Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › Peltier chilling system
New Posts  All Forums:Forum Nav:

Peltier chilling system - Page 5

post #41 of 167
Thread Starter 
Quote:
Originally Posted by toolmaker03 View Post


yea, I can see that as being a definite issue, seeing as I am using 10AWG wire to power, my 12726's, and I had to shorten the 16AWG wires that are on the TEC's, down to 1", so that they would stop getting real hot.
but I am also running the TEC's at 13.57V and 23.1A steady, at start up there at 14.8V and 25.9A but they drop after a few min.

Wow that's a pretty big drop. I noticed around a 0.4v drop myself after long run-times. Knowing short of nothing on how peltiers work, i think it's due to just that. Heat. The wiring/chip gets warm, increases resistance, and causes voltage drop. That's my guess anyways tongue.gif
post #42 of 167
yea, that is what I thought as well, but at this point I do not have any real heat left in this circuit. so, what is left at this point is the real AMP draw of the TEC, even though it is says that it is a 12726, and for the first 3 minutes of operation it is, but once the TEC gets to temp, the voltage drops, and current draw drops, so that, it runs steady as a 12723 TEC. the power supplies I am using have a smart board in them, so they adjust the output power, based on the circuit that it is attached to.
so, the circuit I just described has a 200A relay, 10AWG to the power supply and the short 1" wires on the TEC's. but my other two TEC's still have the long 4" wires on the TEC's 10AWG wire to the power supply and 40A relays. these two circuits still run at 14.6V and 25.2A steady, and 14.8V and 25.9A at start up. the difference between the two circuits is heat, there is heat in the 40A relay, and there is heat in the long 4" 16AWG wires, attached to the TEC's. that is why the power supply is supplying the extra voltage, and the extra amperage to the these inferior circuits.
so, to summarize the power supplies are still supplying the same 23.1A and the 13.5V to the TEC's themselves. the other 2.1 amps and the 1.1V are used by the inferior circuit due to heat resistance within the circuit. as the resistance in the circuit rises due to heat, the circuit it would need extra voltage, and amperage supplied to it, in order to achieve the same output to the TEC itself.
one can see this truth when overclocking a CPU, a hot overclocked CPU, needs more voltage, than a cool overclocked CPU, at the same clock rate, for it to stay stable.
I am not real good at explaining things so I hope this makes since.
Edited by toolmaker03 - 7/6/16 at 1:50pm
post #43 of 167
I have the thermonamic 19940 669w TEC under a custom water block on my 4670k. The spec sheets are theoretical and the TEC's need to be derated quite a lot, ie my 669w TEC performs more like a ~500w TEC.

15a is about the limit for the 40mm TEC's
post #44 of 167
Quote:
Originally Posted by LiamG6 View Post

I have the thermonamic 19940 669w TEC under a custom water block on my 4670k. The spec sheets are theoretical and the TEC's need to be derated quite a lot, ie my 669w TEC performs more like a ~500w TEC.

15a is about the limit for the 40mm TEC's
one of these days I would like to see a pic of your build, I have a idea in my mind of what it is like, but that is only my imagination.
and yea, I totally get what your saying, mine say there 400W TEC's, but I think that applies to the amount of heat that the hot side can generate, and not the amount of heat that the cold side can handle. that is why I don't even bother with the spec sheets any more, and I only look at TEC's from the perspective of how many amps can it draw. the higher the amps, the more heat it can handle. I know that there not quite that simple, but that is about as good of a understanding that I can get at this point.
post #45 of 167
I've posted pics of my TEC block installed in various threads on this forum. Not sure if I ever posted any with my complete system, I'll try to dig up some of my old pics. I'm planning a rebuild so I can fit the TEC block again, but I'm hoping to migrate to a new case that fits more of my radiators.

Re the TEC specs. Ebay TEC's typically name TEC's very simply, ie 12715 = 127 couples (127 couples means 15v as well), 15 amps, they'll also include a size, like 40mm. Then they trick you by saying 231w, but that is maximum power consumption, not Qcmax, ie the cooling power of the TEC, the Qc is actually about 130w. It is a false idea that you are buying a really powerful TEC when in fact it is just power consumption not heat transfer ability.

My 19940 is closer to its specs than the ebay TEC's, the problem is the spec sheet is extrapolated data rather than actual tested data. ie, they will take a single couple of N and P type semiconductors, test it's performance, and then extrapolate the data based on voltage, couple count etc, so in reality the data is quite optimistic. Customthermoelectrics data sheets are pretty good. My TEC is theoretically 669w of Qc, in reality it's not that good, but still just about the most powerful single TEC that can be bought. The power consumption of this puppy at Umax and Imax is actually greater than 960w, ie ~24v x 40a, so they aren't falsely listing the power consumption as though it is the cooling power (Qc) like the ebay TEC's do, otherwise it would sound even more outrageous.

