Why the NanoBlock is better

[UnaClocker]

Ok guys.. I have a new theory on why this waterblock is outperforming a Maze1C.. First, make sure you've read and understand the results of BillA's article at Overclockers.com http://www.overclockers.com/articles481/
As everyone learned in that article, radiators have a happy spot for flow rate, sometimes that happy spot can REALLY improve it's performance.
See where I'm going with this? A couple months ago I moved my whole system up to 1/2" tubing to improve the flow rate. I also went up to a heatercore radiator. My flow rate quadrupled, but my temps remained the same. Mind boggling and hella frustrating.
So. The Maze1C with it's 1/2" fittings and nice wide round channels was not slowing my water down at all.
Now although this little block's whole design is to be unrestrictive, the fact is, my water is being squeezed down into a 1/4NPT fitting, which slows things down alot. This isn't a problem for the block, because with it being so small, it can handle slow waterflow without the water losing it's "bite" from hanging around too long. And with the water flowing so much slower, it spends alot longer in the radiator now. And as BillA taught us, slower through the radiator is a good thing.

[mkosem]

intriguing!

--MAtt

[Sled_Dog]

hehe NanoBlock = funny as hell and requires new mounting system.

[phreenet]

I couldn't quite extrapolate it from your writings there Una , did you mean that the block slowed down the water in the radiator but increased it at the block? or viceversa?

[UnaClocker]

I meant it slowed the water down throughout the system, let's say I was getting 200gph before, now it's more like 100 or so. Now if the water went that slow through a maze block, it'd be bad because it'd be hanging around too long. But in this really small block, it's not hanging around at all. And going through the radiator slower is what's making it perform so much better. IMHO.

[Brad]

or maybe it is related to the thickness of the copper between the water and cpu?

[UnaClocker]

That would be my absolute last guess. I don't think that has anything to do with it.

[pHaestus]

You could plumb in a valve to get the same flow rate in the Nano/ghetto block setup and see how much temps are affected I guess. It seems to me that even at an equal flow rate the nanoblock has the advantage though because its internal volume is so much less.

[ItsSoLARGE]

I thought we all knew the flor rate thing... but for some reason for the last ferw months people have been upping their flow rates like crazy. Anyone know why? was is ever disproved that radiators have a sweet spot and that we should use 1/2in tubing and 750gph pumps?

[redleader]

Wow thats an amazing article. I'm glad somebody actually tested flow rate on waterblocks to confirm it makes a difference.

One thing that I disagree with it the focus on radiator perforamce. Thats probably the least important part of the system. The ratio of differnce of coolant/CPU die temperature to that of air/coolant temperature approaches 3:1 in many systems.

Many of those tests showed up to a 30% improvement in radiator performance at certain rates over max flow. Yet those same temps showed a 240% increase in waterblock performance simply by maxing the flow rate.

Therefore the best performance would be obtained by ignoreing the radiator altogether- its effect on performance vanishing small in all but the worst-case systems and instead maximising waterblock performance.

Furthermore those tests confirm that massively overkill radiators are available for cheap. Most people will never come close to the limits of their BigMama. And if they did they could simply get a larger fan.

Basically those tests indicate that the bottleneck in a watercooled system is the waterclock and not the radiator. Common sense agrees: just compare the surface area of a Maze2 to a BigMama. Its quite clear just from looking at it that more heat can move into the heatercore then can leave the waterblock.

[deeznuts]

i too believe the waterblock is the bottleneck in a system if you have a radiator. that doesn't make the radiator the least important however.

a shit waterblock with a heatercore will perform better than the best waterblock and, say, a Z4 radiator, i think.

to me, that makes it look ike the Radiator is the most important piece. but it is easily taken care of.

deez

[UnaClocker]

Yeah, in my setup with dual heatercores and excellent airflow through those radiators, it's definately up the the block to keep the CPU cool. Whatever makes this block perform so well, I like it.

