Direct Die

[Kabooka]

My direct die machine is up and running, some stats so far.
Keep in mind it was done with take off parts - (proof of concept)

System -

AMD Thunderbird 1.4
ECS K7AMA (ewwww)
On board NIC
128 meg PC100 no name memory
WD 3.2 gig HD
4 spin cd rom
300 Watt Skyhawk PS
Windows XP professional

First boot up successful - no shorts YEA!!

Idle temps - 42
Folding temps - 48

I know that this is high compared to other WC rigs. But it is lower than what I was running on air with this set up. I'm also looking for other benchmarks for a TBird 1.4 on water.

I shut the system down to put together a better socket mount - I hacked together a zip tie arrangement I really didn't like and definitely didn't trust, and I still have to get some water wetter and such.

for the next step I don't know if I should:

a. Buy another MOBO so I can try an OC
b. Give this a go on say a 2100 or maybe a Barton

Any Thoughts???

[Player0]

Thats 48c stock 1.4 temps? Thats really high for a watercooling setup.

My Maze4 runs 38c full load with a 2400+ at 2250mhz 1.85vCore.

What pump and radiator are you running? Any pics?

[Kabooka]

I know that there have been better temps, but I haven't seen temps related to a Thunderbird 1.4 - this thing is notorius for running hot - that's one of the reasons I thought It would be a good test. I'd really like to hear about others experiences with a 1.4 on conventional WC

Heres some pics of the pump and setup the radiator is a chevette heater core, and I'm running with all 5/8" tube.

[Player0]

Yes, the 1400 is hot at 73 Watts stock. However, any AMD processor over 2100+ runs hotter than that. I have watercooled 1400s before, and my temperatures were lower than that. The only way to compare temps is to install a waterblock on that system yourself.

You will eventually see what everyone else who has experimented with this has, and realize that it just doesn't work. I have seen Leufkens blocks and Dickis blocks and a couple other guys and their experiments. Standard waterblocks just work better because of the larger surface area, which is more important than the thermal interface losses with a normal waterblock.

Some thoughts on your design. First off, your sending water from the pump to the radiator. This slows down your water. I suggest changing so that the pump goes directly to the block. You NEED fast flow for a system like this to work. I think your pump is too under powered to use 5/8" ID tubing. And I wonder how effectively you are 'jetting' the water over the CPU die. So I think there is a lot of room for improvement with your design.

[Kabooka]

Thanks for the input Player,

Actually I don't jet water on the die - I'm drawing it from the top of it, I figured that waterblasting the die didn't yield any benefit against drawing a large volume of water across it, but jetting it may have some improvements and it's easy enough to set up.

I do disagree on the 5/8 tube and general flow assumption, the larger diameter tube is actually easier on the pump and allows for high flow. I think there might be some confusion between pressure and flow. A smaller hose diameter would increase resistance which would increase pressure thereby reducing flow. The best example of that is a spray nozzle at the end of a garden hose.

Reversing the direction of the water in a closed loop system will have no beneficial effect on the flow.. it is what it is no matter what direction it is traveling.

[Player0]

Jetting the water against the die provides the best results. The higher the water velocity the better, as proven in several Leufkens tests.

Your pump is probably rated to use 3/8" ID tubing. You could probably use 1/2" ID tubing with success. Using larger diameter tubing reduces head, ie the pressure the pump can provide to the water. Using 5/8" diameter tubing means there is more 'weight' for the tube to pump, and there for it is running at a higher 'head' and will NOT flow as much water. The speed of the water will also be reduced. And high water velocity is exactly what you need for jetting water against the CPU die.

"the larger diameter tube is actually easier on the pump and allows for high flow"

This is simply false. If you have 6 pounds of water in the tubes, rather than 4 pounds of water, how can this possibly be 'easier' on the pump? It is a FACT that larger tubesize reduces a pump's head efficiency, simply due to the mass of all the extra water in the tubing that needs to be moved. Some pumps are designed to be most efficient with 3/8" tubing, some with 5/8" tubing. You are using your pump in a mode which is probably making it less efficient, and you are getting less flow because of it.

"A smaller hose diameter would increase resistance which would increase pressure thereby reducing flow. The best example of that is a spray nozzle at the end of a garden hose."

Not necessarily. A smaller hose diameter DOES have increased flow resistance, but this doesn't overweigh the head loss that a larger hose would introduce. They are all factors, but you are confusing which factors are more important to your system. Water will accelerate itself through a smaller opening. Which means the flow may actually end up being the same depending on the head capability of the water source. Yes, this is getting complicated, I know.

Buy a 5/8" garden hose and a 3/4" garden hose of the same length. Have them fill a 5 gallon bucket. Both will fill the bucket at exactly the same speed, ie same flow. But the 5/8" hose will shoot the water out at a faster velocity because it is reducing the 3/4" outdoor house pipe to 5/8" diameter, squeezing the water down and accelerating it, like putting your thumb over the end of a hose. There is a cut-off point where hose friction will overcome the accelleration potential of the water...like putting your thumb on the end of the hose too much. You get a small jet stream of water, but you end up loosing all flow. You have to cut the diameter by like 75% for friction to overtake the acceleration however, and its not something you have to worry about in your system, except on your waterblock where I suggest something like a 1/4" barb to supply the water jetting. This is going to SERIOUSLY reduce flow, which is why I mention you might need a pump capable of 400-600GPH to make this work well.

