What do you want to see in a Water Block???

[jaydee]

I dropped in my XP1600+ set to 1.95Vcore, 10.5X Multi, 145FSB, 2.7Vmem, for 1523mhz.

Room temp is 22C, water temp is 25C, and load temp is 28C!!! My hose is 5/16" ID, and my connectors are 1/4"ID, pump is 185GPH in-line and my rad is pictured below. So far so good.




Note the mobo is on my test bench, upside down from a normal case install.

[agentwebhunter]

i was wondering if you thought of the reynolds numbers for this water block you want to design. for the many weeks that i have been frequently coming to this site, i have never seen anybody mention anything about reynolds numbers. this is a number that is characteristic of the flow of water through a pipe or an obstruction. to get a higher reynolds number, you must have higher flow and mor turbulance. turbulance helps by almost circulating the water while its moving through the pipe. this helps because it lets the water carry more heat since it creates more water surface area.

all that i;m stating is that you may want to look into this. i know nothing on this subject and parts of the stuff i said up their may be crap. the only reason i mentioned this is to see if anybody knows what this is, and if they know a lot about it.

[jaydee]

I have a design engineer that specializes in heat exchangers as it is what he does for a living. He has thrown in some input and done some calculations on the subject and thinks the design i just came up with should do pretty good. I have a 1/2" wide 3/8" deep channel that lets the water flow very well and with the semi maze design it has planty of serface area. if I can hold 4C above water temp with 1/4" barbs and a 185GPH pump and an Aluminum block I am doing something right. The flow is great. It comes out the water block nearly as fast as it goes in.

I have already got the kinks worked out and the code written up for the CNC mill for the final version. I will make some up after new years and send them off for review. The Aluminum block should perform aswell if not better than Danger Den's Maze 2 and only cost no more than $25 US. the Copper version will be around $30 or so. We will see what the reviewers say in about a month.

[morphling1]

I don't wanna diss your temp results but those are so good that I can hardly believe it. Your cpu pums out 84W of heat at full load and at water temp of 25°C, thermal resistance of this WB is 0.035 K/W!!! if your cpu temp is 28°C. How do you measure temp, if with probe in socket, are you sure it's in contact with cpu. If everything checked, then I must only congraculate for one hell of a WB you made.

[jaydee]

Originally posted by morphling1
I don't wanna diss your temp results but those are so good that I can hardly believe it. Your cpu pums out 84W of heat at full load and at water temp of 25°C, thermal resistance of this WB is 0.035 K/W!!! if your cpu temp is 28°C. How do you measure temp, if with probe in socket, are you sure it's in contact with cpu. If everything checked, then I must only congraculate for one hell of a WB you made.

Thats why I want to send them out to independant reviewers. In case my testing methods are not right.

I have so much flow going through this thing is dosn't have much loss in speed comming out of the block as going in.

I am going to run a hotter stress test on the block today using Toast instead of Folding@Home and that may raise the temps. Even if my temps are off a bit it will still be worth the $20-$25 price compared to Danger Dens Maze 2. I have alot more flow in mine. he only uses 3/8" channels and roughly 1/4" deep. Thats a restriction on its own. Mine has 1/2" wide by 3/8" deep channels. And it has a couple smooth corners to tranfer heat/cold to and from the block.

I will have a pic if the inside later today on my website. After new Years I will send out a few to some independant reviewers.

[jaydee]

This is a prototype. The final will have a slight change as the bottom channel will extend a little more and the channles will be centered in the block. And the starting stopping points will be rounded out a little more to allow a 1/4" NPT threaded barb which is a little over 1/2" wide at the threads. Any size of fittin that uses a 1/4" NPT threaded barb will work. Which should go up to atleast 3/8" hole. That is more than enough to make this block work well. I am only using a 1/4" hole in my barbs for my testing.

Deminsions will be 2"x3"x5/8" overall. Channels is 1/2" wide and 3/8" deep. I left 1/8" of material for the heat to spread on the bottom which seems to work out just right with the design.

[morphling1]

Good, I can't wait to see the inside, your put pic on this forum too. When I tested I saw temp rise of 4°C with Toast compare to Seti.

[morphling1]

Oh, I was offline and didn't saw your post. Simple and efficient, with very little flow restriction.

[jaydee]

Originally posted by morphling1
Good, I can't wait to see the inside, your put pic on this forum too. When I tested I saw temp rise of 4°C with Toast compare to Seti.

I am expecting about a 2-3C rise over Folding. Folding@Home runs hotter than SETI by about 2C on my comps so I used Folding as more of a general use testing. Toast should get the temps up more, but I didn't use it to do my main testing with because no one on the planet runs their comps that hard even with extream gaming so it isn't a good real world test IMO.

I will have some addition results with toast up later this evening.

[jaydee]

2 hours of toast raised the temps up 2C. it hoped up to 2C in 10mins and has stayed there ever since. I just milled a new block this morning which will be the sellable one and I am going to put a 3/8" in/out lets on that one instead if the 1/4" I am using now to see if that will help any. I am very happy with the current temps though.

Also I bumped it up to 2.05Vcore and 1553mhz at 148FSB before I started toast. Tomorrow I will test it at default for the XP1600+.

[Butcher]

You might also try cpuburn as this gives me consistently higher temps than toast.

Also are these temps taken with the on-chip diode or with a socket thermistor? Socket thermistors are known unreliable especially with water cooled setups.

[jaydee]

Originally posted by Butcher
You might also try cpuburn as this gives me consistently higher temps than toast.

Also are these temps taken with the on-chip diode or with a socket thermistor? Socket thermistors are known unreliable especially with water cooled setups.

