|
Post by rushed on Jul 29, 2019 21:12:43 GMT -5
|
|
|
Post by Bones on Jul 29, 2019 22:22:20 GMT -5
Pelters work but aren't very efficient. The design of that one doesn't look very efficient either because, while it may be useful to keep things cool you won't get anything near what a bucket of icewater would give. The coolant would flow fast enough through it there just isn't enough time to chill the water very effectively.
It would work though and yes, it's probrably fine for 24/7 use but if wanting it to use with benching I'd skip it and do the icewater thing instead.
|
|
|
Post by Vinster on Jul 29, 2019 23:03:55 GMT -5
Pelters work but aren't very efficient. The design of that one doesn't look very efficient either because, while it may be useful to keep things cool you won't get anything near what a bucket of icewater would give. The coolant would flow fast enough through it there just isn't enough time to chill the water very effectively. It would work though and yes, it's probrably fine for 24/7 use but if wanting it to use with benching I'd skip it and do the icewater thing instead. I'm with Bones, you'd be better off running straight tap water through a loop and down the drain. that works very well. some people here do it and get great scores/results.
Vin
|
|
|
Post by Bones on Jul 29, 2019 23:22:16 GMT -5
That's what Shrimpy does, nothing wrong with it at all. I prefer to have an external res (No rad) and just stuff it with ice and water as shown in the vid here. warp9-systems.proboards.com/post/19773
|
|
|
Post by ShrimpBrime on Jul 30, 2019 6:04:28 GMT -5
If looking for a 40mm TEC, 12715 is the one you are looking for.
|
|
|
Post by rushed on Jul 30, 2019 6:10:26 GMT -5
If looking for a 40mm TEC, 12715 is the one you are looking for. from my understanding TECs don't work very well under load, but do good for max frequency if i'm just playing around how accurate do you say that is?
|
|
|
Post by dr4g00n on Jul 30, 2019 8:48:51 GMT -5
If looking for a 40mm TEC, 12715 is the one you are looking for. from my understanding TECs don't work very well under load, but do good for max frequency if i'm just playing around how accurate do you say that is? It could work alright if there's no real load. I got a couple 12715's to play with a bit ago, in a cascade config of 5V->12V they kept my 8700K around 0 to 5c at idle with all cores @ 5.1 1.35V. This is with a water loop cooling the TEC's. If your trying to find a cheap way to get some lower temps then your best bet would be to look around for some ratty old window a/c unit and convert it into a waterchiller. I did that with a 5000BTU unit that I got for free, coolant temps usually get down to -25C with it. A 10K BTU one can get down below -30C.
|
|
|
Post by rushed on Jul 30, 2019 10:07:27 GMT -5
from my understanding TECs don't work very well under load, but do good for max frequency if i'm just playing around how accurate do you say that is? It could work alright if there's no real load. I got a couple 12715's to play with a bit ago, in a cascade config of 5V->12V they kept my 8700K around 0 to 5c at idle with all cores @ 5.1 1.35V. This is with a water loop cooling the TEC's. If your trying to find a cheap way to get some lower temps then your best bet would be to look around for some ratty old window a/c unit and convert it into a waterchiller. I did that with a 5000BTU unit that I got for free, coolant temps usually get down to -25C with it. A 10K BTU one can get down below -30C. I'd really like a single stage evaporator set up, but can't really justify dropping like 1000 dollars on one of those... that seems like a good idea too
|
|
|
Post by Vinster on Jul 30, 2019 11:36:43 GMT -5
It could work alright if there's no real load. I got a couple 12715's to play with a bit ago, in a cascade config of 5V->12V they kept my 8700K around 0 to 5c at idle with all cores @ 5.1 1.35V. This is with a water loop cooling the TEC's. If your trying to find a cheap way to get some lower temps then your best bet would be to look around for some ratty old window a/c unit and convert it into a waterchiller. I did that with a 5000BTU unit that I got for free, coolant temps usually get down to -25C with it. A 10K BTU one can get down below -30C. I'd really like a single stage evaporator set up, but can't really justify dropping like 1000 dollars on one of those... that seems like a good idea too
Vin
|
|
|
Post by ShrimpBrime on Jul 30, 2019 15:30:35 GMT -5
If looking for a 40mm TEC, 12715 is the one you are looking for. from my understanding TECs don't work very well under load, but do good for max frequency if i'm just playing around how accurate do you say that is? Everyone has different experiences with X processors used, implementation of the TEC and so forth. I cannot say it will, and at the same time I cannot say it will not. 12nm 2700X before and after with a Cinebench run at 4Ghz 16 threads. TEC cooled by tap to drain (geothermal) loop and a large cold plate over IHS. Plate is 50mmx60mm about 3/16" thick copper. Edit: My experience with Athlon 220ge smaller less cores on 12nm.... was able to remain frozen all the time. Obviously with lower wattage processors, keeping frozen is pretty easy. During bench near end__________After Bench completed__________Idle temps. Note: Voltage of Cpu during bench is accurate.
