Best possible radiator discussion

[fhorst]

Hi All,

I'll need your help (again).

First, some basics. Please correct me if I'm wrong.

First part:
- Single pass rad: The coolant flows only one way, in at the top, out at the bottom. Is (in theory) the least restricting.

- Double pass rad, coolant flows down and up again. in and out at the top. More restricting.

- Triple, quads, etc, more curves in the coolant stream, more restricting.

More curves are more restricting.

Second part:
The surface. More surface (square cm), is better cooling.

Third part:
Tube size. Bigger tube size, is better flow, but for computer watercooling, 1/2" is standard.

-----------------------------------------------------

So, if I want the best cooling radiator what would be better?

Statement: (for this statement airflow and cm2 are equal)
- Take 1 large rad, and let all the coolant flow through that one?
- Take 4 smaller rads (together equal to the large one) and let the coolant flow simultaneously through all 4 rads?

To my idea, the 4 rads should perform a lot better. As the velocity of the flow is divided in 4, the water stays longer in the rads, and has more time to transport the heat. Also, it is way less restricting.

Problem in a computer is that we don’t have enough space. In a big tower, we have space above the PSU, about 15 x 20 cm. from the back, to the CD-ROMs. (That’s about 25 cm)

The “D-Tec Heatercore” is nice thick, about 5cm. but, the way I see it, id better could have been 20 cm. Place one fan in front, and one in the back, and we have the largest possible cooling surface that can simply fit in a computer case.

Only thing is, that IMHO we should use 4 radiators, and split the coolant flow in 2.
Lead the flow though the last 2 rads, and from the last 2 to the first 2 rads.
On this way, the velocity is divided, the hottest coolant comes in contact with the “pre-heated” air from the first 2 rads, and the cooled coolant gets in contact with the fresh air, and looses its last degrees.
(air and rads) back, fan sucking air, rad, rad, rad, rad, blowing fan, front
(coolant flow) pump, split, to back 2 rads, to front rads, combined again, CPU block.

Did anyone tried this? And with what results?
The best radiator should be able to give a coolant temp only a few degrees above ambient temp. I’ve tried quite some radiators, but none seem to be able to manage that (unless I place a lot of fans on them)

Currently I’m using 3 rads, and that goes Ok, but takes a lot of space ;-(

Thanks for your imput!

[BrianW]

- Take 4 smaller rads (together equal to the large one) and let the coolant flow simultaneously through all 4 rads?

To my idea, the 4 rads should perform a lot better. As the velocity of the flow is divided in 4, the water stays longer in the rads, and has more time to transport the heat. Also, it is way less restricting.

It is not about time. It as about deltaT (the temperature difference it takes your particular setup to dissipate x amount of watts through your radiator tubes to the fins between them. This is best achieved by high flow. There is a knee in reduction of deltaT vs flow, but in most of our watercooling systems we are below the knee.

To obtain the real performance of the coolant side of a heater core you need to know the velocity of the coolant through each tube. Then you can use that number in conjuction with amount of tubes to predict the deltaT you will have with the x amount of watts your heat sources (things you have water blocks attached to) are adding to the coolant.

Hope this helps, and hope I am not totally off base...

BrianW

[fhorst]

Thanks for the reply.
I know that a higher flow will give a better dissipation of heat.
Only problem might be that high flow is more restricting.

I tried this with my BlackIceExtreme and My D-Tek First making them parallel, let the flow run from the Dtek to the BIX, and serial, letting the flow run through both rad's at the same time.

I don't have a flow meter, but I could see that the flow was in a serial setup a lot higher.

I don't have a real large rad, so I don't know how restricting they will be.

With WC we all have a space issue, and the only large space I can think of (in most computer cases) is at the top.
So, with that in mind, How and what can we place there?

3 or 4 BIX or Dtec's would be great, with 3 to 5 fan’s sandwiched. (Also higher airflow will boost the performance, but we all want to keep the noise level down)

Any ideas?

Reason for this thread is also: no matter what good waterblock a person has, when the coolant that enters the waterblock is "to hot", the performance is not what it could be.

[fhorst]

One other question, you all might be able to clear-up:

How much degrees of water temp rise do you "need" to give one degree extra CPU temp?

I heard something about 3 degrees water temp = 1 degree CPU

The answer will help me a bit on calculating how much efort I need to put in to the Rad's. ;-)

[Groth]

If all else is unchanged, it's one to one - one degree cooler water means one degree cooler CPU.

A suggestion - if you use multiple radiators in the top of your case, instead of a fan or two per rad, use a single blower. Mount it unshrouded in the bottom front with a big open intake. Then close off all vents but PSU and rads. Positive pressure!

[fhorst]

Thanks for the input and sugestion!

My problem will be with that setup, is that I then have no place for my hard disks....

That's why i want to place all in the top of the case..

I could do some re-desiging, to see if I could make it fit.

Any sugestions?

I got a 2 pass D-tec heatercore, A 2 pass Black Ice Extreme, and a single pass motor rad, about 15 x 30 cm...

I was thinking of letting the blower suck the air through the big rad from the case, and then push it through the BX and Dtek.

[PooJou]

What exactly are you cooling?!

You ever heard of the term 'diminishing returns'?

If you are running some Dual overclocked 3.5ghz Peltier cooled Beast then ignore this

But yeah, for a normal pc, that is sorta overkill... i mean maybe the Dtek and the bix together would be good, but anymore and it's sorta rad for the sake of rad and 1c less. if you know what i mean

[fhorst]

I did not know this "law" so I looked it up:

the Law of Diminishing Returns: when a fixed input is combined in production with a variable input, using a given technology, increases in the quantity of the variable input will eventually depress the productivity of the variable input. (Malthus argued that decreasing productivity of labor would depress incomes).

It's a bit like the 70/30 rule. With 30% effort, you can get 70% result. To get the other 30% result, you need to put in 70% effort.

Overkill? I don’t know. When I talk to ppl about WC, they say its overkill.

Getting a new CPU block to get the temps 3 degrees down is that overkill? Don’t know. But what will a new block do if your water temp is 30 degrees (or hotter)? not much.

Better first get those water temps down.

A chiller is expensive! (And it costs a continues amount of electricity) so better try on an other way.

The idea is to get the best rad for WC, and that means to get as much cooling area in the smallest dimension.

Currently I’m cooling: CPU, 2.8 @ 3.5, GPU, NB, 3x HDD.

With my BIX and D-tec, my temps stay nice low, about 39c (load) on my Abit IC7-Max3 (Abit temp probe is known to give higher temp readings then actual temp).
My water temp is (I guess) about 27c., ambient is 20c. If I set my fans at full blow, my temps drop 2c.
So not bad…. But I guess I’m one of those who will put the 70% in, to get the last 30% of result.
To get the water temp closer to ambient I need to increase my air flow (working on that, got me a blower) and/or increase my cooling area.

I’ll post some pics when I’m ready ;-)