Could this be the future in PC cooling and silencing ?
http://www.eneco.com/technology.html

The chip actually consumes heat to produce electricity or when a power is applied will act as a heat-pump !!
Imagine a double layer CPU : normal CPU next to this : a 0W TDP !!!

(Found via : http://akosh.pcinpact.com/actu/news/329 ... esseur.htm)

Unless they make it out of superconductors there will always be resistance, and therefore the components will also produce some amount of heat. But components made with superconductors would produce no heat anyway.

but superconductors would need to be cooled by liquid nitrogen, thus defeating the point of this technology. what I want to know is what they do w/ the electricity that has been generated?

According to the Second Law of Thermodynamics, the maximum efficiency for a heat engine such as this one is (1-Tc/Th) where Th and Tc are the hot and cold Temps in Kelvins. Thus, sandwiching a Carnot Engine (nothing has better efficiency than a Carnot Engine) between a 100W CPU at 75°C and a Ninja at 25°C results in an efficiency of 14% - 14W of electricty, 86W to be removed by the Ninja. And a Ninja would have to be LOUD to keep its base at 25°C while removing 86W.

You're not going sneak a miracle past the laws of physics. Heat engines simply require large temperature gradients in order to achieve meaningful efficiency.

Wikipedia: http://en.wikipedia.org/wiki/Heat_Engine

Perhaps what i should have said is if someone comes up with a cunductor that works as a superconductor in room temp, it migh twork wel for SPCR folks. Highly unlikely though.

Well, according to the second law of thermodynamics, a heatpump is a physical impossibility, you get 400% efficiency out of a device where Th is colder than Tc. Of course they still obey the laws, but the mechanism upon which they work means that treating them as carnot engines as a whole is a clearly false assumption.

Assuming the chip doesn't work because thermodynamic laws operating on the whole system in one simple calculation is looking for trouble, there's more than one way to skin a cat and even more ways to get the laws of thermodynamics to work in your favour.

There is a confusion here between "efficiency" and "coefficient of performance", which is the metric used to rate a heat pump's output.

http://en.wikipedia.org/wiki/Heat_pump



A heat pump moves heat by mechanical means, so inevitably the conversion of electrical energy into useful work is not 100% efficient.

Anyway, I don't think these guys are claiming 100%+ "efficiency", cos that would be a perpetual motion machine. I think the tech they might be using is this:

http://en.wikipedia.org/wiki/Thermotunnel_cooling

My classes in electronics and thermodynamics date back more than 10 years but I think we are getting mislead here : for me this is more similar to photoelectric components : using some source of energy and turnung back some of it into electivity though of course with no perfect efficiency.
Here it is a part of the energy agitating the atoms (aka HEAT) that is converted in electricity I guess.
Apparently there is no energy applied to the device : just withdrawn.

Having said that :
- heat pump do exist (you can even buy one and to install in your house and it happens to be one of the most eco friendly heating systems)
and it is used to convey heat from a cold source to a hot source : here it would be the other way round anyway.
- the above mentionned formula I believe applies to a specific type of thermodynamic machine and with no connection whatsover with this kind of device,
- supraconductivity has little connection to this either : it is about conducting electicity with no resistance. Here we are talking about a device that will consume thermal energy and produce some electricity nonwithstanding the fact that it can have its own resistance (like any electricity source).

I believe it might work, and here's why: I investigated the heat to electricity conversion myself, and set my eyes on the thermionic conversion. But this usually works at very high temperatures, much higher than silicon based semiconductors would bear. It could be used with SiC though, at 1000+ degrees.
The technology presented by ENECO seems to use the same principle: moving - by their own kinetic energy - charges from an emitter to a collector, not in 'void' but in a thermoelectric semiconductor.
Remains to be seen how effective is in real life. Maybe there's room for improvement, after all this is the first generation.
Another technology that I read about some time ago involves the magnetocaloric effect, a new cheap compound was recently discovered.

Interesting, I never heard about this effect before:

http://en.wikipedia.org/wiki/Magnetocaloric_effect

At present, they have to use superconducting magnets, so it's not very practical for PC's, but maybe if they use a rare-earth magnet?

This is an article from the Institute of Physics News (I can't find it on their site anymore, but had it saved). The technology is implemented by Camfridge Ltd.

Researchers at the University of Cambridge have discovered a material that gives a whole new complexion to the term 'fridge magnet'. When this alloy is placed in a magnetic field, it gets colder. Karl Sandeman and his co-workers think that their material - a blend of cobalt, manganese, silicon and germanium - could help to usher in a new type of refrigerator that is up to 40 percent more energy-efficient than conventional models. Given how much energy is consumed by domestic and industrial refrigeration, that could have a significant environmental payoff. Sandeman will describe the work at the Institute of Physics Condensed Matter and Materials Physics conference at the University of Exeter, on Friday 21 April.

The 'magnetic fridge' envisaged by the Cambridge team would use a phenomenon called the magnetocaloric effect (MCE), whereby a magnetic field causes certain materials to get warmer (a positive MCE) or cooler (a negative MCE). Although the effect was discovered more than 120 years ago, it is only recently that magnetocaloric materials have been known with the right properties for use in everyday refrigeration. But several factors have so far prevented such applications.