Re ebay tecs, I typically compare them to customthermo data sheets, ie for 12715, I find the 15.4v 15a 127 couple TEC, and check its data, then derate it about 10-20% to get an idea of how it will perform.

you really need to avoid thinking about more amps = higher cooling ability though, that only really applies if you are stuck with 127 couples because you are limited to 15v, on the customthermo site you will see there are some high voltage low amp tecs that have fantastic possibilities and can be very very efficient
post #46 of 167
Thread Starter 
Quote:
Originally Posted by LiamG6 View Post

I've posted pics of my TEC block installed in various threads on this forum. Not sure if I ever posted any with my complete system, I'll try to dig up some of my old pics. I'm planning a rebuild so I can fit the TEC block again, but I'm hoping to migrate to a new case that fits more of my radiators.

Re the TEC specs. Ebay TEC's typically name TEC's very simply, ie 12715 = 127 couples (127 couples means 15v as well), 15 amps, they'll also include a size, like 40mm. Then they trick you by saying 231w, but that is maximum power consumption, not Qcmax, ie the cooling power of the TEC, the Qc is actually about 130w. It is a false idea that you are buying a really powerful TEC when in fact it is just power consumption not heat transfer ability.

My 19940 is closer to its specs than the ebay TEC's, the problem is the spec sheet is extrapolated data rather than actual tested data. ie, they will take a single couple of N and P type semiconductors, test it's performance, and then extrapolate the data based on voltage, couple count etc, so in reality the data is quite optimistic. Customthermoelectrics data sheets are pretty good. My TEC is theoretically 669w of Qc, in reality it's not that good, but still just about the most powerful single TEC that can be bought. The power consumption of this puppy at Umax and Imax is actually greater than 960w, ie ~24v x 40a, so they aren't falsely listing the power consumption as though it is the cooling power (Qc) like the ebay TEC's do, otherwise it would sound even more outrageous.

Re ebay tecs, I typically compare them to customthermo data sheets, ie for 12715, I find the 15.4v 15a 127 couple TEC, and check its data, then derate it about 10-20% to get an idea of how it will perform.

you really need to avoid thinking about more amps = higher cooling ability though, that only really applies if you are stuck with 127 couples because you are limited to 15v, on the customthermo site you will see there are some high voltage low amp tecs that have fantastic possibilities and can be very very efficient


Yeahhh i can't even find anything other than the standard 127 couple on ebay anyways tongue.gif I think china just has an extreme surplus of these for whatever reason (127's in general)

Thanks for the customthermo link. They have some great stuff there!


I think i may want to try the - 19911-5L31-15CQ

Based on their charts, running at 12v, i'll see anywhere from a 6-8 amp draw and could produce up to 130-160 QcW. Only that price tho :0 $42.50 Ea.
post #47 of 167
Thread Starter 
Quote:
Originally Posted by mrkubanftw View Post

Yeahhh i can't even find anything other than the standard 127 couple on ebay anyways tongue.gif I think china just has an extreme surplus of these for whatever reason (127's in general)

Thanks for the customthermo link. They have some great stuff there!


I think i may want to try the - 19911-5L31-15CQ

Based on their charts, running at 12v, i'll see anywhere from a 6-8 amp draw and could produce up to 130-160 QcW. Only that price tho :0 $42.50 Ea.

Also this chip is a 40mm footprint. Do you think i should even run that high amperage off such a small footprint? My only issue is; i would need to fast track the water-block. Not a big issue. I have it all ready. Just need to order material and machine them up. Is there a demand in particular for TEC guys to get a 62mm or 50mm water-block.
post #48 of 167
40mm TEC's can be pretty much handled by a lot of existing CPU water blocks, so there is zero demand for custom water blocks of that size.

For 50mm, you can use a 50mm copper shim to adapt it to a standard CPU water blocks, but it's not ideal, same for 62mm. In those instances, a custom water block is usually the only really feasible way to do it, or more to the point, to do it well. There is virtually no demand for custom 50-62mm water blocks, only a handful of guys are interested in it, so don't make those size blocks if you plan on selling them. Foxrena and Ultrasonic already do that and they barely sell any, it's just fun.

That 19911 is a good 40mm TEC. No worries at all running that amount of amps through a 40mm TEC, I ran my 12715's up at ~11amps and as long as they are cooled well it is fine. I was using my old swiftech apogee XT water block. If you use 4 of those 19911's you'll need to use a very good antifreeze or a controller. I use these controllers. http://www.ebay.com.au/itm/TEC-controller-for-Peltier-modules-with-PID-7-25V-max-20A-cooling-mode-/231704893134?hash=item35f2b016ce:g:fJQAAOSwiLdWA-wf or these http://www.ebay.com.au/itm/Mini-TEC-controller-for-Peltier-modules-5-30V-max-20A-/231897113286?hash=item35fe2522c6:g:hfMAAOSwQTVWBolp You'll probably need 2 controllers as they can only handle 20amps each. Very good for holding a set temp and saving power when CPU etc is idle.

