Liquid chiller blocks..

Ive seen some that have liquid chillers here..

and Ive read about them here and elsewere..

Something has me scraching my head..

water blocks are designed for ambent temperatures.. and Ive seen that most liquid chiller users are using waterblocks..

If your liquid temperture is say -30C would it not stand to reason that your system could hold load temperatues better if the blocks were to be designed with more mass?

like the diffence in base thickness between water blocks and direct die phase change evaps.

again just a thought.

I agree with that, more mass means there will be less of a temperature swing in unloaded/loaded situations. I think drastic temperature swings are similar to voltage spikes, they are not very good for stability & overclocking.

I won't pretend to be anywhere near an expert on this, but I don't think you can quite compare a direct-die phase change unit to a water chiller. The way I see it, dry ice/LN2 tubes and phase change evaps benefit dramatically from thick bases since there's no static flow of fluid through them (talking about the strict definition of fluid here, liquid or gas). There's liquid or solid that changes to gas so you're looking at less heat capacity, I think.

A water chiller on the other hand has a constant stream of water. None of the dynamics should really change between it and ambient water. There's no change of phase happening at the waterblock, the difference is simply that the liquid is much colder than in an ambient loop. Between the chilled liquid and the CPU/GPU, there's just a conventional passive removal of heat, so you should get the same delta as with ambient. A good waterchiller should have nearly no variation of liquid temperature, and ambient water loops do as well.

That's the way I see it, hopefully a guru comes and sets me straight, heh.

Gautam.. I would think that you would have practical experance with this..
Im currious how well your cooling system holds load temps. Is your coolent temperature and CPU/GPU load/non-load temps. approximately the same? Or do you have a way of testing that set up?
Thanks in advace.

Should've mentioned this earlier. Under load, coolant temperature increases by a degree and a half or so at most. GPU temperature is about 15C higher than coolant temperature according to the GPU's probe loaded, 10C unloaded. The CPU temp I'm sure is misreported. I haven't had ambient water for a couple of years now, but from what I can remember it was about the same case.

Since the heat exchange happens at the waterblock, the temperature should only depend on the coolant temperature. The fact that it's cold won't change anything. It should be a perfect linear relation between coolant temp and CPU/GPU temp, unless the coolant undergoes dramatic temperature changes, which could be possibly from a poorly designed chiller.

I would think that if the blocks had more mass that the coolant would be able to keep them cold under load better thats all.
Your GPU temp. going up 15C is what I was thinking might happen if the blocks did not have enough mass.
of course there may be another reason for that that Im not thinking of.
Im am just thinking here I had no idea that your GPU wouldnt hold the same load temp. as your coolant, I was thinking something like what you discribed might fight that effect.
thanks for the info.

The reason why extra mass in a waterblock is not going to help anything is because it is rather insignificant compaired to the power output of the processor. Copper has a specific heat of 380 J/(Kg K) and a density of 8.9 g/cm3 meaning that a block 1cm x 4cm x 6cm has a heat capacity of 81 J/K. If such a waterblock were cooling a 100w cpu and the water temperature increased suddenly, then the waterblock temperature would initially increase at a rate of 1.2C per second until it levels off at the new temperature. Now if you double the mass of the block, you still have a very high heating rate.

LabRat23.. your input is appresiated but Im not quite sure I follow..
Im assuming most liquidchillers are converted aircons with 1/2 hp rotary compressors.. I would expect that the compressor is more then enough to keep the temperature down.. so why would the load temps increase much?

in other words the liquid temperature should stay very low and not change more then 1 degree C from what Ive seen..

I guess what Im saying is you lost me..

The point is that a thick copper mass will not slow the temperature much. You need to throw continuous cooling at it or the processor will quickly overheat. Futhermore, the thermal changes on a processor die never spikes. When you switch from a load to no load condition the temperature will lower relatively slowly and approch the operation temperature and vica versa. Temperature doesn't affect stability unless it becomes excessively hot or cold for a die to operate.

Voltage is a different story. Processors have a narrow operation range for voltage. Any spikes or surges above the absolute maximum operation values will result in damage which may or may not be terminal. Going below the minimum will cause errors of all sorts. The introduction of noise can also have drastic operational effects even within the normal voltage range. Noise is basically AC waveforms interposed into a DC current. This should be filtered by an L or Pi passive filter within the 3 phase buck switching power supply to the CPU. A bad cap might allow some dirty DC current to pass. Switching from no load to full load isn't a problem because the switching power supply automatically compensates for voltage changes. The capacitors resist the change in voltage while active cirucits continuously adjust the power going into the filter.

