CryoStar Sli and CPU evaporators by Gosmeyer

[wdrzal]

The way to compare is to test against other evaps on a calorimeter. It must be balanced until it reaches a steady state. All evaps would need to be tested @ a specific temperature with a specific gas.other wise they are still just guessing

[kamongear]

The way to compare is to test against other evaps on a calorimeter. It must be balanced until it reaches a steady state. All evaps would need to be tested @ a specific temperature with a specific gas.other wise they are still just guessing

What if Identical units were made with different evaps. Wouldn't the best performing one have the best evap. I know this is probably wrong I just want to see what the fault is in it.

And what type of calorimeter would be used?

[Raja@ASUS]

True walt, this is something that has really not been done. In my case I will just have to post the results as I get them, and also how LD gets his before hand. An exhaustive real world controlled conditions test would be ideal, if there were to be some takers and of course the time and inclination to do so..

For me as long as my perception of the oc's on the gpu's are fine, I will be happy...

regards
Raja

[runmc]

I myself don't care, but I can't see why this is even an issue without results.
2 people have them now and another will in about 5/6 days.
No more go until we all see results. Good or bad.
I don't like the situation that it's in now. Sorry but I'm not going to sell something that I make just because someone wants it. If it's a performer than that's a different story.

Eric is a little harder to convince than I am. I have tried the evap on a 275+ watt load with a 5.5cc Danfoss compressor. (The specs are in this thread) I know it works. (good) Eric wants to try it on a Quad core. I guess we should have gone ahead and bought one for testing.


I wasn't insinuating any one would copy the design. .

We hadn't considered supplying them un brazed because we wanted to keep the brazing quality controlled, so that the evap would work to it's best. That's not saying gosmeyer is the only person with access to an oven.

We can always make considerations to work with everyone.

[runmc]

Eric has been hard at it - Awesome job Gosmeyer

[LittleDevil]

Really nice

Regards

[Pete]

Ron, wow! They look amazing!

[LardArse]

the fault is mine, I've got DICE cooling pots on my VGA cards ATM, as I wanna go with a speedy bench of LN2 cpu dice gpus while I have a working 3d mark 06 bench setup. once I've some nice results with dice on the gpus, then I'll put the cascade back on the vga and retune. of course, if i can get -80c fully loaded hi speed hi volts on the ultras with runny evap cascade i'll be a real happy man, as the heat load will be tremendous, we have tried ln2 with the slim solid pot and you can see the temp just rise quickly once 3d runs at 938mhz core 1.5v, so you can safely say the Runny Evap will be facing off a formidable foe.

[Fhqwhgads6680]

Just curious, but is that top plate Ridged or something? I am just wondering cause if so wouldn't refrigerant take the shortest path and go through those ridges vs. following the maze? just curious, they look beautiful tho guys, can't wait to get mine!

[runmc]

I asked gosmeyer about those marks also. They can be seen but not felt. Some sort of machining mark. Eric will have to explain.
All that will be coated with brazing alloy when they are being baked in the oven. As you can see from previous picture, alloy coats the cap and even runs slightly up the walls of the channels. Marks will harm nothing.

[gosmeyer]

Bottom of the cap is not ridged. In fact, very flat and parallel within .002". You will always have tool marks (area that the tool just traveled across) on all machined surfaces. Only the finishing process will remove them if needed (not needed in this case) as there is a .010" gap between bottom of cap and top of evap. that flows.



That's why I didn't blast the bottom of the caps, It's covered by solder

[Fhqwhgads6680]

Ok cool, those look alot deeper in the picture than realiity, I figured that wasn't the case you guys know what you are doing, but just curious. I wish I had access to all the machinery Eric does lol! I hope I can make it fly!

[gosmeyer]

Ok cool, those look alot deeper in the picture than realiity,

Yes, look at the profile toolpath. the one that goes around the diameter & then look at the x y toolpath, where they overlap you can see both. That's a real good indicator there flat when all you have to go by is a picture.

[BLHealthy4life]

What is taking so long for results??

[runmc]

What is taking so long for results??

I ran test on the evap myself and was quit pleased. We wanted more than personal results so we sent blocks to SoddemFX and Lardarse.

Soddem got called away from his home town during the weeks because of business. Lardarse has also been distracted because of other more pending problems. We are sending blocks to Little Devil tomorrow (Monday) and some to Germany on Tuesday. We will starts seeing more results soon. (I hope)

[LardArse]

alright some updates:

first, i put the top down cryostar evap onto my gpu 1 8800 ultra, so that of course there is clearance for second gpu.




Below, you see that i used tiny bit of foam tape to cover the oyter perimeter since the evap base is protruded slightly from the foam insulation.




so i put on the ultra:





as u can see above, foam tape to further wrap seal.

Also as previously talked about that tiny space is easy to conquer with bit or cut up armaflex





you see that i can just hang up the evaped card with a hook and cable tie


Now, due to my phobia when testing with gpu 2 cascade, i directly started off gpu1 cascade with cryostar evap with short big cap tube - ~ 1.95m of 0.04". Now i have HX temps of -35 to -40c load.

End result was that flow was indeed fast and -72c idle. uunder load at 918mhz core of the ultra and 1.5v, temps will normally hang ~ -66C but GT3 of 3d mark 06 will bring it down to -63C or so. as a comparision, dry ice will only let me do 900mhz core, unless i pour in some ln2 at GT3 and GT4 to drop temps.
cryostar at current stage for me does one step higher.
blindly pouring ln2 without a temp gauge, card will do 972mhz core.

now my gpu2 cascade is a headache. it couldnt hold the load which i at first thought could be solved with shorter cap tube. that was why i did a short cap tube on gpu1. however, something was definitely amiss, and i even changed out the compressor. suction pressure is high, even when not much refrigerant used, while HX temps is good. so right now im guessing the oil seperator float that could be faulty and bypassing discharge into suction.

so i guess i can definitely do better temps with more throttled refrigerant flow. im hoping 3.5m will give me idle -85c load -80c.

Just a thought, maybe the cryostar evap base can be much thicker, at least a version of it. as many will bench with these evaps, the thick base will hold out some lower temps for short bursts of intensive bench moments - AKA cpu test for quad core. or even some gpu intensive periods can be pulled through as the tests are not too long and you will have time to recover.

[Moc]

and some to Germany on Tuesday

Who is going to get them in Germany?

[Fr3ak]

Those evaps look sexy

[runmc]

Who is going to get them in Germany?

toaster

[Pete]

What is taking so long for results??

Not everyone is focused to clocking there PC every day, some of us have a day job, familys and a life.

I do agree the time is passing by rapidly!

LD will get it done in a day to 3 days on new build unit

[runmc]

LD will get it done in a day to 3 days on new build unit

Little Devil should be getting raju evaps any day now.

[kayl]

toaster

and australia ?

[n00b 0f l337]

Shamino, your problem is you should be using 7.5 meters of .04"

[LardArse]

Shamino, your problem is you should be using 7.5 meters of .04"

i really want to, but skeptical u know, at this stage it seems the flow is just barely fast enough. i really have no confidence in lengthening cap tube anymore, do remember its 1/2hp on the low stage. Has anyone done some really hot loads with similar setups, like a hi clocked and hi volted ultra?

[n00b 0f l337]

Increasing the cap tube length will speed up the flow as pressures will be higher on high side, and lower on low side. Either check your superheat or a mass flow chart. Your so unrestrictive you might have just tuned the unit to air condition your home