a bang-bang controller (such as a standard heating control thermostat and relay) to turn the TEC on and off based on res temp. The latent heat will keep the TEC from cycling quickly, and if the TEC transfer power is more than the heat load, you will maintain your load cooling temps nicely.
I tried that using the setup above, but with an old mercury thermostat mounted on a copper plate with most of the plate in the water. The thermostat controlled an air conditioning relay rated at 60A which I happened to have around. That switched the TEC on and off.
With the TEC at 300W and a 200W heat load, and the thermostat set to 18C (the lowest it could go), I powered up the system. The res got to about 17C in a little over 20 minutes, and the TEC switched off. After about a minute, the water in the res got to 18.5C and the TEC switched on. It stayed on for about 2 minutes, the temp went back to 17 C, and the TEC switched off. I ran it for another 20 minutes or so and the duty cycle was very stable, running 2 minutes on, 1 minute off and maintaining temp between 17 and 18.5C with a 200W load.
NaeKuh, send N0L a camera with the pump.