For one thing, some of the materials - typically metal alloys - that show the strongest MCE contain the element gadolinium, which is very expensive. And some of the best potential alternatives contain arsenic, raising health concerns.

Sandeman and colleagues have now found a material that is neither toxic nor costly, and which generates significant cooling at around room temperature. The key to the magnetocaloric behaviour is a sudden change in the magnetic state of the compound - a so-called magnetic transition. The material is magnetic because it contains metal atoms that themselves act like tiny bar magnets. As it is warmed up from subzero temperatures, there comes a point where these atomic magnets abruptly change the way in which they are lined up. This switch occurs at different temperatures when the material is placed in a magnetic field. So applying such a field can trigger the magnetic transition, and the resulting realignment of atomic magnets can then cause the material to lose heat and become colder - in other words, it shows a negative MCE.

Such a material could act as a heat pump for refrigeration. Applying the magnetic field triggers cooling; then the field is switched off and the material absorbs heat from its surroundings, cooling them down. Once that has happened, the field is switched on again and the cycle repeats, each time sucking more heat from the surroundings.

Sandeman and colleagues say that their new magnetocaloric material is particularly attractive because it can be tuned - depending on the strength of the applied magnetic field, as well as the way the substance is synthesized - to work over a wide temperature range, making it potentially suitable not just for a kitchen fridge working at room temperature but for other cooling applications at higher or lower temperatures. The Cambridge team are now developing a spin-off company, Camfridge Ltd, to bring their new materials system to real applications.

The presentation 'A clean, cheap and green magnetic refrigerant: negative magnetocaloric effect in CoMnSi1-xGex' by Karl Sandeman will be delivered at 14:45 on Friday 21 April 2006 at the Institute of Physics conference Condensed Matter and Material Physics.

Many powerplant types work by generating heat, boiling water with the heat, and making the steam turn turbines that spin generators.

In theory it is definitely possible to use the heat generated by the cpu to produce electricity.

But this process can not be even near 100% efficient. Power companies would be really interested in applying such technique for their coal plants or whatever to reduce fuel expences.

100% efficiency would mean that we could take all the heat produced and change it back to electricity, so the computer would power itself, thus creating a perpetual-motion machine, and these are impossible to implement even in theory.

So yes, some of the heat could be transformed, but no, it can't be very effective.

Basically it all boils down to this: You can move heat from colder to warmer substance by spending extra energy, or you can let heat move from warmer to colder substance by itself.

What I really see as a solution to the TDP problem is that as the technology advances, smaller and smaller machines will be enough for people. Even now the average joe can do basically all of his tasks with a laptop that takes just couple dozen watts.

A machine with efficiency>100% violates the first law of thermodynamics: energy must be conserved.
A heat engine with efficiency > (1-Tcold/Thot) violates the the second law of thermodynamics: entropy can be created but not destroyed. It's a firm physical law, and any machine in violation is a perpetual motion machine.

Therefore, low-temperature waste heat recovery can not be very efficient. However, high-temperature waste heat recovery can be efficient, economically attractive, and green.

I recall having read a paper about using thermocouples to generate electricity from the heat in a car's exhaust. Maybe these chips could boost the efficiency of that system to a level where it's economically attractive.

Peltier's can be used to generate electricity as well and have been around for ages, but I've yet to seen one used for that purpose (I don't think they're internally reversible anyway).

I don't see the difference between a peltier and this "Thermal chip."

Waste heat recovery sounds rather attractive but if you look at a more reasonable temperature difference you're looking at a maximum efficiency of ~12%, say you can achieve 8% in actuality.

Factor in the energy you'd need to make the thermal chip, as well waste products, and suddenly it does not look to great from a green computing standpoint.

The device is very simple. It converts a DIFFERENCE in temperature into electricity. That means that the larger the temperature differential the higher the voltage. Electrical voltage is nothing more than a differntial itself. It is a measure of potential energy just like lifting an item off the ground increases it's potential energy with the height lifted of the difference between where it is and where it wants to come to rest. The chip wants the heat on both sides to be the same. At that poingt it would have 0 potential. As the temperature differential increases heat flows across the resistance of the chip from the hot side to the cold side. This flow across a resistance produces electricity. We produce heat from electricity in just the opposite way by flowing electricity across an electrical resistance.
The other use is called the Peltier effect. By applying a voltage to the chip, heat is transferred from one side to the other. The one side can become quite cold. Cold enought that it can produce destructive condensation in an electronics environment. It would be a good way to get rid of a fan on your cpu or atleast turn it down. By using one of these chips, you can transfer you cpu's heat into the heatsink keeping you cpu quite cool. You heatsink can become quite hot, but the hotter it is compared to the case temperature, the better it will transfer heat.

But Pelts are quite inefficient (someone posted one recently that used 60W!) and it is questionable whether they provide better cooling overall than a good heatpiped heatsink (eg Scythe Ninja) with good TIM (AS5).

Given a peltier of adequate power(?), the component to be cooled can maintain temperatures below ambient. Exhaust heat may be an issue, but it will cool better (at least in certain parts of the system).

People rarely use peltiers on CPU's these days because you'd need a rather large one to be able to over clock (100+W), but they are still used on GPU's.

http://www.hi-z.com/websit30.htm#KansasCity

There are a lot of papers on the US Department of Energy web sites on this topic too. It's some pretty old research; a more efficient TEC module would certainly be an advantage.