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

Peltier chilling system - Page 15

post #141 of 167
Thread Starter 


Was able to test the slitting saw OP today. This was a scrap part I just did a few lanes in. This was my main concern going .175" deep on a .02" slot with only .02" of material supporting the rib. It worked pretty well all in all. Thursday I should be able to wrap up this operation but that might carry into the weekend.
post #142 of 167
Wow ... those are deep. Nice work.
post #143 of 167
very nice work, these blocks should be good, low restriction, plenty of surface area, should have low thermal resistance. Did you scrap that one because of the full size plenum making the channels too short? Or another reason? Just thought maybe you machined that base before our discussion re plenum and channels. Either way these blocks will be bada$$!

One thing that annoys me about those pumps is the config of inlet and outlet being opposite on each side. Would be nice if the 2 on the front were outlet and the 2 on the top were inlet, but it can't be done. It might actually be better to use the better performing single outlet and a "Y" fitting and plug the other outlet, depending on how it splits the flow/pressure out of each outlet, not sure how that will work using 2 outlets straight from the pump, whether the equal restriction of blocks/rads will cause each outlet to flow the same or if one won't work so well just because of the fluid dynamics. Have you got a flow meter or 2 so you can make sure each side is getting equal flow? Just need to test 1 pump and block to see how the outlets behave. I know there is usually a recommended outlet when 2 are available.
Edited by LiamG6 - 8/23/16 at 8:51pm
post #144 of 167
Thread Starter 
Quote:
Originally Posted by LiamG6 View Post

very nice work, these blocks should be good, low restriction, plenty of surface area, should have low thermal resistance. Did you scrap that one because of the full size plenum making the channels too short? Or another reason? Just thought maybe you machined that base before our discussion re plenum and channels. Either way these blocks will be bada$$!

One thing that annoys me about those pumps is the config of inlet and outlet being opposite on each side. Would be nice if the 2 on the front were outlet and the 2 on the top were inlet, but it can't be done. It might actually be better to use the better performing single outlet and a "Y" fitting and plug the other outlet, depending on how it splits the flow/pressure out of each outlet, not sure how that will work using 2 outlets straight from the pump, whether the equal restriction of blocks/rads will cause each outlet to flow the same or if one won't work so well just because of the fluid dynamics. Have you got a flow meter or 2 so you can make sure each side is getting equal flow? Just need to test 1 pump and block to see how the outlets behave. I know there is usually a recommended outlet when 2 are available.


Haha funny you say that. I did all the blocks at the same time but that was my first block, and in the middle of machining it i noticed the pocket was larger that the two lanes the cutter cut. I mistakenly programmed off the first model, rather the second revised version we talked about! I quickly reprogrammed it to the second model and continued running blocks.

I don't know how these ports will work on the pump head, its just something i wanted to experiment with. It may react differently under different flow restrictions but i do plan to test each pump circuit. I had seen data on each set of ports being used but that was individual sets of the ports. I plan to do a pump flow test. Ill snag two 5 gallon pails, send each output to a pail, and measure the output over a minute or so. Yes i desperately need flow meters though haha tongue.gif I believe the port options on this block were more intended to target flexibility of routing your hoses. Though i think it can also be used to reduce pump flow restriction, even if i have to run the three outputs to a junction that has two outputs, i could gain all that lack of restriction and evenly distribute it back out. Or maybe two of the ports will flow even enough. I just wont know until i get to test them frown.gif
post #145 of 167
Thought so ;-) At least you caught it, either way it's a very minor difference in performance with the channels at ~55mm or 62mm.

I think with 1 pump using 2 outlets and 2 inlets into the block, the pump pressure is a constant, so each side will get equal pressure, and each side of the block has equal restriction, so it should get the same flow rate, I guess I was just wondering if the different hydraulic behavior would influence the flow rate on the side that performs worse in single outlet configuration. I guess this is similar to GPU blocks in parallel, even though one of the cards has worse flow path, because of the equal restriction and equal pressure from the pump they both get equal flow. So it should be the same for your setup. Easy to test by connecting 2 pump outlets to the block and then a tube from each side of the block outlet into a separate bucket, see which one fills up first tongue.gif It will certainly help with lowering the overall restriction which is always a good thing. Having 2 x 1/2" ID tubes is similar to having 1 x 3/4" ID tube so that is a good reduction in restriction in the tubing/barbs.
post #146 of 167
Thread Starter 
Oh yeah i knew my .1875 end mill should have cleaned up the whole pocket just by taking a pass on each side of the slot. So i knew right away when it went over and left a huge island in the middle.



