Oh a bunch of us did, Cathar's was two pelts. He found he needed more power.
Printable View
Oh a bunch of us did, Cathar's was two pelts. He found he needed more power.
A single loop for the hot and cold sides of the TEC is the biggest FAIL i've heard of for years. it's impossible to get better temps that way. you can only end up with worse temps. due to the fact the TEC uses electricity that gets dumped into the loop as heat. Which means more heat needs to be dissipated by the rad which means higher temps
The whole premise for the single loop idea is floored.
It seems that people are suggesting that if you slow the flow rate down you will see a low temp coming out of the cold side which can then be used to cool the CPU. Resulting in lower CPU temps.
If we exclude the TEC chiller for now. the above will happen if you slow the flow rates down you'll will get lower temps exiting the rads than if you had a high flow rate. but that does not lead to better CPU temps. All it leads to is a greater delta between the hottest and coldest water temps.
If the flow rate is low the water will have more time to remove the heat via the rad for each rotation of the water but the flip side to this is the water has more time to heat up in the CPU so you dont achieve anything the average temp will be the same as if you had the water moving very fast which would result in a very small delta in and out. the reality faster flow would result in lower temps than slow moving water but cause it decrease laminar effect and increases turbulence (not that these factors matter for our argument)
Ambient temp 20c
Now if we add the TEC chiller in the same applies we may be getting say 5c water out the cold side because of the slow flow rate but on the hot side this would have the opposite affect resulting in higher hotside temps say 35c. the average being 20c so there is no point in having the Chiller
Now things are actually worse than that because the TEC chiller uses electricity which gets converted to heat which needs to be dissipated via the rad. More heat to dissipate for the rad means a greater delta from the air to the water temp. Meaning your worse off with the chiller than with it.
It doesn't matter how much heat the TEC's move from one side to the other it's irrelevant.
Single loop Chillers are a fail
Thank you Ultrasonic, we know how conservation of energy works.
However, it DOES work. Sure the radiator takes a bigger hit, but if the radiator can take it, your fine, just wasting electricity for the gains.
Well im not going to argue because i wont get anywhere. However it's illogical ...
Any possible gain on the cold side will be lost on the hot side. = pointless
It can't possibly work and it doesn't.
However you'll probably say it does so we'll just leave it at that. :-)
Your correct, however your viewing the water temperature as a hole and average.
Using this method, part of the water loop will be colder then the other part. A good bit at that. The TEC's use power to create this gradient.
Trust me, taking a holyer then thou attitude won't work, especially with the person who currently HAS THE SETUP IN HAND. And someone who's been in the TEC game a long time.
If we reversed the effect and had a flow rate of infinity and a loop with C/Ws of 0 then the water leaving hot side would = ambient and the temp leaving the cold side would = ambientQuote:
Originally Posted by n00b 0f l337
Since it's one loop what ever effect you apply to the cold side will be applied to the hot side. Meaning no gains with the chiller since it's all one loop and they are canceling each other out.
Im dont believe im taking the holyer than thou attitude i have only debated the Principal that this chiller works on. i have not suggested that people take my view because im better than any one or that i have more experience than any one else. i have debated the facts.
How long someone has been doing TECing should be irrelevant to any discussion and has no bearing on what's right or wrong.
If you want to get into a credibility contest i can hold my own in relation to TEC. However there's no point in doing that. Im probably an unknown to you and your an unknown to me. I would happily debate the principle that this single loop is based on with anyone.
Indeplth TEC knowledge is mostly irrelevant to this debate anyway as the debate we're having revolves around the effects of flow rates.
I really don't understand why someone would want to do a single loop chiller, really it just makes the concept pointless. The only way to make it feasible is if the rad in question was a car rad but even then I couldn't see if working well since flowrate has an overall effect any increase in flow rate increases cooling of hotside and decreases cooling on coldside. There is no right and wrong way to do it, however there is an efficient and inefficient way and this falls into the inefficient category. However whatever you want to do is whatever you want, I'm inclined to say I'm more interested in Ultrasonics design. You have a very nice thread on OCN and I'm actually very amazed at how well you have done. I think this whole post rad concept is going to fail to some extent since anything you apply to one part of the loop will effect any other given part of the loop. That is just how I think.
