Whats the diff between the high and low flow rate gemini?

[everx]

according to that article http://www.overclockers.com/articles546/index03.asp
they both had same results, so im wondering whats the point of paying more for one that doesnt give better results?
Thanks

 

[mustangman]

according to that article http://www.overclockers.com/articles546/index03.asp
they both had same results, so im wondering whats the point of paying more for one that doesnt give better results?
Thanks

One thing you have to consider is that Hoot used the same flow rate for all blocks. The high flow rate block may pull out ahead with more flow and the low flow block may start to lose. The high flow is rated for flow above 240gph and if I remeber right it was tested at something like a 150 at this flow rate there probably isn't any good reason to buy it. If you think you may invest in a bigger pump later on well then it may be worth the five bucks then again...

 

[Silversinksam]

If your going to use a Peltier, then the high Flow rate would be the best choice.

 

[The Overclocker]

the high flow rate version has more surface area which makes it a better option for people running a peltier or high power cpu, the low floware version also increses the ammount of work the pmp has to do making it produce more heat whcih could be a problem in some computer systems with hot pumps.

 

[BillA]

come on sports, read the posts

Hoot used the same pump for all his tests
but the pump's actual output depended on the sum of all the flow resistances of the various pieces in the coolant system for that specific test

Hoot's flow tests were flawed
do NOT assume anything about 'equal' flow rates, they were not

what Hoot demonstrated was the relative performance of the wbs using his pump (with the coolant at 25^C)
no more, no less

"rated for flow above 240gph"
pure sales B.S., IF you have such a pump you could put 600gph through it
what does "rated" mean ? (seen a flow vs temp curve ?)

surface area ?
do you KNOW the surface area of the 2 ?
or are you just presuming that . . . .

if indeed the 'high flow' Gemini has larger channels to accomodate this higher flow (at a more tolerable head loss);
then one would assume that there is LESS surface area,
no ?

be cool

 

[Highlander944]

Good morning all.

BillA stated correctly the idea of 'sum of flow resistances'. The low flow block is designed to maintain 'fast' water in the presence of lower flows. The High flow is designed to reduce backpressure for those using higher volume pumps.

The high flow does expose more surface area to water. The water channels are the same width in both blocks, however the high flow model has deeper channels. Thus the fins, if you will, have larger area in contact with the water.

The ratings I have stated are more of a guideline. If you are only pushing 100gph flow rates through your system, the low flow block will yield better results do to the more turbulent flow of the water and thus better thermal transfer in the water channel. If you have a large pump and are cooling multiple devices, the high flow unit will provide better flow volumes, but also requires these higher volumes to produce turbulent flow through the deeper channel.

Hope this helps.

Peter

 

[The Overclocker]

surface area ?
do you KNOW the surface area of the 2 ?
or are you just presuming that . . . .

if indeed the 'high flow' Gemini has larger channels to accomodate this higher flow (at a more tolerable head loss);
then one would assume that there is LESS surface area,
no ?

be cool [/B]

the width of the channels are the same on both of the blocks, but the high flow rate version is twice the hight of the normal version. so the hight of the channels have doubled. which means there must be an increse in surface area.

the normal version has internal channels of 1/4 inch, while the high flow rate one has channels of 3/4inch, that accounts for the flow drop

 

[BillA]

good clarification Peter (and the overclocker)

indeed, a deeper channel will provide greater surface area and less flow resistance

and this raises the interesting question:
as the ratio of effective surface area to coolant volume has changed; does this not decrease the wb'c efficiency ?

it certainally does with radiators (which are forced convection heat exchangers just like wbs)

bigger is not better

be cool

 

[schoolie]

It's not very practical to get 240gph with the high-flow block. Maybe an Eheim 1060, or Iwaki monster pump would give you that flow, but the pressures in the hoses would be higher than I like. Highlander944, have you managed 240gph with this block and a radiator?

Thanks!

 

[Aesik]

and this raises the interesting question:
as the ratio of effective surface area to coolant volume has changed; does this not decrease the wb'c efficiency ?

Interesting and very loaded question Bill. I could write nearly a thesis and a half on this alone.

Basically it comes down to if the Reynolds number can be maintained at the same, or nearly the same as with the smaller block. There is alot of other discussion about this and other aspects of the Gemini blocks in other threads for those who want to research and read.

 

[nihili]

Let's not forget that the added surface area is at the top of the fins where it is far less efficient. In order to know whether it was any help at all in cooling, you'd have to know the temp at equilibrium of the top of the fins. I don't think anyone has even attempted this measurement.

