if getting close to 0 is your goal then I concede that cooling would not be sufficient. At any rate I was tired and forgot to say thanks for sharing. No matter what, the way these diy projects get better and better is by people like you willing to take their time to share. Thanks and I'll be looking forward to some more videos in the future ;)
He's right. The dew point is always lower than the ambient temperature (except in the case of 100% humidity). It's definitely possible to achieve sub-ambient temperatures using a peltier maintained above the dew point.
better to have the motherboard upside down. Condensation should make water fall down with gravity. In warm days you will have more water since warm air hold more water.
Mike, I am a Mechanical Engineer (we design systems like this). Yes, combiz is absolutely correct. And not living 'in a world where heat transfer is perfect' has nothing to do with it. Good water-cool temp is going to average around 50-55 Celsius under load. 10C is going to give you EVERY benefit that 0C would give you except some negligible increase in electric efficiency of the cpu.
I've done some more TEC benching since my last comment. I delided the E2160 and it now goes to 3.8ghz on air. Despite upgrading my TEC to a 15amp unit I can't boot Windows over 3.6ghz. The reason I had an improvement in the beginning was because the old thermal paste under the IHS was so inefficient. My results with a E6600 which is soldered are the sames as with the delid E2160 in that I can't go quite as far as with a decent air cooler (Noctua NH15). Basically once you overcome the thermal limit of the TEC the processor temps instantly jump causing a crash. A 15amp TEC is about as big as you can get to sit on top of the processor so I have nowhere to go. I did try fitting the TEC into the heatsink on top of the cpu contact plate then wedging a small heatsink on top and using metal plates to connect the top of the small heatsink to the fins on the large one. This caused a temp increase as some of the heat from the TEC just heated the larger sink making the cpu hotter even with 140mm fans on it at full blast. I then added a TEC to my radiator on another computer but the heat from the heatsink made the air around the rad warm which caused another small temp increase. Last thing to try is if it will allow a higher overclock on my single core Sempron 145 as that has quite a low thermal output. From a practical point of view though a really good cooler is a much more effective way to go than a peltier.
During Summer only in south EU and south US will you ever get 40%. Regular indoor humidity is around 60%, in north EU and US it's rarely under 50%. During winter i've seen as low as 18% inside in north EU (must be almost constant sub-zero). Besides that you're comment will be very well placed in a video about "waterchiller".
Ookey... So spend the night researching. In old-school pre made heatsink/peltier hybrids the most effecient models was -> first a layer of heatpipe (based in the start of the fins) cooled then the TEC, then yet a layer with other heatpipes (in rear end of fins). The elements used was not high watt and they had a control unit where the aim was keeping the CPU at 25-35c at load. About new CPU's high output, my A6-3670k does at stock need same cooling as my A64 3500+ Manchester, to keep same temp.
i think the best way to go is to bulid a circuit that is always checking the cpu temperature in order to increase/decrease the current going to the peltier unit so you can control the cooling. i havent played yet with a peltier unit but if by decreasing-increasing the current you can control how cold it gets then you can build a circuit in wich you will se it to keep the cpu at 30-40c no matter if its iddling or stressing. the circuit will increase-decrease the current so the peltier wont freeze
So been reading about big watt TECs to find pros and cons. -#-Pros: Can transfer mush more heat # Possible to reach sub-zero at load # Allows higher temps = efficiency curve drops slower -#-Cons: Runs very hot # High power input = large power bill # If hot side gets over 50c (likely) efficiency will be low # You might need a new or separate PSU. To sum up a 120w peltier hotside 55c will transfer around the same as a 80w "HS" 40c. So it's a matter of cooling system can deal with over big watts
By now I have researched a bit and the reason the CPU runs hot is due to the fact that the fan-based cooler can't disperse the heat fast enough. This leads to higher CPU-temperatures. To counter that, the Peltier needs even more power, but this is a weak 60 Watts model, plus Peltier elements have an efficiency of only 10%. This probably lead to a temperature build up on both sides of the peltier element. Water cooling would be the way to achieve a lot better results.
To save space you can replace that large heatsink with a smaller one with a smaller fan mounted on top, such as an NMB-MAT which would keep it cooler than your current setup.
Michael Mikkel Tech I think he does have a point although it would not yeild better results. As the efficiency of the heat transfer is dependant on the temperature difference if the cooler gets to say 90C then the heat transfer rate would be much greater than the same surface area at say 50C. This would be really dangerous for a normal cpu but this had the peltier unit which "pumps" the heat away making the cpu way way cooler than it would be ordinarily. Save space = yes Better results = no Better results than you would get without the peltier = probably. Would be interesting to see how the performance decreases with the W capacity of the heat-sink.
He's right, use an arduino or similar with a humidity sensor to keep the temperature just above the dew point. Of course it provides sufficient cooling because the dew point is always much colder than the acceptable CPU temperature.
This was a great video. I've seen some in internet, related to this topic, this one stands out. Maybe you can give more advice in the construction part of the peltier-heatsink part, which in this vid is already done.
alimentei o peltier em 5v e ficou perfeito. sem gelo na pastilha e temperatura de 27c, num a10 7850 4.5ghz com tdp em 65w. estavel. Fiquei muito satisfeito. Proximo peltier sera instalado na VGA, alimentando na corrente das ventoinhas.
google search:"peltier multistage" should answer your question. the highes i found was a 4 layer/stage peltier. almost all of them have the decreasing size towards the cold end so i guess that does indeed increase the max temparature difference.
You're all right :) But MichaelMikkelTechs (and my) goal is sub ambient measured inside the CPU at load. So the goal for the TECs cold side is sub-zero at load. That's because even whit water cooling a semi OC'ed100w CPU will be delta +30-40 (water temp 21C) = 51-61C. So to hit sub ambient (delta 0C) TEC cold side must be cooler then -10c
btw, you will get much better temperatures at load if you get a peltier that has a higher wattage rating than what your cpu is rated for. you will be able to transfer more heat, while also getting closer to the ideal 70 degree C temperature differential.