Of course this is inefficient, we do know that; however technically it has it's upsides.
For instance, the TEC's increase the delta that the radiator operates at, thus increasing its efficiency.
And yes, higher flow rate will bring this system closer to ambient; however you are still operating a heat gap at the processor.
I agree that at low flows you will see a big temp difference between hot side out to cold side out of the TEC.Quote:
Originally Posted by n00b 0f l337
So you'll see a lower temp going in to the CPU block than at high flow rates. This also means the temp exiting the CPU block would be higher than with high flow rates too. Net affect is point less. The same idea applies the the radiator.
Even if the suggested system did work why bother ? Why not go to a duel loop system. one loop with the hot side and rad. and the the other with the cold side and CPU. This way you get tangible results for the rather large investment of chiller TEC's and electricity used. All you'd need is the extra pump.
Martin posted this chiller on OCN and at the time i protested against the proposed plans for it. I tried to sign up to this forum too but that took aaaaggggeeeessss so by the time i could post their wasn't much point as this tread had died.
From the OCN post on this chiller i actually thought the proposed idea was to have 2 loops a hot side and cold side loop but with radiators in both sides which would result in minimal benefits.
Yep, at least 4 x 226W pelts needed for good performance when run at 5V (or about 30% Vmax). Makes for very good CoP. Need a dual-loop setup (which is what I was always running). Single loop setups are for the birds if you want effective results, or unless you're prepared to run about 10x TEC's in serial with about 10x radiators and a very low flow rate, and even then it'd be unspectacular.Quote:
Originally Posted by n00b 0f l337
The biggest issue with TEC's is the limited temp delta across them at the lower voltages. Have been theoretically toying with the idea of a two-stage chiller with 3 loops to reduce temperatures further.
Insulation of tubing is very important for success, and absolutely must not be under-rated in terms of its importance.
yes i agree with that mine attracts condensation badly at low tempsQuote:
Originally Posted by Cathar
Not just for that, but also for better temperatures. The tubing absorbs quite a deal of heat from the environment if not insulated.Quote:
Originally Posted by Ultrasonic2
oh yes for sure
lol...
this project was changed so many times.
At first i wanted to see if it was possibly. I did my own tests, and its very difficult if not impossible to have a rad attached on the same loop. You hit below ambients and the rad takes you back up.
The final config i think i decided for this was to use 4 x 437W pelts, or 5 x 226W pelts, however the heat i calculated would of been massive.
I honestly was expecting this so i did dedicate my oracle station to cool the hot side, and go with cathars original idea:
The hotside will most definitely be cooled by this monster:
http://i125.photobucket.com/albums/p...a/IMG_0119.jpg
I figure that setup is probably good for even the 4 x 437W pelts + load.
All i need back is the TEC blocks, and i think im good to go.
Quesiton tho, what kind of PSU would 4 x 437W Pelts require? That im kinda still scared about.
oh and i know someone is gonna ask with all this money i spent why not just get a phased chiller, the answer to that is i already did. :)
i get bored of things really fast, and i dont know what it is about TEC's.
I just had to give it a shot without sparing any expense.
Question tho, what kind of PSU would 4 x 437W Pelts require? That im kinda still scared about
Well that depends on your goal . presumably you'll be running them at less than imax or umax as you'd be up for 3500 watts ish.
i would take down the entire power grid on my block if i drew that much power.Quote:
Originally Posted by Ultrasonic2
:rofl:
Holy crap NaeKuh, now THAT is some :banana::banana::banana::banana:ed up :banana::banana::banana::banana: you got there :shocked: :D
Question, do the 4 fans in the middle really improve cooling? Wouldn't 1 row of fans between the 2 rads be enough?
The Noobinator brings in the big artillery to pwn mr. aircooling!
And what is cathars idea ?Quote:
Originally Posted by NaeKuh
hot side watercooled.Quote:
Originally Posted by Ultrasonic2
Simple and efficient.
i believe nol had sandwitched 8 tec's on a big heat sink aircooled last.
