EKWB preparing TEC equipped cpu block

A

alxlwson

You Know Where I Live
Joined
Aug 25, 2013
Messages
9,414
So, kind of a news item. Not much information to go on at this point. We do know it's able to keep a cherry-picked 10900K 5.4GHz all core @54°C.

Linus has a little bit of the scoop.

 
Interesting. It's back to the future.

I'd use this, but only if I could get one in my GPU and CPU blocks, and it came with control circuitry to keep things ambient or above. Because otherwise the TEC will raise the loop temp and thus raising the temps of all the non TEC equipped blocks.
 
Zarathustra[H] said:
Interesting. It's back to the future.

I'd use this, but only if I could get one in my GPU and CPU blocks, and it came with control circuitry to keep things ambient or above. Because otherwise the TEC will raise the loop temp and thus raising the temps of all the non TEC equipped blocks.
Click to expand...

It does appear that the unit is controlled with windows software based on the USB thing he was talking about
 
Zarathustra[H] said:
Interesting. It's back to the future.

I'd use this, but only if I could get one in my GPU and CPU blocks, and it came with control circuitry to keep things ambient or above. Because otherwise the TEC will raise the loop temp and thus raising the temps of all the non TEC equipped blocks.
Click to expand...

More importantly not cause condensation...
 
Dayaks said:
More importantly not cause condensation...
Click to expand...

Yes, I did not write that clearly. I want both. I don't feel like dealing with rubbing grease in sockets and that kind of bullshit.

alxlwson said:
It does appear that the unit is controlled with windows software based on the USB thing he was talking about
Click to expand...

Potential bummer but it depends on the implementation. I'd only use it if the controller has its own processing capability and can be programmed.

if it relies on a real time client running in the OS in order to monitor and control things, it is a total fail.

Ideally the unitl would have an RH and temperature sensor embedded in it, which would calculate the dewpoint in real time, and automatically shut itself off to maintain a temperature within a safety margin above the dewpoint.
 
I didnt watch the whole video because he is annoying but wouldn't it make more sense to put the TEC on the radiator? more surface area to heat transfer?
 
sc5mu93 said:
I didnt watch the whole video because he is annoying but wouldn't it make more sense to put the TEC on the radiator? more surface area to heat transfer?
Click to expand...

Maybe. You'd have to figure out how to assemble it though. Usually radiators are essentially a copper tube loop attached to fins. I'd need to put a plate in there, sandwich a TEC inside it, and hvae the finscome off the other side, and still have enough space for air to blow through.

You might be able to make it more effective from a heat expulsion perspective, but it would likely be way more expensive, use a lot more power, and be a much more complex design.

I think I'lss stick with the single flat TEC between a water block and a chip. Much simpler, cheaper and very effective.
 
The issue is tecs are not very efficient to simply put one on your cpu and cool the hot side you need a 200+ watt tec to keep it from cycling below and above ambient (creating a little puddle on your mobo). The coolers commonly accomplish this by making a traditional air cooler with the tec resting above the heatpipes on the CPU. They work decent but suck down a bunch of extra power compared to a air cooler.
 
Last edited:
cdabc123 said:
The issue is tecs are not very efficient to simply put one on your cup and cool the hot side you need a 200+ watt tec to keep it from cycling below and above ambient (creating a little puddle on your mobo). The coolers commonly accomplish this by making a traditional air cooler with the tec resting above the heatpipes on the CPU. They work decent but suck down a bunch if extra power compared to a air cooler.
Click to expand...
The way to solve violent oscillation is to have a solid chunk of metal to even things out. That and to use a specialized controller. Maybe something that has more than just "on/off" modes, but can apply voltages in between for partial effect.

Totally doable if a company with some engineering behind them gets in on it. It would be more controlled than in the old days.
 
Zarathustra[H] said:
The way to solve violent oscillation is to have a solid chunk of metal to even things out. That and to use a specialized controller. Maybe something that has more than just "on/off" modes, but can apply voltages in between for partial effect.

