CryoStar Sli and CPU evaporators by Gosmeyer

**gosmeyer** · 05-05-2007, 03:37 PM

Could someone please give me a quick expiation on how 250/270w on load tester equates to a full tilt quad load.
Sorry for the ignorance, I know stock their value is like 135w. is there a formula or was this done by trial or both?
One more, When captube length is spoke of, is that just to get liquid up or down to ambient ? Other than that I
can’t see where there is a difference in 2m or 5m. Thanks, I know you guys will get me straightened out.

**SoddemFX** · 05-05-2007, 04:03 PM

gosmeyer,

Full load i think it's probably in that region at single stage temperatures...

No formula that we have. The FCV^2 or the adjusted voltage thingy some people use is wrong as leakage is now a big problem and leakage isn't linear with voltage or frequency. Going by comparissons of load tester at known output v's real CPU seems to be a favourite way of guestimating the power output of an overclocked CPU.

The capillary tube is used as the metering device to allow a certain mass flow of refrigerant into the evaporator. Mass flow dictates evaporating capacity (like the power sinking ability) of a system. For example with r402a evaporating at -40C with a capacity of 250W you'd need about 2.5grams per second of refrigerant entering the evaporator. The capillary tube meters this mass flow via the restriction of diameter, length with a given pressure drop (discharge - suction). Finding the "right" length of a certain diameter capillary tube is one of the things people try to do when they optimally tune a system to hold a certain load at the best possible temperature.

Hope this helps

Tom

**gosmeyer** · 05-05-2007, 04:23 PM

Originally Posted by SoddemFX

gosmeyer,

Hope this helps

Tom

Thanks, yes it does - still fuzzy on length.

**SoddemFX** · 05-05-2007, 04:37 PM

Longer lengths of capillary tube give increased resistance and therfore less mass flow.

A good analogy is like winding an electrical resistor from nickel wire. The longer the piece of wire you use the higher the resistance you get.

The physics of actual capillary tube calculation is realy, realy hard, i have no idea how to do it. All sorts of wierd phenomina come into it also, i think one of them is something called "choked flow" but there's lots. Clever people have writen whole dissertations on it. Best method for us is probably trial and error and finding what works for others and yourself

Tom

**gosmeyer** · 05-05-2007, 04:54 PM

Got it
Thanks

**runmc** · 05-05-2007, 04:57 PM

You guys are a tremendous help - thank you very much.

I hope alot of people read this and learn something valuable.

Okay Gray I took your suggestion and this is what happened.

I started adding a small amount of refrigerant and didn't get the spike you were looking for. Never got a big Spike, just a gradual rise in pressure. I actually added until we were able to handle 255watts. I thought that was enough. After running 255w for 45 min to an hour (I would guess) I lowered load to 50watts and let it sit for a few min. Then I went to zero load. I know there was liquid back to my probe because my probe temperature was colder than my evap temp

That was six inches from compressor. At two inches I may have had some superheat because I was not hearing any liquid in oil.

At 255Watts

evap temps were around -19.4C
6 inches from compressor suction line was -11.1C = 8 degree superheat

liquid line coming out of condenser was 103F, discharge pressure was 285 which equals around 107f which gave us about 4 degrees subcooling.

Temps are low, to what we would like to see. With such a heavy charge it brought the idle temps down. But in a situation like this, whose looking for idle temps.

I am happy with results from a smaller compressor. Evaporator did well also.

Now back to cap tube length. I will study my book and these awesome thoughts from you guys. Would lower temps be possible by cutting cap tube.

If the cap tube was shorter and flooded the evap more, could I reduce charge?
Please explain that for me.

That's assuming shortening the cap tube would flood the evap more. If I was able to reduce the charge wouldn't that let the pressures and temps drop? Of course I want to be able to handle the same 255watt load

**wdrzal** · 05-05-2007, 06:17 PM

Capillary tube systems need to be precisely charged, under charged evap is no good as is a flooded evap,over charged.

Runmc was that book written for phase change for CPU or just refrigeration in general????

**runmc** · 05-05-2007, 06:27 PM

Originally Posted by wdrzal

Runmc was that book written for phase change for CPU or just refrigeration in general????

You can call me Ron, if you like Walt.

The book was written for Refrigeration in General. Gary Lloyd wasn't into CPU cooling when he wrote the book.. He did quote from this book all the time he was helping us. He never said, "Don't go by this or that in the book" He always stood behind what was in his book for our application. It's kinda hard to explain. It's one of those, "You would have needed to be there" kind of things.

**wdrzal** · 05-05-2007, 06:39 PM

I read all his old posts before I joined XS, As I'm sure I read all threads in phase cooling section ,I never joined or posted cause I spent 6 years taking care of ill parents,had a lot of time to read at home and hospitals at times.

Seems many of those old archives are no longer accessible, do you know why???

You can call me Ron, if you like Walt.

Ok, Sir Ron

**Gray Mole** · 05-05-2007, 11:53 PM

One of the most important things that Gary mentioned was that you don't always have liquid if you have frost. It just means you're below zero celcius.

What you'll find with that, or a similar method of heat exchange on the suction line, is that you're helping the liquid refrigerant go to vapor even at temps that are very close to evaporation.

The accumulator and heat exchange setup creates the perfect environment for it to vaporise in a short time as it's got both heat and some extra space to do so.

You always have a mix of liquid and vapor, that's a given, but when you have a 'chamber' for the reaction to take place, the mix that's coming out of your accumulator and into the return will be a far better and more homogenous mix that's primarily vapor and very well mixed liquid.

