Overclock.net banner

1 - 20 of 125 Posts

·
TEC Veteran
Joined
·
2,007 Posts
Discussion Starter #1
Just got my Archimedes TEC block in today and it is a thing of beauty. I will be running it with the 62mm Thermonamic TEHC1-19940, which is over-zealously and theoretically rated at something crazy like 630w Qmax @ 24v but is closer to [email protected] in the real world...which still makes it the most powerful 62mm TEC available and a whole 72w @ 12v more powerful then the former top dog CTE 400w that I am currently running.

There are already plenty of good pictures of the block itself, but the first thing I would like to mention is how great of a job the machining was on this piece - I lap all my parts (by habit, probably not much difference with a TEC) and this was by far the flattest and easiest to lap block/heatsink/cold plate I've had the pleasure of doing. It didn't even need a real multi-stage lap...I literally spent under 5 minutes and just hit it by hand with some fine grain 1500 finishing paper before polishing it with a paper towel and a drop of 3M Perfect-it stage 1 (best polish ever BTW if you have a dark vehicle, commercial grade stuff that you can't find in stores). To compare, I usually spend close to 30 minutes total with a sheet of glass and wet paper stuck to it going from 800, 1000, then 1500 before killing myself with rubbing compound and then polish ending with a sore arm
laughingsmiley.gif
.

Before (extremely flat, acceptable surface but has some machining marks):


After (spent literally FIVE MINUTES with that ~1.5"sq piece of sand paper you see and a drop of that polish...thats it!):



Will run some base tests with my current setup by Saturday night and hopefully start installing this Sunday and re-running with everything besides the block identical. Planning to do some simple hyperpi runs at 4.5ghz, 4.7ghz, and 5ghz for people with different OC'ing chips. Since my controller currently limits me to 0c cold plate which it can hold at 4.5ghz I expect the Archimedes to be the same at 4.5ghz, slightly better at 4.7ghz, and much better at 5ghz where my current block gets overloaded and starts creeping up. That should also show in nice easy to follow numbers how a TEC behaves when it starts being ran past its rating...things get nasty QUICKLY :jest:.
 

·
Banned
Joined
·
3,726 Posts
well, now you have some conception of the pain I went through sanding and polishing the 6 X CPU water blocks I have. It took me 8 weeks to do all of them working 2 to 4 hours a day. you did a great job though and the final finish looks like a mirror.
 

·
TEC Veteran
Joined
·
2,007 Posts
Discussion Starter #4
Quote:
Originally Posted by toolmaker03 View Post

well, now you have some conception of the pain I went through sanding and polishing the 6 X CPU water blocks I have. It took me 8 weeks to do all of them working 2 to 4 hours a day. you did a great job though and the final finish looks like a mirror.
Trust me my arm hurt just reading about you lapping 6x waterblocks!

These were so flat and already machined smooth raw copper so it was cake though - 5 minutes tops by hand
biggrin.gif
.
 

·
TEC Veteran
Joined
·
2,007 Posts
Discussion Starter #5
I was busy with other projects today so did not get to do much. I did assemble the block with CLP on both the hot and cold plates, actually running out of it so now I need more for the CPU...

I also am trying a new method for helping insulation...had some Plastidip left over from messing with my beater and have always wanted to try it for insulating. I masked off the hot side, the top, plugged the temp probe hole, and the cold plate raised surface then hit it with a few thick layers both along the side of the TEC+plates and all the surfaces of the cold plate except for the exposed raised area.

I will still be stuffing it with art eraser just in case, but hopefully this not only makes it better insulated but also makes it easier to clean and remove everything. For the motherboard, it is already insulated so I won't do anything with that but replace the paper towel layer on top of the eraser.




I'll remove the masking either tonight or tomorrow morning and report back whether it looks like it will work or not
smile.gif
. If so, I will be using it to insulate my next GPU when I run the two TEC chillers on it...still waiting to see what the 490x brings to the table but 2x8GB 480s for only ~$460 is crazy tempting...
 

·
TEC Veteran
Joined
·
2,007 Posts
Discussion Starter #7
Quote:
Originally Posted by foxrena View Post

The plastic dip thing is interesting.
rolleyes.gif
Did you spray into the gap between cold and hot plate?
Yep, sprayed it into the gap along the edge of the TEC as well. It came out OK, but my last coat was too thick (in a rush) so it bubbled a bit and I let it try too much before removing the masking tape so the edges arent as clean as I'd like. Still keeping it on though, will add some art eraser on top of it and hopefully will make it better insulated and easier to clean.

 

·
The 6502 Still Rocks
Joined
·
768 Posts
Are those 18awg leads on that 40amp TEC? I'll never understand that. Please tell me they're at least 14ga or 12ga and that I'm just underestimating. Do you just clip them short and run a heavier gauge to a splice?