Totally doable if a company with some engineering behind them gets in on it. It would be more controlled than in the old days.
Click to expand...

keeping the chip close to ambient is easy a on/off conroller works fine for that. you can even sense enviroment rh and temp if you want to get as cold as you can. Once you are talking aboout trying to stay bellow that point it is a problem simply to find a tec strong enough to deal with cpus under full load. You could use a large thermally conductive slab of metal and place a extremly oversized tec on it and probrably get by with that. You can also cool fluid with a array of them efficiently.
 
Zarathustra[H] said:
Interesting. It's back to the future.

I'd use this, but only if I could get one in my GPU and CPU blocks, and it came with control circuitry to keep things ambient or above. Because otherwise the TEC will raise the loop temp and thus raising the temps of all the non TEC equipped blocks.
Click to expand...

Umm im not sure exactly what youre saying. A TEC is not going to raise the temp of everything in the loop because you have radiators constanly fighting for equilibrium with ambient already. Whats different than a standard cpu block? Or are you saying there needs to be a way to keep the TEC thrtottled so its not constanly going sub ambient? That is kind of a waste of energy and excessive heat I agree.
 
tangoseal said:
Umm im not sure exactly what youre saying. A TEC is not going to raise the temp of everything in the loop because you have radiators constanly fighting for equilibrium with ambient already. Whats different than a standard cpu block? Or are you saying there needs to be a way to keep the TEC thrtottled so its not constanly going sub ambient? That is kind of a waste of energy and excessive heat I agree.
Click to expand...

It will result in ~100w extra into the loop. That will probably raise the temp of the loop slightly
 
tangoseal said:
Umm im not sure exactly what youre saying. A TEC is not going to raise the temp of everything in the loop because you have radiators constanly fighting for equilibrium with ambient already. Whats different than a standard cpu block? Or are you saying there needs to be a way to keep the TEC thrtottled so its not constanly going sub ambient? That is kind of a waste of energy and excessive heat I agree.
Click to expand...

I'm saying both.

1.) It would need to be controlled to prevent it from going, not subambient that is fine, but prevent it from going sub dewpoint (with a safety margin)

2.) TEC's use lots of power in order to force a cold side and a hot side. They dump the resulting heat on the hot side. So the hot side is not just the heat from the CPU, but also the heat generated by the TEC, which is significant.

So yes, all else being equal, if you put a TEC in your loop, your loop will run hotter. This is not necessarily a problem, as the radiators will be more efficient at removing heat with a higher delta T to ambient, but the problem arises when you have two blocks, let's say one for CPU and one for GPU. If the CPU has a TEC and the GPU doesn't, the GPU will see higher coolant temps.

Now, you can counteract this by running the fans faster, but then you are creating more noise, which we are trying to avoid.

For a system like this I'd insist on having a TEC on each of the blocks.
 
Zarathustra[H] said:
I'm saying both.

1.) It would need to be controlled to prevent it from going, not subambient that is fine, but prevent it from going sub dewpoint (with a safety margin)

2.) TEC's use lots of power in order to force a cold side and a hot side. They dump the resulting heat on the hot side. So the hot side is not just the heat from the CPU, but also the heat generated by the TEC, which is significant.

So yes, all else being equal, if you put a TEC in your loop, your loop will run hotter. This is not necessarily a problem, as the radiators will be more efficient at removing heat with a higher delta T to ambient, but the problem arises when you have two blocks, let's say one for CPU and one for GPU. If the CPU has a TEC and the GPU doesn't, the GPU will see higher coolant temps.

Now, you can counteract this by running the fans faster, but then you are creating more noise, which we are trying to avoid.

For a system like this I'd insist on having a TEC on each of the blocks.
Click to expand...

