waterblocks get better with more flow rate
radiators slowly get worse with more flow rate

billa's radiator article showed this

Sorry to take this thread on a tangent. I should re-read billa's article...I don't argue with the data collected, and for some radiators low flow increases heat loss. It's the explanation that low flow is better because the water spends more time in the radiator that I can't understand yet.

Thanks

More time in rad means that the time the water has to cool off is longer... Just realized that it's kinda hard to explain :(

No ASD it's just about it, you explained it well :)

i dunno how to explain this but the total time the water stays in the radiator is the same no MATTER what rate the water is flowing. Think of it this way, if a water molecule takes ten seconds to circle a complete cooling loop and during each loop, it stays in the radiator for two seconds, it would then be in the radiator for 12 seconds per minute.

60 secs / 10 secs/loop = 6 loops per minute.
6 loops/minute * 2 secs (in radiator) / loop = 12 seconds in radiator per minute.

Let's speed up the flow to complete a cooling loop every five seconds, and that same water molecule only spends 1 second in the radiator per each loop. If you calculate the total time it stays in the radiator, guess what? It's still 12 seconds!

60 secs / 5 secs/loop = 12 loops per minute.
12 loops/minute * 1 sec (in radiator) / loop = 12 seconds in radiator per minute.

This formula also applies to residual time of H2O in the waterblock in terms of heat transferred to the coolant. Even if the flow is as little as .10 gpm, the residual time within the radiator would still be the same. So, as you can see, residual time of a liquid coolant within the radiator (or any other heat exchanger) is a constant as long as there is some sort of continuous flow of the coolant in a looped system.

Thats wrong. As ondaedg stated, time in the rad is constant (duh! Have you ever looked at your rad while the system was running and realized there was no water in it? Of course not!)

The peak probably occurs because you're decreasing the delta T between the coolant and the rad. In low flow situations parts of the loop have "cold" water and others "hot". In high flow the entire system was "warm" water. The lower delta T of course decreases the ability of the rad to move heat.

I guess ASD tried to say, the more time water spends in the radi, the more it gets colder... :D

If the 'hot water' from the 'waterblock' has more time to 'cool off in the radiator', you will water more close to ambient going to the 'waterblock'.

It doesn't take a genius to figure that out.

Ok guys, here's an update...
Swiftech officially informed us (their wholesalers), stating that a new version of MCW462-U featuring 1/2" ID hose barbs -instead of quick connects- will be in production within 2 weeks.

Thanks to this thread here :) You guys are great!

Bruce
www.cooltechnica.com

Xel, thats what I think too, less water at a cooler temp, or more water at a higher temp, there is obviously the perfect point somewhere (and every setup is going to be differant

So do they admit that 1/2" is better than 3/8" now? Or are they just doing it because everyone demands it?

because "everyone" demands it

Gabe is no marketing slouch

open up the 462-U
look at that inlet nozzle

what good is a 1/2" barb gonna do ?
better off adapting to a piece of 3/8" copper tubing

So they didn't change the block for the new fittings, you say?

just drill it out...

mx-6* -
"So do they admit that 1/2" is better than 3/8" now? Or are they just doing it because everyone demands it?"

MX- They ran some tests trying various fittings, and they'll publish their results soon (I have the benchmarks results already but I can't post them). Basically they're thinking it makes a difference with 226W pelt use. But I'm sure demand plays a role, too...

Unregistered - "what good is a 1/2" barb gonna do ?
better off adapting to a piece of 3/8" copper tubing"

Wrong info. They changed the design for 1/2" ID hosing, and the inside hole is even bigger than 1/2" ID. The block will come with standart barbed 1/2" fittings - READ: NOT SPEED FITTINGS-

Contact oppainter if you want he's testing them already.

Bruce

New Water Block
 

Here is some pre info. on a test just recently finished. The Gemini Blocks appears ahead of a few top contenders. GEMINI BLOCKS

I called and talked to Peter this morning. I can tell you he has a brain. His idea's come from a technical background and not from a trail & error method that some use.

Reviews:

Quoted From John Hill @ Overclockers.com

"Pump: Eheim 1048 (158gph)
Radiator: D.D. Heater core (low pressure drop)
Tubing: 3/8" ID Silicon
Vertical lift: 4 inches (insignificant)
All Water Blocks 3/8" ID straight hose barbs
All Water blocks inlet water temperature 25C
CPU: Athlon XP1600 12.5x145 (1822 Mhz) @ 2.05Vcore = 112.5Watts


Gemini High Flow Core Temperature=41C Efficiency=.14c/w (consistent)
Gemini Low Flow Core Temperature=41C Efficiency=.14c/w (consistent)
Swiftech MCW-462u Core Temperture=43C Efficiency=.16c/w (consistent)
D.D. Maze-2 Core Temperature=45C Efficiency=.17c/w (best of three tries)


I was totally surprised by the results, especially the Danger Den Maze-2. I was equally surprised that the Gemini Low Flow block did as well as the High Flow block, considering the slower flow rate that the 1/4" channel causes. (120GPH vs 158GPH)"

Dodgeviper - "I called and talked to Peter this morning. I can tell you he has a brain. His idea's come from a technical background and not from a trail & error method that some use. "

Gemini block is just an imitation of Fixit's Spiral block. So first of all, it's not his unique design (you said Peter, right?) it's Robert "Fixit" Wright's design.
The Geforce + memory blocks that he's selling are also Blade Runner's design.
Therefor we all know where his ideas come from. The source is Trial & Error method that some use...

