News Phononic crowd-funds Peltier-based Hex heatsink

What's the power usage though - I always thought that Peltier cooling in a PC required a separate PSU to deliver the oomph that the Peltier requires?

 

"Banana for scale"

I did lol

 

I thought we got over Peltiers and TECs for CPU cooling years ago...

I'm guessing they completely missed the days of trying to cool OC'd P4s with air or water cooler peltiers and having to have a completely separate 100W+ power supply spitting out hot air of it's own just to deliver the power needed for the heat pump then.


EDIT:-

Even if it somehow gets around the tons of extra power problem you then run headlong into another. They suggest they're only using an 80mm fan. Which is fine, the heat pump allows you run the CPU cool while kicking up the heat exchanger temperature which in turn gives you a larger thermal gradient and better heat transfer to the air passing over it. (i.e. if you allow the exchanger to get hotter it can dissipate more energy) but what about the now much, much hotter exhaust air.

What happens when you're filling you're PC case up with CPU exhaust air at 40C, 50C or even 60C. I'll go out on a limb and say you're probably going to make other components unhappy (see:- adjacent air cooled graphics cards in multi-GPU rigs)

In short, yes, but why? We tried it and water by itself was less hassle and did the job and air coolers are pretty good too these days and don't exhaust air that is stupidly hot into a thermally sensitive environment.

 

only supports 2 sockets.... no good for me then

 

Depends on the pelter power. If they're packing ~100W TEC, at 12V that's only 8.3 AMPs. Old PSUs didn't have much 12V power. They can't pack a higher rated TEC as the small cooler can't take the heat and they can't drop below ambient otherwise it'll condense water.

 

Even a perfect 100% theoretically efficient Peltier requires X watts input power to move X watts heat between its faces. For a 85W TDP CPU, that's an 85W Peltier (for a total of 170W that the actual heatsink needs to dissipate). But Peltiers are not 100% efficient, they require typically 1.6x to 2.2x the power they can move.. so for a 85W TDP CPU, that's a 136W Peltier (total heatsink load 221W).

And all cooled by a single 80mm fan running at a screeching 4500RPM.
No thank you.

 

Like others have said, we've seen attempts at peltier coolers in the past. And they sucked, or at least offered no advantage over a decent air cooler.

I don't mean to sound pessimistic as it's far too easy to belittle a fledging company with big ideas, but if the major players haven't come up with a good peltier based solution, I don't see how a small company can.

 

Dear Gareth,
First, let me thank you on behalf of Phononic for including the press release information and discussion on your website. As a young company, Phononic appreciates the interest and open discussion on our products. To that end, I'd like to answer a few questions about Phononic and the CPU cooler. I'm a technical leader at Phoonic and we are making and shipping our refrigeration products now through our distributer. As far as the use of Indiegogo, we wanted to gauge the interest and spur discussion, but let me assure you, all those that pre-purchase a CPU cooler will get one. I'll add some comments soon on the technical questions.

 

I agree with you that this isn't a great idea, but your statement that "a perfect 100% theoretically efficient Peltier requires X watts input power to move X watts heat between its faces" is simply inaccurate. In fact, you can run the same device as either a thermoelectric cooler (by applying current, heat moves from cold side to hot side) or as a thermoelectric generator (heat moves from hot side to cold side and current is produced).

A "perfect 100% theoretically efficient Peltier" would be a perfect Carnot cycle machine, and would have an efficiency (ratio of work done on the cooler [in the form of electrical energy input] to heat dumped at the hot side) of [1-Tc/Th], where Tc is the absolute temperature (in K) of the cold side and Th is the absolute temperature of the hot side. Equivalently, heat is absorbed from the cold side (which is what you want to measure) with efficiency of [Th/Tc-1]. "Efficiency" is probably a misleading word here as smaller is better - because this is a reverse Carnot cycle (a heat pump rather than a heat engine), the less work you do to pump a given amount of heat, the better.

Putting real numbers to it, if your cold side is at say 300 K (27C) and your hot side is at 350K (77C) and you need to move 100W, then the perfect thermoelectric cooler would need 100 * (350/300 - 1) = 16.7W of power input to maintain equilibrium, i.e. it pulls 1 W for every 6 W of electrical power it shunts from its cold side (which is on the CPU) to its hot side (which is on the heatsink).

However, a real world TEC is more likely to reach 10-15% of maximum Carnot efficiency, which puts it more like 100-150W to cool a 100W CPU under reasonable operating conditions.

 

Apologise if I'm speaking out of turn, but I found this supplier for one of the medical refrigeration products.

 

Yes, HCL (Heal Care Logistics) is our distributer for the clinical medical refrigerator. Thank you, Corky42, for speaking up.

As for the HEX 1.0, we thought indiegogo would be a good site to generate publicity and pre-orders, but we are going to production on this and other CPU/GPU related products and we will be selling direct and/or through distributers. We are funded through our investor group and will begin production regardless of the number of sales on indiegogo. Of course, we hope for a good response and many pre-orders through the site.

 

I remember the cooler master V10 was a similar concept, combining an air cooler and tec. In the end it didn't perform as well as the top air coolers of that time (I think the TRUE was the best back then).

I'm sceptically looking forward to how this will perform.

 

Yep, so with a ~65W chip you could use a 100W TEC. This cooler isn't made for OCing, it's SFF where AIO or large air coolers won't do.

http://www.coolermaster.com/cooling/cpu-air-cooler/v10/

Good memory. And yes, it was pretty pants, but AFAIrecall the TEC was very low power.

 

Just a quick question / point:



In the blurd is says:

Does this use up a PCIe slot with a controller board?

If so is this PCIe 1, 4, 8 or 16 and how long is the data cable to the heat sink (depends where an available PCIe slot is in the computer)?

Tnx.


Look forward to more real world comparisons against water cooling systems. A-I-O and custom.