At that point why not just chill the coolant as much as you can and pump chilled coolant everywhere? And at that point a phase change cooling system would be more efficient.
 
cdabc123 said:
At that point why not just chill the coolant as much as you can and pump chilled coolant everywhere? And at that point a phase change cooling system would be more efficient.
Click to expand...

As long as you keep the coolant above the dew point this is a viable approach. it's more technically challenging though.

A Peltier element is a flat thing Much easier to sandwich between a block and a chip than to integrate into a radiator.
 
If this thing is practical I’d love to add it to my loop to cool my 9900k.
 
Zarathustra[H] said:
As long as you keep the coolant above the dew point this is a viable approach. it's more technically challenging though.

A Peltier element is a flat thing Much easier to sandwich between a block and a chip than to integrate into a radiator.
Click to expand...

Keep it well below the dew point frosty lines arnt the issue as much as cycling below and above that point. And it would be very difficault to find a tec with enough power to handle the tdp of a gpu
 
I'll be interested to see how this plays out at large. I started to get into cooled coolant loops but condensation was always a problem, and there just wasn't software to keep up with temperature management.
 
I don't know why people keep harping on the "puddle of water" issue


Step 1) Assemble rig.

assemble it.png



Step 2) Kick it.

kick it.png



Step 3)
Enjoy dry components.
 
Last edited:
  • Like
Reactions: Axman
like this
I'll just whip out my Swiftech MC2000 peltier block circa 1999 from my Celeron 300A... good retro/throwback [H] times!

Yup, still have it. :D

Proof:

IMG_0397.JPG


Noisy little bastard... but boy did it do some amazing stuff with Celery overclocks...

If EKWB is doing anything serious/substantial as to thermals with a peltier, be prepared for that sweet, sweet jet engine sound - especially if they aren't adding a massively sized radiator/fan combo to effectively cool it. Peltiers add a ton of extra waste heat that has to go somewhere.

I can see it working quite well if designed properly using power regulation coupled with good modern software/sensors though. In the old days, if you didn't keep enough of a load on the CPU, the block would start frosting up on the cold plate side like a cheap motel beer fridge cranked to 11.
 
Last edited:
cdabc123 said:
And at that point a phase change cooling system would be more efficient.
Click to expand...
Oh for sure but loud, big, hard to maintain, and obnoxious.

With peltier cooling all you have to worry about is making sure the main machine is on a different circuit than your peltier's PSU...
 
Yeah I think I will wait to see if this actually works like they say. Never seen a Peltier setup ever work like it should, without drawing a massive amount of power as well.
 
Brackle said:


Better info than the Linus video.
Click to expand...

Was going to post der8aur video since it has more information about the product (including software control of the TEC). That video shows you the cryo mode (looks like it keeps the TEC outside of condensation range) and the full out to -5c (accept the warning that you may short something out). Also, I would trust der8auer a fuck ton more on extreme cooling than Linus, since his other video showing the use of a water chiller... It was umm... Yeah when water ends up in the socket since they cannot prevent condensation...

For those talking about chilling the water instead of the CPU directly, look at these setups.

https://www.overclockers.com/forums...ed-Water-Cooling-Obsession-Is-Now-Version-4-0 (I think this setup uses 300W, cools CPU only)

https://www.overclock.net/threads/chilled-water-cooling-vs-3-0-build-log.1633988/ (Same guy earlier build, not sure what is up with OC.net fourm migrations, looks like it finally took down pictures.)

Than there is this monstrosity... https://www.overclock.net/threads/full-system-tec-cooling.1651377/



Now for this Intel only cooling solution... By the looks of it one should be able to take off the Intel water block and use an AM4 style one? Since it looks like it really is only cooling the TEC hot side, I do like the setup/software for this.
 
Would love this kind of thing for Ryzen. Would buy it in an instant. Hopefully they respond to it and this kind of cooling becomes a real option.

Single threaded performance means everything to me so its quite enticing.
 
You must log in or register to reply here.
Back
Top