Yep, tried that. Some bigger pelts on it I had lying around (breaking down your loop tonight for repack to you and will include them).
Want the big ass heatsink? It'll be expensive to ship, but it does work. (It also weighs like 25lbs)
Its an interesting idea to have a self-chilled loop. However, I believe that it would be wiser to just splurge $30 on the extra pump and get a second dedicated hot-side loop.
Fun read though :up:
I wonder what the ideal flow rates would be in a self-chilled system...
well...I don't want to get drawn in to an argument so I'ii say my 2 cents and butt out. Basiclly I am in argreement with ultrasonic.
TEC's are not chillers they are heat pumps and all they do is move heat from one place to another but they are able to move large amounts of heat from an object hence the "cooling effect".
A "self chilled" single loop only works with additional cooling to the TEC ie rads, they are what actually chills the coolant and you need loads of rad cooling to cool both the heat removed from the object plus the heat generated by the TEC's themselves. In reality there is not much point in using a TEC in such a setup you are just taking heat from a point on the loop and putting it back plus the heat generated by the TEC somewhere else in the loop. there can be no "chilling effect" the only chilling is done by the rads. Hence it appears a very inefficient system...not surprising !!
The only correct usage of TEC's is one cold loop ( or direct attachment of the TEC.) and one hot loop. That way you can remove heat from an object or coolant and then disperse the heat in another location that is the only correct usage of TEC's.
well normally with that big sink i was thinkn of paralelling the blocks. and air cooling the hotside.
since id be doubling the TEC's i would think the hotside could be aircooled efficently no?
well it'll never bee cooled as inefficient as water cooling but it maybe about to give you acceptable results . Not that i've seen the heat sink
It really depends on the C rise from ambient that the heatsink has, and how much air is being passed through it but I'd say the temp would be atleast 10C higher than water more likely 15-20C. I suppose that really depends on heatsink surface area and volume of air blown.
If he did not cut down the heatsink, it should deliver better than .04 C/W with good airflow (1100 LFPM or better). That's not quite as good as the best water rig. You will see a 10C rise for each 250W. The better rads will show 5C for 250W (.02 C/W) with good fans and flow rates.Quote:
Originally Posted by flak-spammer
so u guys are telling me to water the hotside with half the TECs?
Or do i paralell and aircool the hotside?
That is the question im asking.
And lets assume i end up getting 5 x 226W tecs. i know im never going to run them at full blast, so what PSU should i be looking at?
The meanwell S600-12?
You need to be more specific about which TECs you are using. Running the 437's at 12V, you will draw about 12A per TEC. That's one big bunch of cooling - around 700W cooling power at 20C dT for all 5, at about 700W in. If you run those at 10V, you need 9A per TEC, 45A total, and get 550W cooling across 20C dT for 450W in. A Meanwell 600-12 would work for that setup.Quote:
Originally Posted by NaeKuh
Are you talking about the 12726 TEC offered on eBay as the '226W TEC'? If so, I would suggest running those at 5V. You will draw about 8.5A per TEC, and get about 60W cooling across 20C dT. So all 5 gives 300W cooling for about 220W in.
That would require a supply which can deliver 45A or so at 5V. Meanwell SP-320-5 has 55A so that would work.
well im getting the stuff back from nol.
so my options are these at the moment.
1. Get 227 x 5 tec's and run them @ 5V and pray my meanwell S320-12 will hold out.
or.
2. Get my pal bei fei to make me matching base plates to those tec's, run them paralell and have 10 x 227W tecs on it.
however i think id need a second psu for the other side.
Here's a better option - get 4 12730 TECs (the eBay '545W' Ebay 545 HERE ) and wire them as 2 series pairs. They will draw about 10A at 6V, so your Meanwell will easily deliver the total of 20A at 12V.Quote:
Originally Posted by NaeKuh
At that level, each TEC will provide about 80W of cooling at 20C dT, for a total of 320W cooling. Power in is only 240W. This setup is also self-regulating to some extent - at lower heat load (say idle or normal use) the dT will not run up that much, because when below 35% Imax, the higher dT curves don't move much power. You will also have good headroom - the cooling at 10C dT is 120W per TEC, for a total of 480W. So you won't 'hit the wall' if power goes up - you will just get a little less cooling.
okey as soon as i get my chiller back i will run those tec's.
i also intend to use 2 400mm res, for volumn.