It is, google it. Miniinthebox has a 72w/12v (usage) unit in stock. It'll move 50w at "hot side 25c". However looking over TEC's spec sheet (google) it shows over 40w heat emission at 12v with not actual heat transfer present, add in the 50w power transfer *1,2 (resistant lose = more heat. Resistant is different from one the the other tec) = 100w heat emission. 100w + CPUw of heat you'll need to move so fast the hot side remains at 25C(max 35c) els you'll have virtually no effect.
Nice project. Maybe a condensation/humidity sensors working with temperature sensor and voltage controler will do the work. The goal is to keep the peltier cell under control.
I like the idea on using the motherboard port to help control the peltier cooler like a second fan port that would possibly be controlled with or without added extra power. thermostat status.
To answer your question - A Voltage difference moves electrons in the peltier element creating a temperature difference, hence cooling. To answer your PSU statement: I was not saying your PSU cannot handle it, only that peltiers can unsuspectingly draw more power to cool (Wattage) than you would expect if inadequate voltage from the PSU and damage the peltier (burnout) or PSU. An automatic regulator can help you with cooling. Search for WH7016K
Hai MichaelMikkelTech. So I've figured out how to make a dummy load capable of over 500w load, parts have been ordered. Still haven't ordered TECs since i have zero cash atm. So I'm just gonna start the experiment by building and testing the dummy. Been trying to figuer out if the TECs can be "to small" as you've stated and found that you're right. A 5A unit will not make any significant performance, relative to the CPU heat dispatch ofc
There is a way avoiding condensation. You should check Dew point in Wikipedia, for example, Dew point at 25C and relative humidity 40% is 10C, so if you keep the tamperature higher than 10C, you won't get condensation.
Ever tried hanging the motherboard upside down so the condensation would drain toward the heat sink? Wondering if this might simplify condensation management.
had the same idea so if theres any liquid itll not touch the motherboard or the cpu, but mine has water cooling and a water block setup. not piltier direct to cpu.
Liquid cooling would be ideal in conjunction with peltier plate. I don't know if I like the idea of having one pressed directly against me CPU and having to count on it working constantly. Least with traditional liquid cooling, if peltier fails momentarily, you have a cold reservoir of water to keep things going.
+Thomas Rainier Saballa One way is by controlling the power that goes into the Peltier Module. Instead of going 12V 10A, go with something like 12V 4A or something like that. Other method would be by keeping the hot side hotter (around 70°C) so that the cold side & CPU would stay around 18 - 20°C. If the temperature of the cold side is between 10 - 20 C, you should be fine, no condensation should happen.
great idea im going to try this out just bought 2 tec and try them stacked but with thermal silcone sheild to protect the electronics from condensation i see how it works on old mother board mite post a vid on it for you guys
Use high temp epoxy resin to attach the hot side of your peltier to a heatsink and buy a copper cold plate. Attach this with epoxy to the cold side of your peltier. Dry under pressure for 10hours. Now its one unit. Use a step up voltage regulation circuit adjusted to just above 13V (ebay) (with load it drops) between 12V PSU and your peltier to avoid blowing your PSU. 12 to 20 degrees or lower guaranteed!!
Actually it is important where you pull the power for the peltier from. Doesn't matter if your psu costs 5 or 5000 bucks - if you're pulling 10+ amps these leads will melt and fuse together after a while. I recently did a peltier based intake cooler for my pcs and i ran into a whole bunch of peoblems with the 21 amps for the two peltiers alone.
Feedback: Rig it up so that as the CPU temps drop the power to the TEC wafer is decreased - so that the cold-side of the TEC and CPU never falls below condensation levels. :)
Yeah, you are right. The Peltier element, and the cooling fan, should effectively be PWM controlled by the Mobo CPU fan connector, via a driver circuit of course. Then you'll probably never see condensation at all.
@@CXensation Ok really all you guys coming with these ideas, and are so sure about it. Do it, film it and send it to me. Or else don´t think you know the answer.
@@CXensation Most defintely not - but it´s the same that everyone suggest without giving it a second thought. We had condensation EVEN when we waited to start the peltier until the stresstest was already going. So if we get condensation there how should it be avoided? You can see from the temps that we can´t really lower the power of the TEC otherwise we would not get cooling.
If I recall it corectly, it also says it in the video, I think it is a 60w peltier unit. It is an e4400 cpu - 2 cores 2,0GHz. That wouldn't work good enough. When the CPU gets cold the cold will spread out into the socket and from there to the MB. And then those items would cause condensation.
Okey you're last comment in the video; yeah we're gonna need some extensive testing to figure out how to reach sub-zero or at least close to it. I'll see if i can't figure out how to make a variably dummy load with around 300w limit (i7 IB 4.8 GHz +0.25v). I'm thinking cooling -> heatpipe - water 120 rad - 240 rad - 360 rad. Peltiers -> as many as i can get my hands on, so will start with ordering the two from mini, then what they got on a danish "electrical devises" site.
The benefit is temperature differential; the hotter the heatsink gets, the more efficiently it cools with ambient air. The negative consequence is that the peltier module is extremely inefficient, I'm guessing around 10%. So in order to cool a 100w TDP cpu, you need a 1000W peltier and a heatsink capable of dissapating 1000W. There is a 10 times increase in heat output of the entire system before any benefit is seen by the cpu.
You're looking at the wrong figure; that number is absurd. You're using the thermoelectric module as a heat pump, so the relevant figure is its coefficient of performance (which can be larger than 100%). You can find datasheets for Peltier modules that cite COPs of ~100% at 50ºC.
oh yeah btw "condensation" is temp based but also humidity based. To have condensation at delta -10c, the humidity must be very high! Condensation occurs because the air can only hold 7,62g of water per m3 (humidity 99%), no matter of the air preasure and when you force the air onto a cold surface, the air will contract. That forces the humidity over 100% = water pearls will appear. However at a humidity 50% i've not seen condensation before delta -14-15c
Regarding the thermal paste, you've probably tested too much paste vs still too much paste, cause if you apply it correctly (for this cpu size RICE GRAIN is enough), it makes a difference.