[decodeddiesel]

I am still amazed that thing works so well. I wish Chip had a little better mounting system to it. I am betting the reason why this thing works so well deffiniatly has to do with flow rates through the radiator/block.

[UnaClocker]

Yeah, flow rate in the block definately seems to make up for surface area.. Who'd have guessed.. As for a better mounting system, I'm guessing that the full size v-tuned blocks work the same way internally, so you get a decent mounting system with those. This one was just meant to be bargain basement cheap. $12 really is a helluva price..

[gmat]

I really think it's related to the waterblock volume. Less mass = less stored heat. Think of specific heat of copper... throw a few thermodynamics laws in the soup and you've got it.
Also the fact it's a straight-through model can help. Remember, the greater the difference in temperature between coolant and WB surface, the greater the heat exchange is.
You must consider also the thickness of WB copper base. Thicker base = higher heat resistance = lesser heat transfer...
I'm eager to see the results of non-restrictive spiral blocks (not Maze2 but ones with *large* and short channels).

[UnaClocker]

Yeah, that too.

[Tweekster]

This brings up several theory questions...maybe it's time for some real world testing...I think everyone agrees the bigger the rad the better, so how about a 1.4 T'bird @ 1.85V, Autozone #398219 heater core with 2 - Panaflo H1A's, Eheim 1250 with reservoir as a standard test system...that should make the WB the weakest link (sorry, couldn't help myself

Test #1 - Standard nanoblock
Test #2 - Custom nanoblock with 1/2" fittings and a 3/4" to 1" internal bore for max flow
Test #3 - Custom block with 1/2" fittings that feed a large channel that breaks down to 4 - 6 smaller, straight channels then back to a large channel for max flow and area of water contact
Test #4 - Standard Maze 2 with 1/2" fittings

Then, maybe, repeat the same tests with the Eheim 1048 for lower flow.

Una, you up for the task?

[gmat]

Yum yum me wants ! me wants !
Anyways Joe's upcoming WB roundup may answer to the 1st part...

[UnaClocker]

If I weren't so broke right now, I'd jump on that.. I don't think the 3/4" internal area is the way to go, only the water in the center of the block would be moving rapidly, the stuff around the edges would be pretty still.

[Tweekster]

I agree that the 3/4" area would be too big, it would just be interesting to see how bad the temps would change compared to the rest....might not be as bad as we think, edge turbulence may help...

Una, for these tests, what do you need?....maybe a few of us in the forum can get or loan you a few things?....would make probably make a good article

[UnaClocker]

Well, I spoke with Chip at Overclock-watercool today. Told him about the interest in a version of the block that can use 1/2" hose. Came away knowing a bit more on how the block works.
It's not that it flows alot of water, it's the velocity of that water. Turns out the hose barbs on it are 1/8 NPT.. What's going on is the water is being forced through a small passage. This is better than a big passage because more of the water that's going through the block actually touches the block. And with it "blasting" through there, the water around the edges can't sit around while only the water in the center flows.
I bounced a couple bad ideas off him before I landed on the idea he's already working on. Don't want to give away any trade secrets, but it will use 1/2" hose, and should flow more water, and while he won't admit it'll perform better, it won't perform any worse.
Sound cool? I'm hoping to get one when he's done making 'em and verifying they work.

[gmat]

Yahhh Nanoblock round-up
Say, "nanoroundup" ???
Una please make sure you get some flow meter in the loop, so it will answer all those flow related FAQ.

[Output]

So I guess a block with lots of small channels going from one side to the other would work great. The channels would then join into a 1/2 in both ends. The small channels would run parallel

[UnaClocker]

Originally posted by Output:
So I guess a block with lots of small channels going from one side to the other would work great. The channels would then join into a 1/2 in both ends. The small channels would run parallel

Bingo.. We have a winner.. heh, Good guess. Should be sweet after it's built.

[Erik Johansson]

Originally posted by UnaClocker


Bingo.. We have a winner.. heh, Good guess. Should be sweet after it's built.

That´s the design I´ve been using for the last year...

/Erik