And i am looking at getting a dweyr t-stat.
Hi Guys...
Why add a whole second loop?
http://nano.ereuland.com/wp-content/...08-300x225.jpg
http://nano.ereuland.com/wp-content/...79-300x224.jpg
Found this online and it seems like with the principles explained in this thread... You don't need a large element cooling your tec because the rad will be helping dissipate the heat...
The post rad theory and every other theory with a rad in the loop with a TEC is flawed.
You see a rad will only chill to ambient then it starts to heat the water up - it doesnt matter whether it's before the TEC or after - as soon as the TEC starts to take the coolant below ambient the rad will try and warm it up again.
The whole point of using TEC's is to get the temp low - why add something to the loop that will have the opposite effect ? Doesn't make sense....
EDIT - OH Lord....I have just re-read your post. The other problem theory is one of only having one loop and passing by both sides of the TEC....Is that that your refering to ? Your asking what's the point of the second loop ?
Well...TEC's aren't chillers per se they are heat pumps moving heat from one point to another - AND THAT IS ALL THEY DO - if you have the loop passing both sides you just remove the heat at one point via the coldside and plonk it back into your loop some place else via the hotside - in short you don't cool the loop at all.
For efficient TEC operation you use the coldside to remove the heat from your coolant,copper block etc. etc. and then either use a second loop to remove the heat using a rad or instead, if the heatload of the hotside will allow, you can use a fan/heatsink assembly. To use a fan/heatsink to cool the hotside you will need to be using either a low (70w) TEC at fairly high power or a heavily undervolted large TEC....The other possible solution would be to use a heatpipe to ferry the heat to a more suitable site for cooling...and those methods my friend, are the only sensible ways to do it.
Oh not this single loop rubbish again.
:rofl:Quote:
Originally Posted by Ultrasonic2
...
Yeah this was scraped later on.. and was decided to run the tec's by themselves..
Originally it was to use the rad as a safety net on the cold side so i wouldn't dip too low into sub ambient.
Later on i changed the plans so i would just control how hot the hotside got with a controller, and by that i could control the cold side.
Naekuh...how's your 4 TEC block coming....or have I missed it somehow ?
i should hopefully get them back from skinnee soon.Quote:
Originally Posted by zipdogso
After that im rebuilding my entire system into nadeshiko with a TMP-200 as the controller.
Meh... i might need another pedistol to support the TEC stuff, or i might just have bei make me a case to house the meanwell, 2 x mcr320's for the hotside + a typhoon bay res.
Man, you all need to work on your reading comprehension, not to mention picture comprehension.
The hot side would be cooled via heatsink(seriously, did you even look at the pictures?). Not water, so it wouldn't be dumping the pumped heat into my water loop. It would be dumping it into the air.
The idea is that my heating elements dump say 300W of heat into my loop. My radiators dissipate about 300W of heat at a delta of about 7C. So if I add a cold side of a peltier moving even 50W of heat out of my water onto a heatsink.
And the whole POINT of doing it like this is so that you _CAN NOT_ get below ambient, and insulation would be unnecessary. Efficiency be damned. I'm just trying to control my coolant temp a little more, and I've run out of space inside my case for radiators, but I *could* fit an 80mm device like this.
Well your idea probably won't work as well as you think. Lets say you wanted a low-ish watt TEC, So I'd say maybe a 12706 so that would be a Qmax of about 50w at 12v, Only problem is that's also about 45-50w of heat dumped with no load. Okay so air cooling doesn't tend to be so effective when cooling high loads so you wanted to dissipate 50w that's about the max that TEC could handle. So you'd end up with a dt of 0 and 100w to dissipate with that heatsink. Lets say room temp was 20C and the heatsink/fan was 0.3C/w so that's 30C over room temp or a cold side temp of 50C. You arn't going to be doing any cooling with that. See the problem now?