|
|
|
Post by Bones on Jul 30, 2019 16:25:50 GMT -5
There he goes again......
|
|
|
Post by rushed on Jul 30, 2019 17:55:24 GMT -5
On a side note I did order a peltier just to play around with on low wattage chips basically see what happens there...
Looking for the right pieces of metal to come up basically now... or an a/c unit whatever comes first
|
|
|
Post by Vinster on Jul 30, 2019 21:54:04 GMT -5
where are you located?
Vin
|
|
|
Post by rushed on Jul 31, 2019 2:34:38 GMT -5
where are you located? Vin Between Minneapolis and Omaha middle of nowhere really
|
|
|
Post by Vinster on Jul 31, 2019 6:20:55 GMT -5
where are you located? Vin Between Minneapolis and Omaha middle of nowhere really ah, too far.. I was thinking if you weren't far I'd get you my AC chiller. but it's too big to ship, but that's too far to drive.
Vin
|
|
|
Post by rushed on Jul 31, 2019 7:05:01 GMT -5
Between Minneapolis and Omaha middle of nowhere really ah, too far.. I was thinking if you weren't far I'd get you my AC chiller. but it's too big to ship, but that's too far to drive.
Vin
Are you one of the fellas around the Chicago area? those are probably the closest people to me from the looks of the map
|
|
|
Post by Vinster on Jul 31, 2019 8:18:50 GMT -5
Nope, I'm near the top. Canada just south of Toronto about 2hrs.
Vin
|
|
|
Post by ShrimpBrime on Jul 31, 2019 16:00:09 GMT -5
ah, too far.. I was thinking if you weren't far I'd get you my AC chiller. but it's too big to ship, but that's too far to drive.
Vin
Are you one of the fellas around the Chicago area? those are probably the closest people to me from the looks of the map I'm one of the fellas in the Chicago land area. Which TEC did you purchase?
|
|
|
Post by rushed on Jul 31, 2019 16:01:38 GMT -5
Are you one of the fellas around the Chicago area? those are probably the closest people to me from the looks of the map I'm one of the fellas in the Chicago land area. Which TEC did you purchase? 15A 230w TEC1-12715
|
|
|
Post by ShrimpBrime on Jul 31, 2019 16:07:14 GMT -5
Yea that's a good choice.
Yellow and Black wires on a spare PSU, try not to use the one you're running the system on. Also be sure to feel the wires out. Small gauge wire PSUs can be scary and run hot. Be sure a good ground and even additional cooling for the PSU. I upgraded the fan in my 1000W Antec CP to keep it running cool. No need to try starting fires.... Don't leave it running without your presence either.
|
|
|
Post by rushed on Jul 31, 2019 16:14:08 GMT -5
Yea that's a good choice. Yellow and Black wires on a spare PSU, try not to use the one you're running the system on. Also be sure to feel the wires out. Small gauge wire PSUs can be scary and run hot. Be sure a good ground and even additional cooling for the PSU. I upgraded the fan in my 1000W Antec CP to keep it running cool. No need to try starting fires.... Don't leave it running without your presence either. 16 Ga wire copper strand wire work, or should i get something better?