There is just a huge amount of those 127 TEC's because they are cheap to make and nobody really cares that they don't perform as well as they should. People mostly just buy them to see what TEC's are and what they can do. They use cheap materials, which is why they don't perform as well as they should, but for the price its hard to beat. If you were to use 8 12715's and 8 of those generic ebay water blocks it will perform extremely well. It's hard to say how much better your custom water blocks and expensive tecs would be. Only way to know is to test tongue.gif

I want to build a chiller with up to 8 of these TECs http://www.customthermoelectric.com/tecs/pdf/28711-5L31-06CY_spec_sht.pdf and a 24v Meanwell RSP-1000-24 PSU. I need custom 62mm water blocks and a massive amount of hot side cooling though, so I've shelved the idea for a while. Maybe one day if I build a chill box this will be how I cool it though as I won;t have to worry about condensation in a chill box.
post #49 of 167
Here is a picture of my Custom 62mm TEC block that I bought off foxrena, I'm using it with a 19940 Thermonamic 669w TEC at 12v. Really I need to push this TEC to about 15v to get what I want out of it.

fb9309c8_image.jpeg

It's currently not installed as I was having issues with my RAM, I mounted the TEC block to my CPU without the IHS and socket hold down, it worked perfectly for a while and I was able to OC my 4670k to 5.2ghz, but I rebuilt my loop to add a small air cooled chiller in, and during that I must have bumped the block or applied too much pressure and bent a pin in the socket and I couldn't run my ram at 2600 XMP anymore. I took it off, but for ages I couldn't get my RAM to run at XMP because of the bent pin, as my socket was full of dielectric grease to protect against condensation I couldn't see which one was slightly bent, and the grease was preventing it from springing back into place by itself. After months of just running RAM at 1600 with no issues, the pin is now making good contact with the CPU pads again and I'm able to run my RAM at 2600+ again. Now I am planning a rebuild in a Phanteks Enthoo Evolv ATX and squeezing a 120, 280 and 360mm radiator in, I can't fit the 280mm rad in my old Silverstone TJ09 without a lot of hassle, so I'll retire that and build a hard tube liquid loop in the evolv with slightly more cooling for my hotside by adding the 280mm rad in. Plus it will look a whole lot better and be a lot smaller than my old TJ09.
Edited by LiamG6 - 7/7/16 at 6:21pm
post #50 of 167
Thread Starter 
Quote:
Originally Posted by LiamG6 View Post

Here is a picture of my Custom 62mm TEC block that I bought off foxrena, I'm using it with a 19940 Thermonamic 669w TEC at 12v. Really I need to push this TEC to about 15v to get what I want out of it.

fb9309c8_image.jpeg

It's currently not installed as I was having issues with my RAM, I mounted the TEC block to my CPU without the IHS and socket hold down, it worked perfectly for a while and I was able to OC my 4670k to 5.2ghz, but I rebuilt my loop to add a small air cooled chiller in, and during that I must have bumped the block or applied too much pressure and bent a pin in the socket and I couldn't run my ram at 2600 XMP anymore. I took it off, but for ages I couldn't get my RAM to run at XMP because of the bent pin, as my socket was full of dielectric grease to protect against condensation I couldn't see which one was slightly bent, and the grease was preventing it from springing back into place by itself. After months of just running RAM at 1600 with no issues, the pin is now making good contact with the CPU pads again and I'm able to run my RAM at 2600+ again. Now I am planning a rebuild in a Phanteks Enthoo Evolv ATX and squeezing a 120, 280 and 360mm radiator in, I can't fit the 280mm rad in my old Silverstone TJ09 without a lot of hassle, so I'll retire that and build a hard tube liquid loop in the evolv with slightly more cooling for my hotside by adding the 280mm rad in. Plus it will look a whole lot better and be a lot smaller than my old TJ09.

Very cool bro! Its funny you mention the chill box because i was thinking something like this myself. If you take a look at my OP i listed a link to a thread where i did a mineral oil bath. I half considered chilling a mineral bath like the one i already made. This would prevent condensation from ever forming. And if for some reason it did form on the surface (never had that problem when i chilled my old one down to 0c) The motherboard could be elevated 1-2" off the floor of the tank and since water is more dense than mineral oil, it would sink to the bottom and perhaps could be captured by something and removed. A simple drain tube at the bottom of the tank.

As for the peltier controller. I won't need a proprietary unit. The arduino that i'm using to control this whole system can manage thermostatic PWM control (exactly what the unit you linked does). I have a round about controller already programmed in, ill just need to add two pins for each array of pelts. Y'all really should look into arduino stuff. Awesome digital way to keep track of and control these units.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Peltiers / TEC
Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › Peltier chilling system