Finally, a real answer. Thank you very much for clarifying it for me flak-spammer.Quote:
Originally Posted by flak-spammer
So the moral of the story is the hot side would either require a jet engine fan cooling it, or it would have to be a lot bigger than the guy used as shown for his aquarium(the mcx462 does in fact have a c/w of .29 with a reasonably quiet fan).
2 pictures that really was not clear what it was about ( 1 showed a TEC/heatsink with 2 tubes going to it - water cooling perhaps !!! the other just showed the same unit in situ.) and 40 words which weren'nt particularly clear either...you know what you were talking about but did'nt clarify it to anyone else. A link to the thread you were referring to would of been nice...But on the face of it since your query had nothing to do with this thread you should of started a new one - and clarified yourself a bit better - we aren't flipping mind readers....Quote:
Originally Posted by zoson
Yes...now you have given up a given us a clear indication of your thoughts...Quote:
Originally Posted by zoson
well done FLAK...
To do what you thought but it still wouldnt be great... you would need to use a large TEC undervolted - this would give you a Dt at 50w which would get you a small measure of cooling but you would still have 100w maybe a bit more coming off the hotside - roughly one can always assume the heat coming off the hotside will be 2 to 2.25 times the cooling on the coldside. It is pretty tricky to cool any TEC of a size worth using with a passive heatsink but on that point a fan will AWAYS greatly improve things.
it's a pretty small heat sink tooQuote:
Originally Posted by zipdogso
If you read the first few pages of the thread, this model here is EXACTLY what this thread was originally about.Quote:
Originally Posted by zipdogso
Right, I am planning to have a fan on it. The heatsink is actually fairly large... It's a Swiftech MCX462... Over 450g of copper/aluminum pins. Their site rates it at .29c/W with a 20dBA fan. I had originally thought that I would take a 80W 16v Peltier and just plug it directly into my PSU's 12v rail, to get some better efficiency... But you guys seem to indicate I would be better off taking like a 226W 16v and running it at 7v(12v/5v rail molex thing)?
What I don't understand, is that people used this particular heatsink with 226W 16v pelts undervolted to 12v right on top of CPU's at sub zero temps... This is the EXACT heatsink I have, just it doesn't have the pelt or coldplate.
http://www.swiftnets.com/products/mcx462plusT.asp
Which they rate at:
# Heat dissipation capacity: 70 Watts continuous thermal load to ambient temperature with 80CFM fan.
# Maximum continuous thermal load: 100 Watts
Anyway, I wasn't looking for a huge amount of cooling. Just a little more at high loads.
Ohhh....so I did comprehend you correctly originally then......when I said any theory placing a rad in the same loop as a TEC is flawed....it is, really.Quote:
Originally Posted by zoson
As it happens your plan of an 80w at 12v would get you a highish Dt for 60w cooling .... A voltage drop from 16v (actually 15.6v) to 12v is standard fare because that gets you the best cooling/high input voltage ratio but don't confuse this with efficiency even at this point TEC's are damned inefficient. Efficiency comes at the other end of the voltage scale but it's not a given right and can be difficult to achieve.Quote:
Originally Posted by zoson
If you only want cooling by a few degrees it is not easy (with a high Dt.) but it is more achievable if you use a large TEC undervolted quite a lot - this will give you a low Dt and hence your low temp reduction requirement it will also give you better efficiency (so long as your under 50% - but the actual Dt is critical for efficiency.). BUT to get it all to work you must have very good cooling on the hotside.
Hmm yeah well I think most decent heatsinks with a big fan on will go 100-150w. Probably they count on people not realizing what you have.....they want people to buy the whole unit not just the heatsink...go figure...Quote:
Originally Posted by zoson
Actually in reality undervolting to 12v knocks a chunk off the 226w cooling then there are the inevitable thermal transfer inefficiencies You might be surprised at the amount of cooling some people are getting from 226's....certainly nothing like 226w. I will hanker that those 226w units are not exactly that in the first place, anyway.
The whole point of using a TEC is to get things COLD not just knock off a few degrees - TEC's arent really designed to run like that - If your ambients not too high you could find it easier/cheaper to use a dedicated CPU only water loop.Quote:
Originally Posted by zoson