+sanshinron Usually we use a thermal conducting compound but hey, lets go check with rice instead :D Or did you mean the size of a rice grain of TG is enough? :D
Part #1 My first TEC cooling was done ~2000-2002, using an AthlonXP 2100+. My initial method, as yours, was direct contact with the CPU, which I found to be unsatisfactory, primarily due to the condensation/ice buildup, and for me, slopping goop of any sort on my MSI KT266 Pro2 R/U motherboard and my precious CPU were absolutely NOT an option… I did try a few crude methods of dehumidification, none of which were viable long-term, however, so I gave up for a while… ****Continued****
Here are a couple of notes: 1. DO NOT cover you CPU VRM in any kind of material. If you cover it up it will overheat. 2. Use a high quality power supply. The ones under €30 are rated at 550w but will actually blow up at about 250-300w.
I'm using a PMW controller that only costs about $10 US to control the Peltier and regulate the temmperature so that I can keep it just a bove ambient temperature and prevent condensation. Just go to ebay and search for PMW Controllers...they have a ton.
thats why you would put plastic around it, or you could just completely insulate around the TEC and CPU so that nothing that touches air gets cold enough to condense, i would say plastic or hot glue or even ceramic
next time try to stack two peltiers. hot side of one on the cooling side of the second one. big cooler on the heating site of the second one. this should increase the temperature difference a lot if i am not mistaken.
that because of the peltier what is below zero,and processor just heating it up, when processor is 2 grad, the peltier is below. I think you should try to cool a processor with water cooling,and peltier must be in front of radiator, that will make a soundless cooling at all most zero grads,or with +10 with no condesation,or just use a specific material around peltier,thats will not allow air to touch the pelter,couse thats the condensation is all about.
use a thermostat and power controller to throttle the power to the thermoelectric cooler to prevent and condensation, if you control the power you can regulate the heat flow.
Ofc :) But i've seen a handful of ppl setting up peltier cooling solutions for everyday use, in a PC first aimed to be change in a couple of years. Nothing wrong with vaseline, like i wrote silicon is for long lasting setups. Vaseline is way to messy to cover the entire board at vertical mounting.
I would use an Arduino board that could PWM-controll the cooler depending of ambient temperature(dew point) and CPU temperature. You don't need the peltier to cool your rig in Idle, and cooling below ambient is a b*tch. Just lower the CPU temp to a couple of degrees above dew point, and your safe :)
I'm planning something like this too! And also stuck with the question. My idea: Put a copper plate in my water reservoir and 1 (maybe 2) TEC units attached to the copper plate. I don't know if it would give any cooling as the water is not really long in the reservoir... Hope he (or someone else) responds to this.
Why not use the cell to cool the radiator instad of the CPU directly? You can mount a cell in between two disipators, one for the CPU, and one to disipate the hot from the cell. It would be much cleaner, easier and also faster to replace if it burns (cause it will, eventually).
On the socket just use some vaseline, to prevent the condensation on lower side of cpu. But I've seen no one whoose motherboard would cause condensation. Instead I've seen how melted ice from CPU shorted MB, so if you stop the condensation on CPU your MB would be fine. Its just my oppinion... The guy in this video "freon cpu cooling" did the same and had no problems.
Right about that, however the 'moisture build up' that will short a board ain't going to occur in a the situation you described where there is 1 degree temperature difference. And shorting the board is really the only consideration here.
harshal singh Thanks for the feedback, makes us feel great about what we do!Ehm, the vaseline is just there to act as a shield against the condensation, not ot stop it.BTW, We made som new videos please check them out!
Current regulation. You need a way to use the CPU fan control pin to limit the current to the peltier cell. There's a hundred ways to do that, but the best option is probably a programmable microcontroller board, because the efficiency of your cell, the heat generated by your CPU, and the size of your heatsink are going to all make this a very specific application that you'll want full control over. Ideally you want to keep the cell at 70 degrees or so.
Harshal Kallianpur kneaded rubber eraser, surround everything around the cpu except the cpu head, obviously, and then go from there... Best of lucks my friend!
Condensation don't occur when you don't have water. Between the 2 surfaces you have no air = no h2o = no condensation. Delta temp = ambient temperature/ case temp or what ever you wanna call the surroundings. Condensation is when a surface is over delta -10c. Moving air/ fans on that surface will only make the condensation happen faster (unless the fans managed to keep the surface at just over delta -10c)
Is there a way to avoid condensation? Is there a way to have a water collector when the condensation melts. Is there a pacific voltage to use on the peltier to avoid getting condensation. How about the fans voltage? are there other ways to insulate the mother board from water when the condensation melts?
What power Peltier module you use in W? And what CPU are you cooling, how many cores, GHz? I've seen a guy who put his temps below 0 C, on his quad intel... And for condensation just use some thermo insulating foam or wool. Tightly wrap your peltier's sides with it and it will prevent hot air to touch your peltier module, so on it will prevent condensation and ice.
it is necessary to put a thermostat on the peltier plate with a probe, to regulate to tempature bassea 30 degrees celsius .. of this way the temperature does not go down too low and there is no condensation!
Part #2 Next I built my own HSF, from aftermarket units, it employed a vertical tower, and then off to the side heat pipes went to an additional HSF that cooled the TEC (hot side), and allowed the cold side to cool the CPU (Athlon64 X2 6000+), I found that this was not a great solution due to size. Though it worked well, it was cumbersome, sadly I required a lot of space for the aftermarket cooler for my GPU (8800GTS 320), which took up a TON of space, thus preventing both. ****Continued****
Add exterior res. Cool that with pelteir and heatsink on hot side then colder water and no hassle with condensation. One 120 gpu fan and gpu cooling block. Add 240 id doing cpu gpu. Colder than with no peltier.
Why don't you use capillary fiber around the CPU so that it sucks all the water out and away from the boarder? I would love to see you guys try that out in a video, or even try a capillary tubing so that the water flows into a cup or something like that so that we can see how much water there is.
