Trending

EKWB Could be Releasing A Electrothermal CPU Waterblock Soon

By

Could Peltier cooling be making a comeback?

CPU waterblock
(Image credit: Shutterstock)

As spotted by hardwareluxx.de, EKWB could be building a prototype CPU waterblock that uses an electrothermal transducer, assisted with water, to cool a CPU. LinusTechTips gave a sneak peek at this "prototype EKWB" CPU cooler in one of its latest videos. There was no explanation of exactly what it was, but they were able to hit 0 degrees Celcius on a 10900K at idle and 50C under load without the need for LN2, suggesting some sort of electrothermal technology was being used.

Thermoelectric cooling relies on the Peltier effect, which allows electricity to transfer the heat (energy) from one source to another. The discovery was made in 1834 by a physicist named Jean Charles Athanase Peltier. If used effectively, this could be applied to cool CPUs or anything that requires cooling in general. The interesting thing about this technology is the fact it uses current, not air for cooling, meaning sub-ambient temperatures are totally possible (as shown by LinusTechTips with the 0C idle temps).

However, as harewareluxx points out, many manufacturers have already tried this.  The problem with this technology stems from a large amount of waste heat the cooler builds up -- as well as condensation. So it'll be interesting to see whether EKWB can solve these problems if they plan to launch this prototype CPU water block to the masses.

Topics
Cooling
12 Comments Comment from the forums
  • escksu 10 November 2020 01:08
    Peltier again??
    Reply
  • Lucky_SLS 10 November 2020 08:23
    The success depends on the support and reliability to this early adopter tech. Watercooling being an exotic hobby, ppl won't mind adopting it if the price n performance are worth it.

    Hope there aren't any extra parts that can make an itx custom loop difficult...
    Reply
  • w_barath 10 November 2020 09:33
    This is a fundamentally flawed concept. Peltier devices obtain at best 8% efficiency at cooling. That means in order to cool 100W of waste heat from the CPU you will need 1200W of input power.

    Your 80% efficient PSU will need 1500W from the wall to achieve that - and that's without powering the mobo, cpu, gpu, peripherals, fans, lights... In short, this is entirely undoable at load. Yes it will work at idle and at low load levels but the wall power and the PSU upgrade costs will be ridiculous.

    If your'e already buying a water-block... just use a water-block as intended If you want your water-block to give you sub-ambient cooling, put its radiator in the way of the output of your aircon.

    Second flaw: The one and only reason to use a Peltier device is to try to get below ambient. This must not (meaningfully) occur or it will cause condensation. So you will need to apply active heating to evaporate the condensation. If 8% of that heating enters the Peltier device you have just completely defeated its gains
    Reply
  • escksu 10 November 2020 11:06
    Lucky_SLS said:
    The success depends on the support and reliability to this early adopter tech. Watercooling being an exotic hobby, ppl won't mind adopting it if the price n performance are worth it.

    Hope there aren't any extra parts that can make an itx custom loop difficult...

    Peltier is been around for decades..... I have personally played with peltier a long time ago during Pentium 4 days. Its just too inefficient and too much waste heat. Today's CPU makes it even harder to use peltier. Eg. the 10900K consumes 330W of power when overclocked. So, you need an even bigger peltier (eg. 400W) to cool it. Together thats 730W....lol....

    Another is the big temperature difference between idle and load. So, you could achieve sub-0 temps when CPU is idle but rises to 40-50C under load. This is due to the way peltier works. If you want lower temp, you need even bigger peltier.....Imagine over 1000W power dissipation. Madness.....
    Reply
  • spongiemaster 10 November 2020 14:52
    The cooler in the Linus video had one PCIE power connector. It wasn't specified which one, but even an 8 pin, and the spec is "only" up to 150W. The video didn't address condensation, and they didn't have any issue with it while running their test system. So why don't we wait to see what EKWB came up with before wildly speculating about 400W+ coolers that sweat all over the rest of your system?
    Reply
  • Integr8d 10 November 2020 17:11
    Admin said:
    EKWB Could be Releasing A Electrothermal CPU Waterblock Soon. LinusTechTips shows a prototype block from EKWB working on a 10900K.

    EKWB Could be Releasing A Electrothermal CPU Waterblock Soon : Read more

    How badly did the author of this title fail basic English? Seriously?
    Reply
  • hotaru.hino 10 November 2020 17:28
    This is kinda reminding me of Dell's H2C system, which was a Peltier + AIO water cooling hybrid. The scant reviews I could find on the machines that had it said they kept a higher end Core 2 Quad exceptionally cool. But no word on power consumption.

    Old reviews are kind of painful to look at these days.
    Reply
  • TJ Hooker 10 November 2020 17:43
    w_barath said:
    This is a fundamentally flawed concept. Peltier devices obtain at best 8% efficiency at cooling. That means in order to cool 100W of waste heat from the CPU you will need 1200W of input power.
    I agree that peltier coolers don't make sense for PC cooling, but you're being a bit too pessimistic. Having a peltier cooler with a coefficient of performance (COP) of 0.08 as you describe would be an especially poor device. COP being (useful heat transfer)/(input power). You can definitely have a peltier cooler with a COP of 1 or higher (and COP for a given device will differ with operating conditions). E.g. this peltier CPU cooler that draws 20-40W while cooling 100s of W worth of CPU load. But even with a COP of say, 5, you still have to provide and dissipate 20% more power compared to what the CPU itself is drawing, which is obviously isn't ideal.
    Reply
  • TJ Hooker 10 November 2020 17:49
    Integr8d said:
    How badly did the author of this title fail basic English? Seriously?
    Are you referring to the use of "a" instead of "an"?
    Reply
  • escksu 11 November 2020 01:35
    spongiemaster said:
    The cooler in the Linus video had one PCIE power connector. It wasn't specified which one, but even an 8 pin, and the spec is "only" up to 150W. The video didn't address condensation, and they didn't have any issue with it while running their test system. So why don't we wait to see what EKWB came up with before wildly speculating about 400W+ coolers that sweat all over the rest of your system?

    The reason for 400W+ cooler is very simple. According to review by PCgamer, their 10900K consumes peak of 331W.

    Assuming you are using a 331W rated peltier, what happens is that the hot at cold side of the peltier will have 0 temp difference. So, its as good as not using peltier. Then, 331W from peltier and 331W from CPU, you get 662W of heat......

    In order for peltier to work, you need one that is much higher than the peak power consumption of your CPU. If its lower, your CPU will overheat instead.

    You can prevent condensation by controlling the power to the peltier. Long ago, it was done using a thermal probe, but since CPUS all have internal temp sensor, that can be used instead (likely through USB).

    One more thing to take note, if your CPU achieves the same clockspeed with or without peltier, then having peltier makes no sense right? So, you want to go higher using a peltier, this means even higher power consumption... so even bigger peltier needed.
    Reply