That means that even if you have a small amount of liquid exiting, you'll likely have enough heat at the compressor to vaporise any of the leftovers.

Anyway, GREAT result Ron

-20 at that load with that compressor is a very nice result. I think it's a credit both to your evap and the build in general.

If the captube is able to give you the capacity you want, don't change it. You're probably on the edge of what it can do for you, but like I said before, if you have the longest/most restrictive captube that will still do the job, you've got the best temps that you can get from it.

You may be able to use marginally less refrigerant with a slightly shorter cap, but you'll also find that your low pressure will be slightly higher on average, bringing up your evap temps along with it.

So, you've made a system, with a small compressor, probably very quiet, that will hold the load of a Quad core cpu and then some (with those temps, probably about 230w is the most you could pull with the overclock it would allow anyway) with a nice temp.

Remember all the comments about the old OCZ unit not being strong enough? I think you're proving the point as well, that it's not so much size that matters, it's how you use it

Nice job Ron

**Unknown_road** · 05-06-2007, 03:35 AM

Originally Posted by runmc

I'll strap a probe on the discharge between condenser and filter/drier. Unknown_road - I know you remember Gary Lloyd - well last night I pulled his book out and did some refreshing of my memory -

lots of my current phase knowlegde I owe to him, at the time this forum was more of a learning place then anything else. Gary thought theory and bowman hands on.

anyway, the result with superheat within 10K and subcooling within 4K is a very good result. If you shorten the cap tube, subcooling will go down a bit but with lower load this will result into more flooding of the compressor. So I would keep it this way.

btw you did use suction pressure for superheat calculation right? and not 19.4 - 11.1...

gray, the pressure spike your talking about seems weird to me. Reason why discharge pressure is higher with high subcooling is because part of the condensor is filled with liquid, this liquid makes the condensing part of the condensor smaller. smaller condensor -> less heat removal at same deltaT -> condensor temp and pressure rises . This is a lineair process not a spike.

**Gray Mole** · 05-06-2007, 03:49 AM

I wouldn't have mentioned it, but I've seen it on 3 of the smaller units I've worked on, and been 'hunting' for the best capacity I could manage.

The larger units I haven't had issue, but they follow more closely with the captube sizing that it's easier to find.

Smaller compressors seem to need a little more captube capacity given depending how small. Captube sizing guides reflect this.

What I get when I'm stretching to the limit of tune-able capacity on a system is that both high and low pressure begins to rise. Up to that point, it's basically linear yes. At a point where it doesn't seem to want to go further, it begins to rise quite quickly as refrigerant is added, and it happens on both high and low side pressure with no real gain in capacity. It's always under load that I see it.

My suspicion is that the captube is actually at a point where it's a solid liquid head, and the liquid is backing up into the condensor. It's only a suspicion, but normally there's a mix of gas and liquid into it, even with a heavier charge.

It's something I now watch out for when I'm trying to get max length for the capacity I want, as I'm finding that the longer it is, the better my temps at whatever load I'm after, if I'm within the actual range of the captube.

Doesn't really matter if it makes sense or not, it's something I've been seeing happen while tuning (only ever on higher loads relative to the captube capacity) so I use it as a tuning tool, like everything else

Gray

**runmc** · 05-06-2007, 04:56 AM

The community thanks you guys for your valuable information and knowlege. Thanks for sharing. I will read and re-read through this thread and try to absorb as much as I can.

Originally Posted by Unknown_road

btw you did use suction pressure for superheat calculation right? and not 19.4 - 11.1...

Jurgen -

You caught me

It should have read -24C minus -11.1C which would have given me approximately 13 degreeC of superheat. What I did was use the evaporator temperature, I should have converted the suction pressure to temperature and use that. Thank you

Originally Posted by Gray Mole

So, you've made a system, with a small compressor, probably very quiet, that will hold the load of a Quad core cpu and then some (with those temps, probably about 230w is the most you could pull with the overclock it would allow anyway) with a nice temp.

Yes Gray the compressor is very very quiet. When I turn off the fan, all I can hear are my fluorescent lights buzzing and not the compressor. What fan do you guys use that is strong and quiet?? I need to change.

Thank you - Tom

- Graham

- Jurgen

and all of you

for your efforts and time

**[XC] gomeler** · 05-06-2007, 06:20 AM

Great results there for a relatively tiny compressor, goes to show that 3/4HP behemoths aren't necessary for heavy-load applications. Are you and Gosmeyer finished testing the evap or will you move on to lower-temp applications? I'd be interested in seeing how the block handles a second-stage gas like r1150 or r744.

**gosmeyer** · 05-06-2007, 06:37 AM

Originally Posted by [XC] gomeler

Are you and Gosmeyer finished testing the evap or will you move on to lower-temp applications? I'd be interested in seeing how the block handles a second-stage gas like r1150 or r744.

There are a few different thing's were looking at. Lower temps are certainly one.

**Vuco** · 05-06-2007, 08:59 AM

Great results Runmc, great evap job Gosmeyer and also great thread with info about subcooling from all the members

Know I know much more what to do with subcooling calculations, and what the numbers mean

Regards,

Sven

**runmc** · 05-07-2007, 04:15 AM

Ambient was a little cooler in my garage this morning. I bumped the load to 275w and the cooler did fine at -21c for an hour. It may hold even more load. I'm off to work so I can't play now.

Gosmeyer milled a couple more evaps to test with considerably more surface area. They are NICE!! Suction and cap tube entrance are interchangeable by user. I'm going to test them on this same setup to look for performance differences.

I give credit for enclosure design to Jinu. He sent me a prototype he made and we used it for a pattern.