Already tried CPU Burn it did about the same as Folding.

Temps are taken with seperate external probes and the onboard probes. Both are within a few degrees F of each other.

[Butcher]

I'm assuming you used the latest cpuburn... the earlier vers didn't do so well on athlons for heat up. On the other hand I have a tbird not an XP so that might affect things. ON my friends p3 cpuburn was about 8C hotter than toast whereas on my tbird it's a degree or two.

[jaydee]

I am not sure actually, it says v1.0. I found it looking for Toast and tried it out. Is there a newer version and if so where can I find it?

[Butcher]

ok just double checked:
toast 46C stable
cpuburn4 48C STOP error
as for where to get it I always have to search for ages to find it, no known homepage for it, attached my copy here.

[futRtrubL]

You do realise that your going to always get the same flow out as in right? Restriction has nothing to do with that. If they were different water would be disappearing in the block.

Great temps, can't wait to see Cu results.

Edward

[jaydee]

Originally posted by Butcher
ok just double checked:
toast 46C stable
cpuburn4 48C STOP error
as for where to get it I always have to search for ages to find it, no known homepage for it, attached my copy here.

Thanks, I got it running now. I will run it for a while and post back.

[jaydee]

Originally posted by futRtrubL
You do realise that your going to always get the same flow out as in right? Restriction has nothing to do with that. If they were different water would be disappearing in the block.

Great temps, can't wait to see Cu results.

Edward

Ok, volocity then. i am expecting about a 1C difference in the Cu version.

[futRtrubL]

With the same cross sectional area you'll get same velocity from the same flow.

Edward

[jaydee]

Originally posted by futRtrubL
With the same cross sectional area you'll get same velocity from the same flow.

Edward

I was always under the impression corners cause things to slow down. Being pumps loose pressure to resistance, a water block with a lot of corners and resistance should slow the flow down because of the limitations of the pump? I havn't seen a pump that can keep the same volocity for an indefinate distance. Most small pumps loose a lot of their flow rate in a foot, so if you have a block with 5 inches of channels it has to slow down inside the block or pumps would not loose flow at all. I have less amount of channel distance than a Maze 2 so the water isn't in there as long and dosn't get slowed down because of the limitations of the pump. But the chnnels are wider alowing for a lot of serface area like the Maze 2. My design is more effecient at keeping the pump from loosing flow rate in other words.

The reason I said that the water is going in alsmost as fast as it is going out is because I have only a 185GPH pump which looses a lot of flow with the more distance it has to push the water. If you had a 500GPH pump it may not be as much of an issue. It may not be anyway.

Thats the best argumant I can come up with, even if it isn't right at least I am thinking about it.

And the CPU Burn hasn't raised temps to much yet. 23C ambient, 27C water, 33C CPU. So far.

[Butcher]

Originally posted by jaydee116

Most small pumps loose a lot of their flow rate in a foot, so if you have a block with 5 inches of channels it has to slow down inside the block or pumps would not loose flow at all.

That's in 1 foot of head surely, i.e. 1 foot vertical. It'd be a really poor pump that lost much pressure pressure pumping down a 1 foot horizontal straight tube.

And the CPU Burn hasn't raised temps to much yet. 23C ambient, 27C water, 33C CPU. So far.

What was the toast temp by comparison?
Still 33C gives a C/W of 0.07 which is nice

[jaydee]

Originally posted by Butcher

What was the toast temp by comparison?
Still 33C gives a C/W of 0.07 which is nice

Toast was pretty much the same. 22C ambient, 26C water, and 32C CPU.

.07? Wow. My goal was to keep it 10C or less of the water temp under load. So far so good. I hope the reviewers have good luck with it. The price will be pretty good so even if it is a couple C over the maze 2 it will cost almost half as much for the aluminum version. I am going to order the Copper and Aluminum stock tomorrow so probably will ship the test units to the reviewers The following Monday. By then I will have the selling price aswell.

i should be able to keep up a stock of them as it dosn't take very long for that mill to mill the block. After soldering/welding the tops on and putting together the mounting hardware I should be able to make far more than I can sell. I am not expecting a lot of sales. i have no idea what the market demand is for water blocks. i don't think it is very high though.

I have another project in the works which will be a chiller. It will hold 3 50mm pelts. It will have a maze design milled into it. I had a few people ask to make some so I will make a few in the future. i already have to code made up for the mill I just need to come up with a way to hold the stock down as my vise isn't wide enough. Shouldn't be to hard though.

[Fixittt]

Pretty impressive!

I love to see people getting into making blocks,

I can tell you this, if you get a block that people like, then it can me HELL!

That little machine is not capable of putting out the volume that I need to put out. So I machine my blocks at work. Am still working on them.

Good Work. Keep it up.

[Brad]

hehe fixittt (wait is that how it is normally spelt) knows all about that.

I know that now you have heaps of demand fix, but do you think you will eventually expand the product line, to a GPU cooler, or another socket a cooler?

[jaydee]

Originally posted by Brad
hehe fixittt (wait is that how it is normally spelt) knows all about that.

I know that now you have heaps of demand fix, but do you think you will eventually expand the product line, to a GPU cooler, or another socket a cooler?

Humm... I can make one in 30mins from start to finish. it only takes 12minuts for the machine to mill the block. In a regular 10hour work day that is 20blocks. I really doubt i can sell 20 a week let a lone 20 a day.

Yeah GPU cooler is on my mind, as is a north bridge cooler, and a few different socket A blocks.

The mills I have only costs $1,500, so it will be easy to buy more if the demand is there which I doubt it will be.