My guess is that there is very little benefit in virtue of the increased surface area. However, there may be benefits in virtue of other factors soch as decreased resistance.

nihili

 

[BillA]

gee, the heavy hitters are gonna beat up on the Gemini again
hi ho
but it is fun

ok, the question was pretty much rhetorical in that Peter cannot answer it

but there is quite a bunch that is known, it impacts directly on wb design
- and most wb designs are headed in quite the wrong direction

the distance from the source of the heat is the dominating factor,
all that real estate 1.5 in. from the die is so much junk
it is the heat convection rate that needs optimization within 1 in. of the die
long channels, folded or curved, don't do it
and deep channels are worse yet

I skirted the question of geometries on the other thread, but lets compare several:
how is it that an un-finned design (Swiftech) can be about equal to a labyrinth ?
different 'levels' of execution ? or different intrinsic 'merit' to the design ?
can the mods that improve a Swiftech ALSO improve a labyrinth ?
(am doing so now, should know in several days)

from my perspective its all about that prime real estate in the die area

and of course ALL readers are thinking INTENTLY about the BMW of wbs:

image from THG blah blah (thanks to EMC2)

be cool

 

[schoolie]

My favorite water block!!! A little pricey, but worth it for the excellent mounting mechanism and great performance.

 

[nihili]

gee, the heavy hitters are gonna beat up on the Gemini again
hi ho
but it is fun

Just to be clear, I don't mean to be beating up on the Gemini. If I were going to buy a waterblock right now, I'd probably buy a Gemini. At the very least it would be one that I would consider very seriously. I do think there are open theoretical questions about why the block works. And I don't think the difference in surface area between the high and low blocks will have a significant impact on their effectivenes except as it contributes to other factors such as cross sectional area of the flow.

nihili

 

[BillA]

easy, easy
only meant as a joke (and I am among the MOST guilty)

going with the flow for a moment;
if one has a larger pump, would it not make more sense to use the smaller wb to get even higher velocity/turbulence/convection rates ?

be cool

 

[nihili]

easy, easy
only meant as a joke (and I am among the MOST guilty)

going with the flow for a moment;
if one has a larger pump, would it not make more sense to use the smaller wb to get even higher velocity/turbulence/convection rates ?

be cool

I suppose that would depend on the rates of increase of the various functions. If they were all linear, it wouldn't make a difference. But I suspect none of them are linear. So it would be a matter of trying to do some optimization for the specific parameters. I don't have a good grasp on the how the relevant funcitons increase... I guess it's time to start studying.

nihili

 

[Aesik]

easy, easy
only meant as a joke (and I am among the MOST guilty)

going with the flow for a moment;
if one has a larger pump, would it not make more sense to use the smaller wb to get even higher velocity/turbulence/convection rates ?

be cool

In the particular case of the Gemini block, I think it would be a wash if just the CPU were being cooled.

Now if there are multiple other devices being cooled (NB, GPU) the high flow model might have more merit because it will have less overall head loss and allow higher flow to the other cooling blocks. Hard to say without some more in-depth analysis (like I don't have enough to do anway!) or some more serious testing.

 

[Highlander944]

hey Billa, it's always fun when u get involved. haha

not to debate my 'design' method, but I did look at the block as a system. I agree with you bill, somewhat, on the flat vs maze type approach. Given a huge flow rate, sure the flat can work. BUT given a slow flow, what works better?

That is really my approach. Given reality and some practical limitations, what design is better suited to remove the heat? As I said before, you don't 'need' huge GPH to remove the core heat. Most use copper, so that part is constant amoung competitors. The primary difference is how our copper interacts with the water (and a few other differences).

just my 2 cents.

 

[BillA]

some numbers (all at 70-71W with a heat die, ~ 100W 'CPU power'):

gpm . . . die^C . . . wb
1.00 . . . 44.8 . . . . MCW462U (@71W)
1.00 . . . 45.8 . . . . MCW372
1.00 . . . 46.8 . . . . Innovatek
1.00 . . . 46.8 . . . . Maze1
1.00 . . . 49.9 . . . . Cooltech

now there are a few minor problems with this data (which is why it is not "published")
but its going to take a LOT more than your words (and Hoot's single test) to prove what you assert (without data)

be cool

 

[Hoot]

Boy, I take a few days off from the cooling forum to play with a new motherboard and look what I miss. Sure, the Gemini blocks worked better than the D.D. and Swiftech with my puny pump, but explain to me then why both D.D. and Swiftech sell that same model pump with the intention of having it used with their blocks? Are they both ignorant of the importance of high flow rates, or do you suspect that perhaps they tried higher flow rate pumps with no appreciable improvement in performance?

Read the water cooling posts. That puny pump I am using represents the "upper crust" of what I see a lot of other water cooling enthusiasts using. Why not design a waterblcok that is optimized for the majority of pumps out there. In pursuing the ultimate theoretical block design, don't loose sight of the average water cooling system. If people have to buy a pump that delivers an honest 600gph to gain 2C improvement with a particular block, that block is doomed.

Hoot