The one fact that isn't understood very well is how sensitive CPU's are to noise. They don't use regular bipolar transistors as some are lead to believe. Enhancement mode MOSFETs are used. I was playing around with large discrete eMOSFET with a light bulb attached. I was able to turn the light on by grounding one hand then using a finger on my other hand to touch the gate on the MOSFET. A little bit of noise can go a long way in throwing off the processor.

your saying clean power is more inportant.. I originaly thought that that was a given when OCing..
and colder doesnt matter?
thats sounding odd.. maybe I misunderstood.

ok.. whats the diffence in cold plates for pelts and having a aircon power your loop?

Green, I missunderstood what you were asking. The mass of a block means nothing. What matters is the pathway of the heat to the watter. Too thin and the sides don't do any work. Too thick and the center doesnt do as much. Its a tradoff. Waterblocks tend to be thinner because the heat doesnt have to travel far to get to the water. In a DD block, it has to go a bit farther, hence more thickness.

LabRat23.. no problem.. I was just wondering why there isnt a design diffence for liquid chilling blocks that would make them diffent then the ones designed for water cooling.. I would personaly thing that SOMETHING would need to be diffent to operate at much lower temperatures. can you think of anything?

The only thing that comes to mind would be the O-Ring. Everything else really doesn't change due to the lower temperature.

In most simplest terms, chilled water is just like taking a regular water loop to a very cold place.

LabRat23 said:

The mass of a block means nothing.

Click to expand...

I disagree with this. More mass in any block is going to maintain it's temperature in a narrower range regardless of whats cooling it, air, H2O, Chiller, Phase, whatever. Temperature swings on die are not a big deal when running under air cooling, but when you get into the ultra high clock ranges holding a specific temperature means everything for stability.

Look at Cathar's blocks, the best H2O block's in existance. These are not lightweight blocks, they take advantage of mass as well as internal design.

El<(')>Maxi said:

I disagree with this. More mass in any block is going to maintain it's temperature in a narrower range regardless of whats cooling it

Click to expand...

Thats not right at all. Any block will have a temperature under load and a temperature at idle. The amount of mass in the block does not change these temperatures. The only thing mass will do is change the time it takes a block to reach its load temp when the cpu goes from idle to full load, which has nothing to do with cpu stability.

Mass doesnt magically reduce temps, and temps are the only thing that matters.

Cathar's blocks have a lot of mass in them because they are built solid to resist warping and because he needs thick material to cut the cups into.

Mass is very useful to help keep stable temperatures when using liquid nitrogen. The more mass you have, the easier it is to maintain a given temperature.

When using a chiller with active flow then the liquid becomes part of the thermal ballast. Water and ethylene glycol have very high specific heats.

Copper 380 J/(kg K)
Water 4186 J/(kg K)
Ethylene Glycol 3378 J/(kg K)

The only real factor is getting the best waterblock setup that allows for the easiest heat transfer from the CPU to the liquid. Thicker metal is actually more heat resistive through it's thickness. Sometimes that loss is traded off for somewhat better heat spread.

J/(kg K) are being quoted.. but not for the coolants that could make this sound more logical..

LN2 ? J/(kg K) ?
refergerant at phase change DD ? J/(kg K) ?

the explantations are starting to sound much clearer guys.. so far so good..

J/(kg K) in words: The amount of energy it takes per kilogram of a given material to raise one degree Kelvin.

J are Joules (1 watt is 1 joules per second, 100 watt processor puts out 100 J per second, 10 seconds results with 1000 J of heat)
K is Kelvin which is Celcius +273 degrees. (water freezes at 273 K and boils at 373 K)

I quoted Specific heats, this is what it takes to raise a solid, liquid, gas or fluid's temperature. Liquid nitrogen and other phase changes use Latent Heat. LH is the amount of energy to boil or melt 1 kg of the material. The temperature must be at the material's boil or melt point. It will stay at that temp. untill all of material changes it's phase.

Latent heats of vaporization (Temperature where it happens at)

LN2: 199,000 J/kg (-196 C)
r12: 166,000 J/kg (-30 C)
r22: 199,000 J/kg (-41 C)
r134a: 193,000 J/kg (-27 C)
r290 [propane]: 365,000 J/kg (-44.5 C)
Helium: 21,000 J/kg (-269 C)
Hydrogen: 445,000 J/kg (-253 C) [Very Dangerous!]