[dicki]

you saw my article right? i had much better temps (though with a much cooler chip running at a heavier overclock...) i know the 1.4's are hot but try jetting water to see what happens.

finally... i seriously think ramsinks... what have you got to loose!

dicki

[Kabooka]

Originally posted by dicki
you saw my article right? i had much better temps (though with a much cooler chip running at a heavier overclock...) i know the 1.4's are hot but try jetting water to see what happens.

finally... i seriously think ramsinks... what have you got to loose!

dicki

No I hadn't seen your article, drop a link... I've already started the reroute for jetting, shouldn't be too difficult - one fitting and it should be good to go!!

I want to benchmark with the jetting - then I can experiment with the ramsink... the block I built lends itself to trying different things like that...

[WesM63]

Hey m8,
Dicki's article has been around for a long time, HERE'S (http://www.liquidninjas.com/reviews.php?op=showcontent&id=3) the link, enjoy! :D

[Kabooka]

Player,

I'm enjoying the mental exercise it's been long overdue for me!!

Jetting the water makes a lot of sense.

"This is simply false. If you have 6 pounds of water in the tubes, rather than 4 pounds of water, how can this possibly be 'easier' on the pump? It is a FACT that larger tube size reduces a pump's head efficiency, simply due to the mass of all the extra water in the tubing that needs to be moved."

I did some quick calculations and came up with a difference of 10 cubic inches of volume between 5/8" and 1/2" tube (about a 24" length) which equates to about 6 ounces of water. A potent amount if we're talking Jagermeister but somewhat inconsequential in my application.

"Buy a 5/8" garden hose and a 3/4" garden hose of the same length. Have them fill a 5 gallon bucket."

What you’re doing here is taking the pressure capacity of the source and matching it against the smallest optimum hose diameter. The bucket would fill up the same with a 2" hose. I agree that the difference is the velocity of the water exiting the hose, but necking the 3/4" hose down to 5/8 at the end would produce the same velocity.

"I suggest something like a 1/4" barb to supply the water jetting. This is going to SERIOUSLY reduce flow, which is why I mention you might need a pump capable of 400-600GPH to make this work well. "

I agree - I have to get to a smaller orifice above the die, but I'll be looking for a happy medium. Don't know about the pump though... I'd be going for something capable of more pressure to maintain a rated GPH rather than a higher GPH which will be defeated by back pressure....

I found a neat program to calculate HEAD LOSS (http://www.pipeflow.co.uk/theory.html) and used it for the attached image, which proves that your assumption is correct in that the smaller diameter increases the velocity of the water, but at the expense of head loss. A larger diameter hose clearly has less head loss, but far less velocity. So my challenge is to lay out the system and hit that magical balance of GPH and velocity at the die orifice, but for the main plumbing I will stick with the larger tubing.


Again thanks for the debate!!

[Kabooka]

Wes - Thanks

Dicki

Nice read, that's encouraging about drying things out and giving it a go again!!!

[dicki]

cheers wes ;)

yeah i think i came to some different conclusions after a while... for instance i now don't think it leaked around the edge of the die i think the glue broke down (common problem) but other than that its accurate.

dicki

[WesM63]

No prob guys. ;)

[Player0]

"This is simply false. If you have 6 pounds of water in the tubes, rather than 4 pounds of water, how can this possibly be 'easier' on the pump? It is a FACT that larger tube size reduces a pump's head efficiency, simply due to the mass of all the extra water in the tubing that needs to be moved."

I did some quick calculations and came up with a difference of 10 cubic inches of volume between 5/8" and 1/2" tube (about a 24" length) which equates to about 6 ounces of water. A potent amount if we're talking Jagermeister but somewhat inconsequential in my application.


Yep, actually the difference is 7.24 ounces with only 2ft of tubing. With 4ft of tubing, the difference is nearly a pound of water. I wasn't quantifying the numbers, my line of thought was simply: more water = more mass = more work for pump = pressure loss = less head. Since that is only 2ft, and half a pound extra of water, it may not be that significant. I was then thinking that given 4-6ft of tubing, the extra weight would be much more burden on the pump.

And for some reason, I am totally wrong. It just doesn't make sense. How can moving more mass take less effort?? But you are right, I work in an office with civil engineers and I checked out the fluid mechanics book, and I'll be damned. Greater diameter = less head loss. Not even counting friction there is less head.

Apparently, the actually acceleration of the water in the tube is where all that pressure energy goes. These flow rate equations look like a derivative of Newtons law. Ive been thinking about this stuff the wrong way, I need to sit down and regroup ;)

[Kabooka]

Well now that we've solved that for the benefit of mankind (and watercoolers everywhere)

Waddaya think - " Tastes Great" or "Less Filling"


hehehe :drinky: :drinky:

[Farabomb]

I think a shot a jager is in order. :D

I love the stuff but if you buy a bottle, better have some bail money too. ;)

[Kabooka]

Originally posted by Farabomb
I think a shot a jager is in order. :D

I love the stuff but if you buy a bottle, better have some bail money too. ;)

I had a room mate that kept a bottle in the freezer at all times, the problem was he was a terrible drunk... the key was getting that fifth shot into him so he would pass out..