direct die, partial success...

[JFettig]

it works, its running on my cpu, made a slight jet for the water to shoot on the cpu, and my temps are up a few degrees.... its real perdy and all, just the temps are up....

pictures

I am going to try another Jet design later, and move the inlet closer to the core... make all my measurements more precice..

 

[Albigger]

nice!

is that o-ring sealed i see? or you use some epoxy?

and what size is the diameter inlet in the plexi?

keep us informed

 

[JFettig]

silicone sealed. there is a .25inch jet in the center... its lexan, and the center is off center a little......... its not perfect.. I think i will do an aluminum one, with a different jet, and hopefully an Oring
were can i get a good Oring? al, were did you get yours?

 

[Diggrr]

Yeah, I was at 38C with my direct die block. I thought it was just the waterflow through the PVC cap I used. Maybe a more angled jet, like shooting in the water at 45 degrees from one edge, and the outlet on the opposite side of the core at a 45.

Just a thought. Your block sure looks sweet.

One question, do the "frosted" milled channels disappear when they're filled with water? That would be cool to see the core while it's running (something noone ever has, for long IMHO).

 

[JFettig]

yep, they are clear as day i took a pic but it didnt turn out good enough to post. I wonder if i can read the coding on the core Pepsi was able to read his, he put a magnifying glass in one of his you could read it clear as day!

 

[JML]

I don't think direct die cooling can match a convential waterblock simply because the heat has no where to spread out except into the water, and waters does not conduct heat like copper.

 

[Albigger]

that would be sweet to read the core while you are running.

so it IS on an AMD then? Stock voltage? or what's in your sig?


I got my o-ring from the tool and mold shop where i work, we keep an assortment on hand, i just found the size i needed.

but I asked one of our purchasers, he said we usually order from a place called bearings inc. but now I think they are applied industrical technology.

but they are pretty standard, i haven't found them at mcmaster.com but i bet you should be able to....


EDIT: if those lexan tops are fairly easy to manufacture (i would think they are) then try some different size holes. In my block I'm doing a tapered inlet "jet" - of course my block is not direct die. But it might be good to maybe get your hands on a 10 or 15 degree taper drill and you can start small and just keep drilling bigger, testing in between to see if it makes a diff.

what pump and what kinda flow rates do you get using that? if you know?

 

[JFettig]

it was at 1900mhz, 1.95v, rite now im at 1600mhz 1.39v and 28c and 21.7c water............ I can read the LON on my cpu, the barb is in the way of the rest..
I would be interested in getting some sort of gasket, so i can simply take the block off my cpu, instead of having the pry at it... and I do plan on making an aluminum one, and i was thinking about making interchangeable jets for it, just use a few small screws.....

 

[JFettig]

ooh, the flow is pretty good, i got a maxijet 1200, 295gph pump, im sure its around 100-120gph.. thru it...

 

[frostmeister]

Good suggestion from Diggrr up there about the 45 degree thing IMO. Am I correct in assuming that flow is even more important in direct die situations? The thing that'd put me off would be the possibility of a pump failure - the advantage of a waterblock there is that the waterblock takes some of the heat and acts more as a heatsink I would have thought. Good effort and great looking - maybe a few LED's in there too? I'm going to be buying some ultraviolet LED's for illuminating some flourescent dyed water when I get my a** in gear. With ultraviolet LED's though, you'd have to be carefull they wouldn't make the plastic go brittle after a while, and put them almost in the water otherwise plastic and glass blocks the UV too much. Keep up the good work!

 

[JFettig]

give me a way i can incorporate the 45degree thing into a 1/2inch thick block... thats all i have. unless i wanna machine stinky crackable acrylic.........................................................................................................

 

[Albigger]

i dont' think the 45 degree would offer so much improvement anyhow, the flow will be hitting the die much harder straight on, instead of just hitting and glancing off laminarly (word?)

anyhow, the multiple jet sizes is a good idea of course, i hope you can figure it out and do some testing

 

[Cathar]

You really need to get those jets much smaller. I'd be trying for an array of 3 x 3 using 1mm wide jet holes distributed evenly over the CPU die area.

Direct die cooling really relies more on peak water velocity. Your 1/4" jet is too big and the water velocity is therefore too low.

 

[Albigger]

You really need to get those jets much smaller. I'd be trying for an array of 3 x 3 using 1mm wide jet holes distributed evenly over the CPU die area.

Direct die cooling really relies more on peak water velocity. Your 1/4" jet is too big and the water velocity is therefore too low.

funny that you mention that. i was just running some calcs at 100 gph - that's only about 3.3 m/s or 10.9 ft/s

---too slow (i feel) like cathar said. I like the idea of multiple jets like Cathar suggested, and have been toying with this thought for a little while..... It has benefits, definately.

 

[JFettig]

i cant really make the small jets, but i can make one smaller jet, and should i make it closer to the core? rite now its like 3-4mm from the core, im thinking rite around 2 then make a jet like a + sign, kinda.... how small do you suggest? all i have is a 1/8inch end mill..

 

[Cathar]

i cant really make the small jets, but i can make one smaller jet, and should i make it closer to the core? rite now its like 3-4mm from the core, im thinking rite around 2 then make a jet like a + sign, kinda.... how small do you suggest? all i have is a 1/8inch end mill..

You can't go to a hardware store and buy a 1mm (or 3/64") dill bit?

 

[JFettig]

oh, i do have a 1/16inch drill bit thats broken, it would drill thru it..
or melt thru it it should? shouldnt it? how do you suggest? ill get on cad....

 

[JFettig]

heres the plan if cathar thinks its good, it seems REALLY restrictive to me....

 

[lemmyslender]

Any reason to try carburetor jets. They are relatively cheap, can be easily interchanged, and come in a large variety of sizes.

I guess it would be hard to get several together in a small space (over the core) though.

Just a thought.

 

[Albigger]

i cant really make the small jets, but i can make one smaller jet, and should i make it closer to the core? rite now its like 3-4mm from the core, im thinking rite around 2 then make a jet like a + sign, kinda.... how small do you suggest? all i have is a 1/8inch end mill..

yes....now how small is the question....


i would go closer to the core...not sure on this (not experienced with direct die) but I would think the fast water velocity gets slowed down by the more water it has to travel through to get to the core. of course you need some clearance for the water to get out of the way for the fresh water. i would try 1 - 2mm though, if possible.

i think the cross thing might work, plunge the endmill through and crank up .0625", then go back and crank down .0625", etc.... that would give you slightly smaller area than the .25" hole you have now. if you want even smaller than don't make the "arms" of the plus sign so big.


EDIT: actually the 9 1/16" holes looks good to me. That's a little more than half the area of the .25" hole. At 80 gph with those 9 holes you would be around 4.7m/s or 15.5 feet /second. combine that with slightly less clearance above the core (especially with the spacing you have with those) and i think you'll be in business