I also highly suspect the results that you posted...

Bruce

dacooltech, Now that you have gotten that off your chest. Maybe you should take a tour on the net. You will find that I did not report those figures but someone who is VERY honest and up front with his post. Someone who many look and turn to when they want an honest answer. I can't even begin to reply to your comments of who makes what. I know that Gemini has been in business for over 20 years and they have the products in stock....

DodgeViper - I didn't mean you posted the results your self, I meant the results of John Hill that you posted here...

I don't know for how long Gemini is in what business, but one thing I know is that they're new in the watercooling business.

Other than that if you can add links to your posts that helps. I love to surf the net for every bit of information, but I have other things to do, such as running a business called Cooltechnica.com.

It clearly states on the Gemini website that the high flow block wont do any better unless the flow is 360GPH.

Does anyone have that kind of GPH in a 3/8" system?

Comments on test results posted by DodgeViper
 

41C for a water cooled system is NOT very good.

The comparison of the blocks is only valid for a low flow system using 3/8" tubing, a Eheim 1048, and the configuration (whatever it is) of the rad/fans.

The test data is lacking some info - what was the air temp going into the rad and exitting the rad? What was the airflow rate through the rad? Was the fan mounted with a spacer/shroud or simply tied to the face of the rad?

The results of the two different Gemini blocks is a very good example of the fact that different blocks work best at different flow rates and that if the flow rate isn't set properly for the total system (block,rad,hose) then you can get much poorer performance than you would if the flow is properly matched to the system.

To date I've only seen what I consider 2 decent tests of water cooling system components published on the web - one of which was Joe's, the other was the radiator testing done at OC.

NOTE- decent tests in my view include all the information (temps and flows), are conducted at multiple points (air and water flow and heat loads), use proper measurement techniques, and are conducted in a repeatable manner using scientific methods. Without this type of testing, a user is left with at best an incomplete (and at worst misleading) set of information which can lead to poor results in their system configuration.

here is a test where the gemini's win, how close are these to fixittt's block? fix's block may perform better than th gemini's still, and have a big lead over the other blocks

Did you notice that this was highly OC computer that the blocks were tested on?

I posted these results only to let others know that there are other blocks on the market. The complete test will be made public soon. John stated to me today when he gets a few free hours he will post his complete cooling test. Those results were a teaser to us over at OverClocker because we had been waiting. There were to be other blocks tested but it seems a few of the top dogs did not want to be included... The rad that was used looks just like mine. In fact John's rad I copied, I just made a few changes with the tapered sides.

http://forum.oc-forums.com/vb/attach...&postid=568007

This sounds eerily familiar. :p

And I didn't know that a waterblock could refuse testing when it saw the competition. :rolleyes:

And I agree, 41C isn't very impressive at all from what I've seen. Do you really think that this guy is the only person to overclock a computer that much? :shrug:

You do realize thats over a 25% overclock?

Running a XP 1600 at 1800mhz(way over XP 2000's clock speed) with temps under 45C to me is amazing in itself.

No, but when ask to provide a water block to be tested they did not.
Well most testing I have seen are on a rather basic computer and typically not OCed.

My system will have a GEMINI HIGH VOLUME SPIRAL WATER BLOCK.

EXL, Glad you have a pair of eyes and can think for yourself.

Dodge - yes, I had indeed noticed he had a nice OC going for a XP1600 system, although not untypical of the later steppings in wc systems. And yes, I agree that it is good to see testing done at other than default speeds/power loads.

Don't take what I said the wrong way, I am NOT degrading his testing. Heck, his full test data and methods aren't even published yet :) When they are, they might very well contain some of the missing data. However, my general comments regarding good testing methods and results for wc systems stands.

Simply consider what I stated as a precautionary note regarding making general assumptions about the relative merits of one widget over another based upon a test done within narrow parameters and/or with incomplete data.


Now if we can just get Joe to use higher flow rates (like in the 300gph range) for his next waterblock roundup :D


Xel - it really depends on the stepping of the XP1600 used. Remember that the only difference between an XP2000 and an XP1600 of the later steppings is which bridges are laser cut and the testing done. Usually the bottom end of the speed bins are simply filled with what is left after the other bins have met their quota. As a result, the "true" OC in this case might be more in the 10% range ;) As I said, a nice OC, but not earth shattering

http://www.case-mod.com/ is selling them.

Do you think they would give me one for my comparison? :D

EMC2, I doubt it, Joe just runs the hydrothruster at full speed, and each block gives him whatever waterflow it gives him, depending on the restrictions in each system

Are there any major design differences between the Gemini and Fixittt spirals? IIRC, somebody did some casual testing and found the Fixittt spiral and maze2 to be tied with about a degree or so of difference. When Joe's socketA roundup is finished, it'll be interesting to see the results.

On another note, both the Swifty and maze2 can be used on socket-478 platforms with little hassle. This was actually a major decision point for me when I ordered my block since I wasn't sure of my next CPU socket. Of course, I'm sure that somebody could rig a hold-down for the Gemini given the people in this forum :D

(I'm assuming that the Gemini wasn't designed for socket-478. I didn't see any mention to P4 systems on the Gemini and Case-Mod.com sites.)