I was playing around with solidworks sim for thermal. My god is that stuff cool. It looks as tho it could be extremely accurate if you input the correct data samples. I have more or less everything set properly aside from the convective heat transfer coefficient of the water cooling itself. In doing more research, i find more college papers out there than anything and its based off of veryyy old principles and laws. Its more or less the convection power of the water moving through the channels. Its dependent on so many things that i wont be able to calculate. BUT what i can do is keep testing samples, and once i have real data from my test, should be able to duplicate the cooling power off water temp and block temp combined with how much heat energy i'm throwing into it. I'll drop a screen shot of what iv'e been playing with a bit later on.
post #147 of 167
Thread Starter 




Here's a section view of the heat transfer. This is suppose to be my hot block for the TN600. My parameters were -5C water with a convection coefficient of 1000 W/(m^2.K). I'm pretty sure the coefficient factor will be higher... but i have no way to test yet. YET haha. Also I set the heat load to the ceramic at 600W. Not sure if that's high or low, but i think its a safe number. At these parameters it shows the block maintaining the -5C without a problem. The coolness soaks into the ceramic even showing that the top surface of the ceramic should maintain pretty close to that -5C. Iv'e tested the blocks for the CTE400's with similar parameters showing roughly the same results but around room temp instead of this cascade chilled number.
post #148 of 167
Thread Starter 
Well i made a mistake on the TEC itself. I called the ceramic plate porcelain ceramic and that was a mistake. Apparently these run aluminum oxide or Alumina. Which has a thermal coefficient roughly 20 times great than porcelain ceramic. I reran the sim with those parameters and it made a huge difference. The heat soak moves far more into the copper, and the alumina plate gets the cold soak far deeper into the plate. In fact, if my numbers are correct, and i'm producing -5C chiller water coefficient close to what i'm assuming.. the ceramic itself should not rise above 2C at its hottest point. This would be ideal for the cascade setup and should allow my larger TN600 to reach maximum DT. Going to run a few more sims with similar parameters to see how well my water cooling chills the stage 1 chill blocks.

post #149 of 167
Thread Starter 
Here is the visual of my Stage 1 block. Parameters are the same 1000 W/(m^2.K) with water temp at 25c accepting a 428W heat load from each CTE400. The heat load is a completely assumed value. I used the tec calculator to see what the heat output of a CTE400 would be while moving 200 watts of heat in addition to its heat load. I say its moving 200 watts of heat based on the fact that i believe my TN600 will produce a solid 600W of heat cooling a processor with a heavy overclock. Since i'm running 3 CTE400's, i'm averaging that heat load over the three of them. That been said, the CTE at 12v moving 200 watts of heat would produce roughly 428W of heat, each. This accurately represents those values, however accurate those values may or may not be.


post #150 of 167
Nice sims.

What voltage will you be running the TN669 at? What CPU do you have, 4790k? I think your CPU will be more like 150w and if you run the TN669w at 12v it's about 220w, so ~370w total heat load. If you run the TN669 at 15v thats more like 23-25amps so 345-375w + 150w = 495-525w. I think 600w is a bit high. Going beyond 15v on the TN669 will really punish your chiller and reveal any weaknesses in thermal resistance of your direct die block just because of the heat density.

I think 14-15v is perfect for the TN669, depending on how well your CTE400's handle it. 12v might be better on the TN if the CTE's are struggling. This is a really good overall setup and you'll be able to tune the voltages of each stage to get the best out of it. Coming along very nicely!

Can't recall if you will be controlling the TEC's to a set temp or if you are just running fixed voltage?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Peltiers / TEC
Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › Peltier chilling system