There is no 'usually', you always want to test which side it is that gets hot. The unit I have has the textless side get hot, while various other videos on youtube show units that similarly have the textless side get hot, with others showing the opposite.
Probably better if you use it In a binary liquid loop (Binary aka) 1 Rad loop (hot), 1 CPU / GPU loop (cold), and a unit in the center being the Peltier....
+Delta Destiny The best way of doing this is to sandwich the peltier between two Cpu water blocks and using silicon gel around them to make an air tight seal No air = no water forming.
+Delta Destiny +Michael Mikkel Tech Would this solution actually work ? Disregarding the condensation, will this way be able to cool it better than a regular watercooling setup ?
Take a look at binary geothermal it works in the same way. and i should know, i'm running one right now. (only for my GPU) as my cpu isn't overclocked..
+Raimar Lunardi (Random Sidequest) Yes but this is a proof rather then practical application. Ideally you would keep it at or around room temp as opposed to negative temps. But that requires an extra circuit and temperature probe.
That's not actually true, you just need to use thermal regulation. Your CPU fan speeds up and slows down based on the temperature of your cpu, you would want to regulate the power to the peltier cell to keep the cold side above the dew point (which is pretty low in a air conditioned environment). You could EASILY regulate your CPU temperature to stay right around 70 degrees farenheit with a peltier cell. It would also use far less power that way. You need to know what you're doing, but it's a viable system. I would probably use an Arduino with a motor control breakout board to limit the current to the Peltier cell, and I'd use a smaller heat sink without a big fan attached. Honestly there's not much point in doing this though unless you need a silent PC for audio recording or something, the efficiency is crap.
You sound like a danish guy. The accent, i would bet you are danish. Nice job anyways. I've thought about making something like this. and probably will one day :P
Wouldn't it be a solution for condensation if we apply completely black silicon to the motherboard from the bottom and top? Will the motherboard be damaged if you apply silicone all over it? Even if we put the peltier on the processor first, we draw black silicon around it, then we put a cooler on it.
isn't there any material or special fabric etc.. thay you can mold as a side and top cover of the peltier that absorb condensate and absorb cold ice on all 4 side of the peltier then running to the top cover to cool the hot side of peltier? then no need for heatsink?
if you limit the amount of air that touches the cold parts, then that could cut down on condensation since it is condensing the water from the air. or use liquid cooling, and use alcohol not water, it wont freeze and will still pump around the system. if you used water it could freeze and block the flow, or you could use mineral oil its less flammable, well not like alcohol. maybe put a fire retardant in the liquid . or even use mercury but it carry's current and could short stuff out but it wont burn
Hi. Liked the idea. Any consequences for ice formed at peltier border? Ice melting and turning to water straight on the mainboard doesn't seems good to me. Please let me know about it.
Why not have a mix of both the Peltier and air or water cooling so instead of cooling the flowing water with fans use Peltier as a cooler for the water and the water now cools the CPU wouldn't that be safer or much more applicable?
I love your selfmade wood mountings!
Thanks :D
if getting close to 0 is your goal then I concede that cooling would not be sufficient. At any rate I was tired and forgot to say thanks for sharing. No matter what, the way these diy projects get better and better is by people like you willing to take their time to share. Thanks and I'll be looking forward to some more videos in the future ;)
the red wire on a molex connector is not 12v, it's 5v. yellow is 12v on all your psu's connectors. that's the whole point of colour coding the wires.
+Maltfalc orange is 3.3 too :-D
Make a PWM circuit to control the TEC power and keep the temp 2-3F above ambient, then you won't have any condensation problems.
He's right. The dew point is always lower than the ambient temperature (except in the case of 100% humidity). It's definitely possible to achieve sub-ambient temperatures using a peltier maintained above the dew point.
This guide was very helpful and well-edited. Thank you
+iwashungry Thank you :D
prevent condensation by overclocking
Gam3rsCentral said a guy who recently lost his sense of humour
completelyunknown1 LOL
Once it's toasty again, up the TEC watts.
You just don't understand.
better to have the motherboard upside down. Condensation should make water fall down with gravity. In warm days you will have more water since warm air hold more water.
you guys speek better english than most people
+dj red hare aus Thanks :D
Mike, I am a Mechanical Engineer (we design systems like this). Yes, combiz is absolutely correct. And not living 'in a world where heat transfer is perfect' has nothing to do with it. Good water-cool temp is going to average around 50-55 Celsius under load. 10C is going to give you EVERY benefit that 0C would give you except some negligible increase in electric efficiency of the cpu.
I've done some more TEC benching since my last comment. I delided the E2160 and it now goes to 3.8ghz on air. Despite upgrading my TEC to a 15amp unit I can't boot Windows over 3.6ghz.
The reason I had an improvement in the beginning was because the old thermal paste under the IHS was so inefficient. My results with a E6600 which is soldered are the sames as with the delid E2160 in that I can't go quite as far as with a decent air cooler (Noctua NH15).
Basically once you overcome the thermal limit of the TEC the processor temps instantly jump causing a crash. A 15amp TEC is about as big as you can get to sit on top of the processor so I have nowhere to go.
I did try fitting the TEC into the heatsink on top of the cpu contact plate then wedging a small heatsink on top and using metal plates to connect the top of the small heatsink to the fins on the large one. This caused a temp increase as some of the heat from the TEC just heated the larger sink making the cpu hotter even with 140mm fans on it at full blast.
I then added a TEC to my radiator on another computer but the heat from the heatsink made the air around the rad warm which caused another small temp increase.
Last thing to try is if it will allow a higher overclock on my single core Sempron 145 as that has quite a low thermal output. From a practical point of view though a really good cooler is a much more effective way to go than a peltier.
During Summer only in south EU and south US will you ever get 40%. Regular indoor humidity is around 60%, in north EU and US it's rarely under 50%. During winter i've seen as low as 18% inside in north EU (must be almost constant sub-zero). Besides that you're comment will be very well placed in a video about "waterchiller".