|
|
|
Post by ShrimpBrime on Jul 31, 2019 16:56:57 GMT -5
Should be OK. The cooler you keep it, the more efficient it gets the less load you'll have. Bucket of Ice water and a pond pump would yield the best results if your doing Direct to Cpu cooling (which I recommend)
TECs and water chilling just does not work. Unless you want to cover some (min) half of the surface area of the water you are cooling. That means super small res, no rads. Small cool running pump too. (for the cold loop). Ambient air coolers on that TEC is impossible. Please don't bother to try. You can fry a processor.
You want a decent sized cold plate. You'll run the system idle for a few minutes to bring down the plate temps. This helps a lot just like it does with LN2 and Dice pots. It takes the cpu longer to heat that plate than if you where to use a TEC right onto the cpu IHS plate.
(sorry, I'll type tips as I think of them just looking to help you have a good experience without trying to learn the hard way)
|
|
|
Post by ShrimpBrime on Jul 31, 2019 22:47:32 GMT -5
I also have some more information for your viewing pleasures.
Calculation of Peltier module clamping force Torque (Т) of the torque wrench may be calculated with the following formula: T (torque) = (2.8 × 10-4 × p × d)/n, (kg × m) where: p is required clamping force (kg) d is screw diameter (mm) n is number of screws (2-4) used for module installation. Example: Installation of a 40х40 mm module (e.g. ТМ-127-1.4-6.0М) requires a total hold-down force of 210 to 240 kg. When two 4 mm screws are used to install a module, torque of 0.11 - 0.12 kg ×∙m must be set on the torquing screw. When four 4 mm screws are used for the same purpose, torque must be reduced to 0.05 - 0.06 kg∙× m. When designing products using thermoelectric modules and if thermal grease or thermal pads is used, it should be taken into account that heat losses result in temperature difference losses of 1.5 - 2.5 °С at the available capacity of 50 W.
Update
table for reference in the above example. 0.11 - 0.12 kg - The 0.11 is inch pounds of force.
Note: The TEC 12715 would be the same torque specs being the same size. _____210 inch pounds over 6/32nds inch diameter screws x4 or x2 config. You would recheck the torque every 24 hours the first three days at least.
|
|
|
Post by rushed on Aug 1, 2019 15:18:43 GMT -5
I also have some more information for your viewing pleasures. Calculation of Peltier module clamping forceTorque (Т) of the torque wrench may be calculated with the following formula: T (torque) = (2.8 × 10-4 × p × d)/n, (kg × m) where: p is required clamping force (kg) d is screw diameter (mm) n is number of screws (2-4) used for module installation. Example: Installation of a 40х40 mm module (e.g. ТМ-127-1.4-6.0М) requires a total hold-down force of 210 to 240 kg. When two 4 mm screws are used to install a module, torque of 0.11 - 0.12 kg ×∙m must be set on the torquing screw. When four 4 mm screws are used for the same purpose, torque must be reduced to 0.05 - 0.06 kg∙× m. When designing products using thermoelectric modules and if thermal grease or thermal pads is used, it should be taken into account that heat losses result in temperature difference losses of 1.5 - 2.5 °С at the available capacity of 50 W. Updatetable for reference in the above example. 0.11 - 0.12 kg - The 0.11 is inch pounds of force. Note: The TEC 12715 would be the same torque specs being the same size. _____210 inch pounds over 6/32nds inch diameter screws x4 or x2 config. You would recheck the torque every 24 hours the first three days at least. Got it! Quarter turn til it breaks, sounds good!
|
|
|
Post by ShrimpBrime on Aug 1, 2019 15:32:39 GMT -5
It might be funny, but you'll find out clamping force makes a difference.
Can't wait to see your results.
GLHF!!
|
|