That use of plasti-dip is really interesting. Looking good.

My block is in the mail.
smile.gif
 

·
TEC Veteran
Joined
·
2,007 Posts
Discussion Starter #9
Quote:
Originally Posted by Master Chicken View Post

Are those 18awg leads on that 40amp TEC? I'll never understand that. Please tell me they're at least 14ga or 12ga and that I'm just underestimating. Do you just clip them short and run a heavier gauge to a splice?

That use of plasti-dip is really interesting. Looking good.

My block is in the mail.
smile.gif
Yea, 18AWG leads. I would love to see larger but they aren't really necessary once you keep in mind that data sheet is basically all theoretical and bogus.

They are not ran anywhere near their theoretical Imax which is at an impossible 74Dt. At 12v w/~150w load it will be more like ~35Dt, 216Qmax, and 18a. Since I know you know your stuff with electronics, thats probably plenty of info for ya. For those without a ton of electronics experience I'll explain a bit more...

While 18a is still slightly higher then the 16a rating of 18AWG, that is a conservative rating due to many factors that effect a wires load handling ability like ambient temperature and length of wire. It depends heavily on length of the wire, temperature of the environment, and how many load carrying wires are in the bundle. For 1-3 load carrying wires together in a ~23-25c case over a very short distance it will be fine
smile.gif
. More important then the wire though, it is too much for one connector - which is why we solder them to multiple wire PCIE connectors to spread the load. While 18a on an 18AWG is fine for a short length, we wouldn't want 18a on one molex pin...it will melt the connector.
 

·
The 6502 Still Rocks
Joined
·
768 Posts
It occurred to me that the wires may terminate on the cold side of the TEC in which case some of the heat will be drawn out of them. As the current goes up, more heat is drawn out.

I'm looking to run one of these with a peak input of 850 watts ... 85% of their rated 1KW input. I don't think there's any point in running past 85% on just about any TEC as the COP curve really gets ugly beyond that point. I know the impedence can shift a bit with temperature differential but generally this TEC seems to be about 5/8th's of an Ohm. That 850 watts would put me at 23 volts and 36.9 amps at that impedence. Until I test, I'm just estimating of course. For example, in spite of its ratings, I may not want to run this TEC past 750 watts when I wring it out. You guys have indicated that this TEC's ratings are a bit aspirational and so I may have to adjust my assumptions.

Again, that 850 watts is just the occasional peak ... not a continuous condition. If I order some of these, I'm thinking of trying to get the Aluminum Nitride substrate, metalized surfaces, and maybe a slightly larger wire gauge if they offer it.

I agree though, you're perfectly fine where you're running given all of the mitigating factors you mentioned.

I'm really looking forward to your testing. I think I noticed that you are changing out both the block and TEC at the same time on this. I assume some of your gain will potentially come from being at a better spot on the COP curve with this larger TEC and perhaps not just with the advanced block alone.

I'm off to build some boards today on an unrelated item ... a pump driver. I'll keep an eye out for your progress.

 

·
TEC Veteran
Joined
·
2,007 Posts
Discussion Starter #11
Quote:
Originally Posted by Master Chicken View Post

It occurred to me that the wires may terminate on the cold side of the TEC in which case some of the heat will be drawn out of them. As the current goes up, more heat is drawn out.

I'm looking to run one of these with a peak input of 850 watts ... 85% of their rated 1KW input. I don't think there's any point in running past 85% on just about any TEC as the COP curve really gets ugly beyond that point. I know the impedence can shift a bit with temperature differential but generally this TEC seems to be about 5/8th's of an Ohm. That 850 watts would put me at 23 volts and 36.9 amps at that impedence. Until I test, I'm just estimating of course. For example, in spite of its ratings, I may not want to run this TEC past 750 watts when I wring it out. You guys have indicated that this TEC's ratings are a bit aspirational and so I may have to adjust my assumptions.

Again, that 850 watts is just the occasional peak ... not a continuous condition. If I order some of these, I'm thinking of trying to get the Aluminum Nitride substrate, metalized surfaces, and maybe a slightly larger wire gauge if they offer it.

I agree though, you're perfectly fine where you're running given all of the mitigating factors you mentioned.

I'm really looking forward to your testing. I think I noticed that you are changing out both the block and TEC at the same time on this. I assume some of your gain will potentially come from being at a better spot on the COP curve with this larger TEC and perhaps not just with the advanced block alone.

I'm off to build some boards today on an unrelated item ... a pump driver. I'll keep an eye out for your progress.

Sounds good! That driver looks way overkill for most pumps, is it for an Iwaki or a D5 Strong? PWM control on that would be pretty cool.