Ookey... So spend the night researching. In old-school pre made heatsink/peltier hybrids the most effecient models was -> first a layer of heatpipe (based in the start of the fins) cooled then the TEC, then yet a layer with other heatpipes (in rear end of fins). The elements used was not high watt and they had a control unit where the aim was keeping the CPU at 25-35c at load. About new CPU's high output, my A6-3670k does at stock need same cooling as my A64 3500+ Manchester, to keep same temp.
Man thats finally some cool homemade project, totally want to see more of that *-*
i think the best way to go is to bulid a circuit that is always checking the cpu temperature in order to increase/decrease the current going to the peltier unit so you can control the cooling. i havent played yet with a peltier unit but if by decreasing-increasing the current you can control how cold it gets then you can build a circuit in wich you will se it to keep the cpu at 30-40c no matter if its iddling or stressing. the circuit will increase-decrease the current so the peltier wont freeze
So been reading about big watt TECs to find pros and cons. -#-Pros: Can transfer mush more heat # Possible to reach sub-zero at load # Allows higher temps = efficiency curve drops slower -#-Cons: Runs very hot # High power input = large power bill # If hot side gets over 50c (likely) efficiency will be low # You might need a new or separate PSU. To sum up a 120w peltier hotside 55c will transfer around the same as a 80w "HS" 40c. So it's a matter of cooling system can deal with over big watts
By now I have researched a bit and the reason the CPU runs hot is due to the fact that the fan-based cooler can't disperse the heat fast enough. This leads to higher CPU-temperatures. To counter that, the Peltier needs even more power, but this is a weak 60 Watts model, plus Peltier elements have an efficiency of only 10%. This probably lead to a temperature build up on both sides of the peltier element. Water cooling would be the way to achieve a lot better results.
To save space you can replace that large heatsink with a smaller one with a smaller fan mounted on top, such as an NMB-MAT which would keep it cooler than your current setup.
No it would not.
You need a big heatsink (as in how many w's it can cool) to do this.
Michael Mikkel Tech I think he does have a point although it would not yeild better results. As the efficiency of the heat transfer is dependant on the temperature difference if the cooler gets to say 90C then the heat transfer rate would be much greater than the same surface area at say 50C.
This would be really dangerous for a normal cpu but this had the peltier unit which "pumps" the heat away making the cpu way way cooler than it would be ordinarily.
Save space = yes
Better results = no
Better results than you would get without the peltier = probably.
Would be interesting to see how the performance decreases with the W capacity of the heat-sink.
He's right, use an arduino or similar with a humidity sensor to keep the temperature just above the dew point. Of course it provides sufficient cooling because the dew point is always much colder than the acceptable CPU temperature.
This was a great video. I've seen some in internet, related to this topic, this one stands out.
Maybe you can give more advice in the construction part of the peltier-heatsink part, which in this vid is already done.
alimentei o peltier em 5v e ficou perfeito. sem gelo na pastilha e temperatura de 27c, num a10 7850 4.5ghz com tdp em 65w. estavel. Fiquei muito satisfeito. Proximo peltier sera instalado na VGA, alimentando na corrente das ventoinhas.
google search:"peltier multistage" should answer your question. the highes i found was a 4 layer/stage peltier. almost all of them have the decreasing size towards the cold end so i guess that does indeed increase the max temparature difference.
You're all right :) But MichaelMikkelTechs (and my) goal is sub ambient measured inside the CPU at load. So the goal for the TECs cold side is sub-zero at load. That's because even whit water cooling a semi OC'ed100w CPU will be delta +30-40 (water temp 21C) = 51-61C. So to hit sub ambient (delta 0C) TEC cold side must be cooler then -10c
btw, you will get much better temperatures at load if you get a peltier that has a higher wattage rating than what your cpu is rated for. you will be able to transfer more heat, while also getting closer to the ideal 70 degree C temperature differential.
the hot side depends on the polarity, invert it and the hot will get cold and viceversa
It is, google it. Miniinthebox has a 72w/12v (usage) unit in stock. It'll move 50w at "hot side 25c". However looking over TEC's spec sheet (google) it shows over 40w heat emission at 12v with not actual heat transfer present, add in the 50w power transfer *1,2 (resistant lose = more heat. Resistant is different from one the the other tec) = 100w heat emission. 100w + CPUw of heat you'll need to move so fast the hot side remains at 25C(max 35c) els you'll have virtually no effect.
this is very cool. i was thinking about the same thing when i was doing a report on thermocouples. nice video :D
Nice project. Maybe a condensation/humidity sensors working with temperature sensor and voltage controler will do the work. The goal is to keep the peltier cell under control.
I like the idea on using the motherboard port to help control the peltier cooler like a second fan port that would possibly be controlled with or without added extra power. thermostat status.
Well, this video is cool, you guys are brave in experimenting stuff with your own computer :D and the result is awesome!~!
+Donny Slander Thank you :)
Please give us a subscribe, that will make us do more videos :)
+Michael Mikkel Tech haha totally forgot to sub. Thanx for the reminder, I was on youtube spree hahaha done and done :D
+Donny Slander Thank you! :D
No problem mate, u guys are awesome im waiting for videos hahah Michael Mikkel Tech
To answer your question - A Voltage difference moves electrons in the peltier element creating a temperature difference, hence cooling. To answer your PSU statement: I was not saying your PSU cannot handle it, only that peltiers can unsuspectingly draw more power to cool (Wattage) than you would expect if inadequate voltage from the PSU and damage the peltier (burnout) or PSU. An automatic regulator can help you with cooling. Search for WH7016K
You should do a revamp with the new tec units you guys got in, Love the video!
Thank you!
I lost the TEC's when I moved, but perhaps we will buy some new ones :)
Hai MichaelMikkelTech. So I've figured out how to make a dummy load capable of over 500w load, parts have been ordered. Still haven't ordered TECs since i have zero cash atm. So I'm just gonna start the experiment by building and testing the dummy. Been trying to figuer out if the TECs can be "to small" as you've stated and found that you're right. A 5A unit will not make any significant performance, relative to the CPU heat dispatch ofc
There is a way avoiding condensation. You should check Dew point in Wikipedia, for example, Dew point at 25C and relative humidity 40% is 10C, so if you keep the tamperature higher than 10C, you won't get condensation.
Ever tried hanging the motherboard upside down so the condensation would drain toward the heat sink? Wondering if this might simplify condensation management.
had the same idea so if theres any liquid itll not touch the motherboard or the cpu, but mine has water cooling and a water block setup. not piltier direct to cpu.
Liquid cooling would be ideal in conjunction with peltier plate. I don't know if I like the idea of having one pressed directly against me CPU and having to count on it working constantly. Least with traditional liquid cooling, if peltier fails momentarily, you have a cold reservoir of water to keep things going.
+shaun lastra that is also logical, just cooling the water in reservoir. hmmm how to not cause condensation on the cold plate, any idea?
+Thomas Rainier Saballa One way is by controlling the power that goes into the Peltier Module. Instead of going 12V 10A, go with something like 12V 4A or something like that. Other method would be by keeping the hot side hotter (around 70°C) so that the cold side & CPU would stay around 18 - 20°C. If the temperature of the cold side is between 10 - 20 C, you should be fine, no condensation should happen.
5haun 1 bro I think work and don't go so cool so there will make no ice
great idea im going to try this out just bought 2 tec and try them stacked but with thermal silcone sheild to protect the electronics from condensation i see how it works on old mother board mite post a vid on it for you guys
Use high temp epoxy resin to attach the hot side of your peltier to a heatsink and buy a copper cold plate. Attach this with epoxy to the cold side of your peltier. Dry under pressure for 10hours. Now its one unit. Use a step up voltage regulation circuit adjusted to just above 13V (ebay) (with load it drops) between 12V PSU and your peltier to avoid blowing your PSU. 12 to 20 degrees or lower guaranteed!!
Actually it is important where you pull the power for the peltier from. Doesn't matter if your psu costs 5 or 5000 bucks - if you're pulling 10+ amps these leads will melt and fuse together after a while. I recently did a peltier based intake cooler for my pcs and i ran into a whole bunch of peoblems with the 21 amps for the two peltiers alone.
Feedback:
Rig it up so that as the CPU temps drop the power to the TEC wafer is decreased - so that the cold-side of the TEC and CPU never falls below condensation levels. :)
Yeah, you are right. The Peltier element, and the cooling fan, should effectively be PWM controlled by the Mobo CPU fan connector, via a driver circuit of course. Then you'll probably never see condensation at all.
@@CXensation Ok really all you guys coming with these ideas, and are so sure about it.
Do it, film it and send it to me. Or else don´t think you know the answer.
@@MichaelMikkelTech No more suggestions from here, you are certainly the pro and have set the defined standards.
@@CXensation Most defintely not - but it´s the same that everyone suggest without giving it a second thought.
We had condensation EVEN when we waited to start the peltier until the stresstest was already going.
So if we get condensation there how should it be avoided? You can see from the temps that we can´t really lower the power of the TEC otherwise we would not get cooling.
@@MichaelMikkelTech It was nothing but a suggestion. If that is how you react, then - as I wrote ^
If I recall it corectly, it also says it in the video, I think it is a 60w peltier unit.
It is an e4400 cpu - 2 cores 2,0GHz.
That wouldn't work good enough. When the CPU gets cold the cold will spread out into the socket and from there to the MB. And then those items would cause condensation.
Yes, you should apply thermal paste on both sides.
It is really important that you do so!
Okey you're last comment in the video; yeah we're gonna need some extensive testing to figure out how to reach sub-zero or at least close to it. I'll see if i can't figure out how to make a variably dummy load with around 300w limit (i7 IB 4.8 GHz +0.25v). I'm thinking cooling -> heatpipe - water 120 rad - 240 rad - 360 rad. Peltiers -> as many as i can get my hands on, so will start with ordering the two from mini, then what they got on a danish "electrical devises" site.
Subscribed!! you guys are awesome! keep the cool vids coming!
The benefit is temperature differential; the hotter the heatsink gets, the more efficiently it cools with ambient air.
The negative consequence is that the peltier module is extremely inefficient, I'm guessing around 10%. So in order to cool a 100w TDP cpu, you need a 1000W peltier and a heatsink capable of dissapating 1000W. There is a 10 times increase in heat output of the entire system before any benefit is seen by the cpu.
Wrong (said in trump voice)
You're looking at the wrong figure; that number is absurd. You're using the thermoelectric module as a heat pump, so the relevant figure is its coefficient of performance (which can be larger than 100%). You can find datasheets for Peltier modules that cite COPs of ~100% at 50ºC.
Sweet. Hope they can fix that. Would be awesome. Thanks for the video.
oh yeah btw "condensation" is temp based but also humidity based. To have condensation at delta -10c, the humidity must be very high! Condensation occurs because the air can only hold 7,62g of water per m3 (humidity 99%), no matter of the air preasure and when you force the air onto a cold surface, the air will contract. That forces the humidity over 100% = water pearls will appear. However at a humidity 50% i've not seen condensation before delta -14-15c
Nice idea, the chiller is the easiest way of using TEC units.
I would try with two Q150w peltier units.
Im building one of these as well, but I am using pwm to control the current given to the peltier to try to eliminate condensation
The red wire in the Molex is the 5V
In ATX PSUs, always the yellow is the 12V
Regarding the thermal paste, you've probably tested too much paste vs still too much paste, cause if you apply it correctly (for this cpu size RICE GRAIN is enough), it makes a difference.