Foxrena tested the TEC, I've never run that model before, but states it is 15-20% better overall then the CTE400 which would make it closer to a 480w QMax then the 630-670 they rate it. I would probably hold off on the Al substrate since I would imagine losing the ability to run CLP and using MX4/Ceramique2 would cost you a bigger difference then the ceramic vs al nitride would. No math to back that up though, just a guess! CLP loves cold temps and becomes a solid metal at 8c effectively soldering the parts together...I know that on my TEC setup going from Ceramique2 under the IHS to CLP was worth a whopping 5-6c lower on my cold plate:core temp delta.

I'm sure the bigger TEC would be worth some efficiency, and am interested to see the comparison at lower clocks like 4.5ghz to see how much...but with my overkill hotside loop I don't think it will be worth that much more until I start getting to where my TEC was becoming overworked (around 4.7+ghz). At my 24/7 setting of 4.8ghz, with just my block my TEC is at its limits and cannot hold anything lower then 10c at 100% load, giving me ~40c during gaming and ~45c load temps in hyperpi (which I use because it is quick and easy to simulate a more normal day to day load vs something like IBT which is a much better stability test but gives temps your CPU will never ever get to during gaming or even benchmarking).
 

·
The 6502 Still Rocks
Joined
·
768 Posts
Quote:
Originally Posted by Puck View Post

... I would probably hold off on the Al substrate since I would imagine losing the ability to run CLP and using MX4/Ceramique2 would cost you a bigger difference then the ceramic vs al nitride would.
Aluminim Nitride (AlN) is a ceramic just like Aluminum Oxide (AlO) ... the normal white substrate. I don't think it will have a problem with CLP/CLU. Plus, I will probably have them metalize the surface ... copper over nickel I believe. I'll have to see ... maybe some with, some without.

The pump controller will do a D5 Strong or an RD30. It would probably handle 2 normal D5's in parallel as well. It was painful doing all of those 0603 placements by hand but it's done and ready to test. My eyes are killing me though.
 

·
TEC Veteran
Joined
·
2,007 Posts
Discussion Starter #13
Quote:
Originally Posted by Master Chicken View Post

Aluminim Nitride (AlN) is a ceramic just like Aluminum Oxide (AlO) ... the normal white substrate. I don't think it will have a problem with CLP/CLU. Plus, I will probably have them metalize the surface ... copper over nickel I believe. I'll have to see ... maybe some with, some without.

The pump controller will do a D5 Strong or an RD30. It would probably handle 2 normal D5's in parallel as well. It was painful doing all of those 0603 placements by hand but it's done and ready to test. My eyes are killing me though.
Very cool! That may work well since the ceramic is so smooth it doesnt break the surface tension on the CLP so you cant spread it - it just rolls around. I had to put the CLP on the copper plates themselves, which wastes some since you have to overlap to make sure it covers the whole TEC.

I tried every which way to apply it to the TEC but it just wont spread - it makes tiny little balls and leaves stains, thats it
frown.gif
. Worked fine on the copper though.
 

·
The 6502 Still Rocks
Joined
·
768 Posts
I can see the hassle in chasing little balls of CLP around a TEC surface. I'm not sure how they deposit the metalized surface on a TEC. Something similar to ENIG on a printed circuit board perhaps.

I'm not really sure what to expect from the Aluminum Nitride. It does have 9.5x the thermal conductivity of Aluminum Oxide though.

Is your rig setup to deliver over 12V to the TEC? I'm unfamiliar with Mindchill's components other than some reading from older threads. I recall you mentioned a while back that you had picked up some of his stuff on the forum.
 

·
TEC Veteran
Joined
·
2,007 Posts
Discussion Starter #15
Quote:
Originally Posted by Master Chicken View Post

I can see the hassle in chasing little balls of CLP around a TEC surface. I'm not sure how they deposit the metalized surface on a TEC. Something similar to ENIG on a printed circuit board perhaps.

I'm not really sure what to expect from the Aluminum Nitride. It does have 9.5x the thermal conductivity of Aluminum Oxide though.

Is your rig setup to deliver over 12V to the TEC? I'm unfamiliar with Mindchill's components other than some reading from older threads. I recall you mentioned a while back that you had picked up some of his stuff on the forum.
On the first build I had the TEC at 14v, but am not using that PSU anymore so they have been on 12v since then. The CTE 400w TEC used does perform better at 14-16v, it is a bit underwhelming at 12v and cant handle over 150w.
 

·
The 6502 Still Rocks
Joined
·
768 Posts
Earlier you mentioned using CLP on your blocks for TECs. Is there a reason to favor CLP over CLU in this environment? Just curious if you've found CLP to be eaiser to work with on TECs as a lot of people seem to find CLU easier to spread on an IHS. Something to do with the raw copper blocks maybe?
 