+sanshinron Usually we use a thermal conducting compound but hey, lets go check with rice instead :D Or did you mean the size of a rice grain of TG is enough? :D
Part #1
My first TEC cooling was done ~2000-2002, using an AthlonXP 2100+. My initial method, as yours, was direct contact with the CPU, which I found to be unsatisfactory, primarily due to the condensation/ice buildup, and for me, slopping goop of any sort on my MSI KT266 Pro2 R/U motherboard and my precious CPU were absolutely NOT an option… I did try a few crude methods of dehumidification, none of which were viable long-term, however, so I gave up for a while…
****Continued****
Here are a couple of notes:
1. DO NOT cover you CPU VRM in any kind of material. If you cover it up it will overheat.
2. Use a high quality power supply. The ones under €30 are rated at 550w but will actually blow up at about 250-300w.
I'm using a PMW controller that only costs about $10 US to control the Peltier and regulate the temmperature so that I can keep it just a bove ambient temperature and prevent condensation. Just go to ebay and search for PMW Controllers...they have a ton.
thats why you would put plastic around it, or you could just completely insulate around the TEC and CPU so that nothing that touches air gets cold enough to condense, i would say plastic or hot glue or even ceramic
next time try to stack two peltiers. hot side of one on the cooling side of the second one. big cooler on the heating site of the second one. this should increase the temperature difference a lot if i am not mistaken.
that because of the peltier what is below zero,and processor just heating it up, when processor is 2 grad, the peltier is below. I think you should try to cool a processor with water cooling,and peltier must be in front of radiator, that will make a soundless cooling at all most zero grads,or with +10 with no condesation,or just use a specific material around peltier,thats will not allow air to touch the pelter,couse thats the condensation is all about.
I think what you are refering to is a chiller.
And that works great! But will still cause condensation if build properly.
awesome guys... hats on
use a thermostat and power controller to throttle the power to the thermoelectric cooler to prevent and condensation, if you control the power you can regulate the heat flow.
Ofc :) But i've seen a handful of ppl setting up peltier cooling solutions for everyday use, in a PC first aimed to be change in a couple of years. Nothing wrong with vaseline, like i wrote silicon is for long lasting setups. Vaseline is way to messy to cover the entire board at vertical mounting.
did you overclock the cpu?
what was the frequency?
if you didn't then 50 degrees celcius isn't that great a result
No we did not.
And yes, the temps where high.
We will hopefully make a new video some time soon.
Seems a Coolermaster Hyper 212 would do this job very well. Any advices on how to dry out the water from condensation? Subscribing!
I would use an Arduino board that could PWM-controll the cooler depending of ambient temperature(dew point) and CPU temperature. You don't need the peltier to cool your rig in Idle, and cooling below ambient is a b*tch. Just lower the CPU temp to a couple of degrees above dew point, and your safe :)
@MrTombstoneMedia yea, a really good cooling! You should cool the hot side of the Peltier with the water loop that would make a killer CPU cooler!
I'm planning something like this too! And also stuck with the question. My idea: Put a copper plate in my water reservoir and 1 (maybe 2) TEC units attached to the copper plate. I don't know if it would give any cooling as the water is not really long in the reservoir... Hope he (or someone else) responds to this.
Why not use the cell to cool the radiator instad of the CPU directly? You can mount a cell in between two disipators, one for the CPU, and one to disipate the hot from the cell. It would be much cleaner, easier and also faster to replace if it burns (cause it will, eventually).
+Yagami Clan Dude If you had a watercoolingloop that would be possible.
On the socket just use some vaseline, to prevent the condensation on lower side of cpu. But I've seen no one whoose motherboard would cause condensation. Instead I've seen how melted ice from CPU shorted MB, so if you stop the condensation on CPU your MB would be fine. Its just my oppinion... The guy in this video "freon cpu cooling" did the same and had no problems.
Right about that, however the 'moisture build up' that will short a board ain't going to occur in a the situation you described where there is 1 degree temperature difference. And shorting the board is really the only consideration here.
Yes, we know this. We have a 150w Peltier in the workshop that we tested but we did not film that. We had roughly 15-20 degrees Celsius under load.
this video was awesome... any other way to prevent condensation other than vaseline?? and... more hardware tinkering videos plz...
planning to do this...
harshal singh Thanks for the feedback, makes us feel great about what we do!Ehm, the vaseline is just there to act as a shield against the condensation, not ot stop it.BTW, We made som new videos please check them out!
Current regulation. You need a way to use the CPU fan control pin to limit the current to the peltier cell. There's a hundred ways to do that, but the best option is probably a programmable microcontroller board, because the efficiency of your cell, the heat generated by your CPU, and the size of your heatsink are going to all make this a very specific application that you'll want full control over. Ideally you want to keep the cell at 70 degrees or so.
Silicon spray
Harshal Kallianpur kneaded rubber eraser, surround everything around the cpu except the cpu head, obviously, and then go from there... Best of lucks my friend!
Condensation don't occur when you don't have water. Between the 2 surfaces you have no air = no h2o = no condensation. Delta temp = ambient temperature/ case temp or what ever you wanna call the surroundings. Condensation is when a surface is over delta -10c. Moving air/ fans on that surface will only make the condensation happen faster (unless the fans managed to keep the surface at just over delta -10c)
Is there a way to avoid condensation? Is there a way to have a water collector when the condensation melts. Is there a pacific voltage to use on the peltier to avoid getting condensation. How about the fans voltage? are there other ways to insulate the mother board from water when the condensation melts?
I guess that would be a viable option!
And easy to apply!
OMIGOD TAKE IT EASY WITH THE THERMAL PASTE MAN! YOU AREN'T SPREADING MAYO ON A SANDWICH.
or like watching an American make a Vegemite sandwich! you only need a little bit :)
Never have never will, although if I ever do I'll keep that in mind.
What power Peltier module you use in W? And what CPU are you cooling, how many cores, GHz? I've seen a guy who put his temps below 0 C, on his quad intel... And for condensation just use some thermo insulating foam or wool. Tightly wrap your peltier's sides with it and it will prevent hot air to touch your peltier module, so on it will prevent condensation and ice.
Depends on unit design, they are not all the same.
it is necessary to put a thermostat on the peltier plate with a probe, to regulate to tempature bassea 30 degrees celsius .. of this way the temperature does not go down too low and there is no condensation!
And you won´t get good enough cooling.