·
Banned
Joined
·
3,726 Posts
if your looking for a power supply that provides a good constant power at 14.8V, the power max power supplies I have are doing that. with decant relays and the proper wiring they stay constant at 14.8V even with a 41AMP or 52AMP draw on them, that is with both the 45AMP, and the 55AMP, power max power supplies, if you get the 75AMP power max power supply, you can easily adjust the output from 12V to 16V with a turn screw.
http://www.amazon.com/PowerMax-PM3-55-Amp-Power-Supply/dp/B004W8MQA8/ref=sr_1_13?ie=UTF8&qid=1465243473&sr=8-13&keywords=max+power
I know what is says, but I can also show a pic of the actual read if you would like. I have done some great testing with these power supplies. with the cheap 20AMP relays, 20AMP switches, and the small 18AWG wire, that I had on the test build, they held a 90% load at 13.8V, but now they hold steady at 14.8V with a 90% load on the power supply, proper wiring of 10AWG wire, and decant relays at 40AMP, with 40AMP switches, made a great difference in how well the power supply can preform.
 

·
TEC Veteran
Joined
·
2,007 Posts
Discussion Starter #18
Quote:
Originally Posted by Master Chicken View Post

Earlier you mentioned using CLP on your blocks for TECs. Is there a reason to favor CLP over CLU in this environment? Just curious if you've found CLP to be eaiser to work with on TECs as a lot of people seem to find CLU easier to spread on an IHS. Something to do with the raw copper blocks maybe?
Nope, just what I had at hand from delidding
smile.gif
. CLP performs slightly better on extremely smooth surfaces like under the IHS on CPU dies, so it is what I first purchased. CLU is supposed to be easier to apply and remove though.
Quote:
Originally Posted by toolmaker03 View Post

if your looking for a power supply that provides a good constant power at 14.8V, the power max power supplies I have are doing that. with decant relays and the proper wiring they stay constant at 14.8V even with a 41AMP or 52AMP draw on them, that is with both the 45AMP, and the 55AMP, power max power supplies, if you get the 75AMP power max power supply, you can easily adjust the output from 12V to 16V with a turn screw.
http://www.amazon.com/PowerMax-PM3-55-Amp-Power-Supply/dp/B004W8MQA8/ref=sr_1_13?ie=UTF8&qid=1465243473&sr=8-13&keywords=max+power
I know what is says, but I can also show a pic of the actual read if you would like. I have done some great testing with these power supplies. with the cheap 20AMP relays, 20AMP switches, and the small 18AWG wire, that I had on the test build, they held a 90% load at 13.8V, but now they hold steady at 14.8V with a 90% load on the power supply, proper wiring of 10AWG wire, and decant relays at 40AMP, with 40AMP switches, made a great difference in how well the power supply can preform.
I've thought about it, but my system is all inside a (big) case that already has mounts for 2x ATX PSU's so I decided to stick with them. Since my second PSU powers not only my TECs but also my pumps and fans, I don't want to run over 14v anyway.
 

·
Registered
Joined
·
1,045 Posts
The 18awg leads get slightly warm at constant 12v. Should be ok at up to 14v but keep some air around them to be safe, airflow on them even better.

If anyone can test a nitrided TEC please share the results, I wanted to but the extra cost from thermonamic was ridiculous and they couldn't really tell me how much better it was even with the ~10x higher thermal conductivity compared to AIO. nitriding cost 200% more than the TEC if only buying 1 tec. 1 fully specced TEHC19940 from thermonamic with AIN and C200 solder and lapped was USD$450+shipping. Maybe they will quote less if you can order in Cantonese/Mandarin, not sure though.

great to see some action in the TEC threads again.

It's a nice block Puck, 2 D5's or 2 DDC's is plenty of pumping power depending on how much is in your loop. I got 1.7-8gpm with 3 rads and 2 MCW80 gpu blocks in parallel and a AQ5 mosfet block and mobo mosfet block and a flow sensor.

Temps were good at 12v, should be perfect at 14v.
 

·
Banned
Joined
·
3,726 Posts
well after looking over the specs of the different types of TEC's the only real difference between standard TEC's and the nitride TEC's is the reliability of the TEC's. so, for example, if a standard TEC has a duty life span of 10,000 hours, than the nitride TEC will have a duty life span of 100,000 hours. so, the difference in price, is for the extended life span of the TEC. the same thing applies to vacuum pumps, the more expensive the vacuum pump, the longer it should be expected to run. another way of looking at, is that your paying for a guaranty, that the TEC does not fail just after 1000 hours.
 
1 - 20 of 125 Posts
Top