Nice, I always wondered about doing this to a computer, glad I didn't... it would be really hard to keep that condensation off it.
well the peltier can run only as cool as the hot side allows it, so you could monitor the temps and adjust the heatsink fan accordingly
SlinkiestPopi Still wouldn't Work. To cool the CPU sufficiently you are going to need to reach temperature below dewpoint.
Michael Mikkel Tech really? hmmm... you could limit the voltage and prevent as much dew as possible
SlinkiestPopi attaching adjustable resistor,to find a sweetspot maybe?
Dave Chippy maybe being able to control the voltage to the peltier
Part #2
Next I built my own HSF, from aftermarket units, it employed a vertical tower, and then off to the side heat pipes went to an additional HSF that cooled the TEC (hot side), and allowed the cold side to cool the CPU (Athlon64 X2 6000+), I found that this was not a great solution due to size. Though it worked well, it was cumbersome, sadly I required a lot of space for the aftermarket cooler for my GPU (8800GTS 320), which took up a TON of space, thus preventing both.
****Continued****
Add exterior res. Cool that with pelteir and heatsink on hot side then colder water and no hassle with condensation. One 120 gpu fan and gpu cooling block. Add 240 id doing cpu gpu. Colder than with no peltier.
Thanks!
We have just startet studying so we have almost no spare time, but tell us what you'd like to see and we will try and get it done :)
So, if i was to use a TEC unit with a water cooling block, I would guess you could get a good cooling going?
what temperature does it run with the same heatsink and no cell?
Why don't you use capillary fiber around the CPU so that it sucks all the water out and away from the boarder? I would love to see you guys try that out in a video, or even try a capillary tubing so that the water flows into a cup or something like that so that we can see how much water there is.
There is no 'usually', you always want to test which side it is that gets hot. The unit I have has the textless side get hot, while various other videos on youtube show units that similarly have the textless side get hot, with others showing the opposite.
Answer:
1) It could be a problem for some, for us it was not a problem.
2) We are 100% confident that this PSU could keep up.
How about solder wick it mite suck up water and if you heat it up the other end it mite help evaporate the water, its worth a try.
Probably better if you use it In a binary liquid loop (Binary aka) 1 Rad loop (hot), 1 CPU / GPU loop (cold), and a unit in the center being the Peltier....
+Delta Destiny Hitting the head on nail :D Finally ;)
+Delta Destiny The best way of doing this is to sandwich the peltier between two Cpu water blocks and using silicon gel around them to make an air tight seal No air = no water forming.
+Delta Destiny +Michael Mikkel Tech Would this solution actually work ? Disregarding the condensation, will this way be able to cool it better than a regular watercooling setup ?
Take a look at binary geothermal it works in the same way.
and i should know, i'm running one right now. (only for my GPU) as my cpu isn't overclocked..
Delta Destiny
I know the basics behind it, just wondering if it would lower the temp from my current setup..
Already on a full costum water setup..
Couldn't this be used to maintain around 30-40 full load instead of negative temps??
+Raimar Lunardi (Random Sidequest) Yes but this is a proof rather then practical application. Ideally you would keep it at or around room temp as opposed to negative temps. But that requires an extra circuit and temperature probe.
one thing everyone forgets is something call dew point and how that will cause moisture. you can't use a peltier in this kind of environment.
Yes we can, we just get condensation.
you can't afford to have condensation so close to electronics like this. this setup is way too impractical
That's not actually true, you just need to use thermal regulation. Your CPU fan speeds up and slows down based on the temperature of your cpu, you would want to regulate the power to the peltier cell to keep the cold side above the dew point (which is pretty low in a air conditioned environment).
You could EASILY regulate your CPU temperature to stay right around 70 degrees farenheit with a peltier cell. It would also use far less power that way.
You need to know what you're doing, but it's a viable system. I would probably use an Arduino with a motor control breakout board to limit the current to the Peltier cell, and I'd use a smaller heat sink without a big fan attached. Honestly there's not much point in doing this though unless you need a silent PC for audio recording or something, the efficiency is crap.
Richard Garnache seems not vaseline but transparent silicone
+houston we have a problem
yes you can... see my comment*
You sound like a danish guy. The accent, i would bet you are danish. Nice job anyways. I've thought about making something like this. and probably will one day :P
Crispy Logi 4:32 he said he's danish ;)
Isalys555
didn't even notice :P
Wouldn't it be a solution for condensation if we apply completely black silicon to the motherboard from the bottom and top?
Will the motherboard be damaged if you apply silicone all over it?
Even if we put the peltier on the processor first, we draw black silicon around it, then we put a cooler on it.
Very cool
isn't there any material or special fabric etc.. thay you can mold as a side and top cover of the peltier that absorb condensate and absorb cold ice on all 4 side of the peltier then running to the top cover to cool the hot side of peltier? then no need for heatsink?
if you limit the amount of air that touches the cold parts, then that could cut down on condensation since it is condensing the water from the air. or use liquid cooling, and use alcohol not water, it wont freeze and will still pump around the system. if you used water it could freeze and block the flow, or you could use mineral oil its less flammable, well not like alcohol. maybe put a fire retardant in the liquid . or even use mercury but it carry's current and could short stuff out but it wont burn
Have a look at Vigor Monsoon II Active TEC CPU Cooling System
Hi. Liked the idea. Any consequences for ice formed at peltier border? Ice melting and turning to water straight on the mainboard doesn't seems good to me. Please let me know about it.
did you see all the nasty shit we put on the MBO?
yea that is to seal it off from water.
Thanks your bro :) İntel i7 6700K Peltier Cooler System
What if to off the fan completely? What will be the temperature, or when will it reach 100°? At full load.
how do you solve the problem of water, coming from the condensate, which will leak onto the motherboard, only with grease ???
red wires are +5v, even on molex connectors...
yellow ones are always +12v
Why did you cover your motherboard in petroleum jelly?
Why not have a mix of both the Peltier and air or water cooling so instead of cooling the flowing water with fans use Peltier as a cooler for the water and the water now cools the CPU wouldn't that be safer or much more applicable?