WOW - huge temp drops with TRIPLE RAD setup

[jackal2513]

long post this so i can share all my latest findings.



The main goal of my rig is low noise. Up untill last night I had the innovatek P4 stock kit rigged up with just x1 virtaully silent Papst 120mm 7v fan pushing air through the Maxxepert (approx 900rpm, <20 cfm). Becuase of the very low fan speed and prolly cos innovatek isnt the best in the world and also cos of my relatively high ambient temps.. these were my results last night (note that i have removed the Geforce Block and gone back to air cooling the GPU 'cos i was getting crashes when playing games - this has now been solved):




IDLE

cpu = 38 C
water = 29.2 C
ambient = 28.3 C
case = 28.3


UT2003 continous until temps stabilised, max resolution, all options turned on - more of a real world "load" IMO

cpu = 50 C
water = 37 C
Geforce = 52 C





Today I did avery rough first install of my triple rad system with a new 1048 eheim. This consists of the original maxxpert rad that comes iwth the innovatek kit, and two further black ice micro rads. These are installed behind teh 2 stock lian li case fans at 7v. Now i dont have all my bits yet so tonight all rads are running in series like this (see pics below)

eheim > cpu > gpu (diconnected) > Maxxpert > BIMicro 1 > BIMicro 2 > Reservoir >eheim

Anyway, ambient temps tonight seem a little lower but here are the results:





IDLE (after cooling from prime 95 to stabilise system)

ambient: 27
water: 25.4
cpu: 34
case: 27.2


PRIME 95, 42 tests ... temps stabilised

ambient: 27
water: 29.4
cpu: 45


UT2003 continous until temps stabilised, max resolution, all options turned on

water: 29.1
cpu: 42/43 oscillating


Then back to IDLE

ambient: 29
water: 25.6
cpu: 34
case: 28.9



note that in the final idle the ambient has gone up as its now besdtime and it always gets warmer in my lounge at this time. Also, after UT2003, it took about 10 minutes to get the CPU back down to 35/36 C, it hovered at around 38 for sometime.

Now these temps arent the greatest in the world but they are a hell of a lot better than they were, especially the continuois UT2003 temps. The load difference is much bigger than the idle presumably indicating that my system is a bit more efficient now. I didnt run Prime 95 last night but i bet it would have had teh CPU pretty near 55 C. The thing is this system is a temporary install till i get all my bits. I intend to run tehe rads in parallel, remove the GPU block altogether (its just creating extra tubing and blocking flow at the moment), and fit 1/2 inch tubing from the pump up to the block and down to splitter before the rads. Also, i had to use some real crappy innovatek tubing as well to connect the micros up so this will go as well. Hopefully with that little lot i'll get things down another few degrees or so.

The idea behind teh 3 rads it to cut down on noise. With the singel rad I found i could reduce temps by 4 degrees by putting a noisy YS tech fan on the Maxxpert (as well as the papst) in a push pull config but that was way way too noisy. On the BIMicors I will evetually have x2 Papst 8412NGL so all three rad fans will be almost inaudible.

One thing i find odd is that my water temps are lower than ambient .. i didnt think this was possible. Maybe the place where i take my ambient temp isnt strictly true (just in front of the top part of the case, away from the fan intakes). Also, my water temp is taken after the rads in the reservoir. I definitely trust the thermometer ive got .. its been very very consistent all along. One thing that would be interesting to know is how accuarte the MBM cpu temp is from the gigabyte board ?

Anyway, hope some of that is of interest to some of you. Im prettty damm pleased... esp as before after load my case was getting hot and the tubes and parts of the rad actually felt hot to touch - this is bound to help my passive cooling on teh northy and Mobaord etc.. When i get my Canterwood board next week I will also reseat teh cpu as for teh first time i now have some artic silver 3. After that I will be back again concentrating on silence and thinking about my 2nd WC setup to deal with the HD's, Northy and GPU. Sadly, my GeForce fan is now back on and it doesnt half increase the noise. If theres an easy way to get a whitewater to the UK then i'll also be interested in that as well.

One last thing .. i find it quite curious that during prime 95 and ut2003 tests the water temp was pretty similar yet the cpu temp differed by more than 2 degrees ?

cheers

 

[emericanchaos]

nice looking case. seems roomy. what one is it?

 

[SemiCycle]

Another reason you temps dropped is because you increased the total volume of water you are using. The water you have in the loop, the harder it will be to incease the tempature of the total mass of the liquid.

 

[WejRepus]

nice looking case. seems roomy. what one is it?

looks like a pc-70

 

[Sniperboy]

Another reason you temps dropped is because you increased the total volume of water you are using. The water you have in the loop, the harder it will be to incease the tempature of the total mass of the liquid.

The volume of water will not change overall temperatures; it will only make the temperatures change slower.

 

[Cathar]

nice looking case. seems roomy. what one is it?

I think that's the reason why he says what it is in his sig.

Look there for your answer. No need to ask.

 

[Mikeonatrike]

Nice, im gonna try my system with two chevy heater-cores when i get my WB all done and proper barbs on one of the cores.

 

[Korndog]

I think that's the reason why he says what it is in his sig.

Look there for your answer. No need to ask.

Damn, cather, u gotta sharp eye.

 

[jackal2513]

"The volume of water will not change overall temperatures; it will only make the temperatures change slower."



True, but in real world usage terms it surely amounts to a visible lowering of temps. I mean imagine if you had a reservoir the size of the Pacific.. would take an awful long time for a P4 to heat up that little lot.

 

[Cathar]

True, but in real world usage terms it surely amounts to a visible lowering of temps. I mean imagine if you had a reservoir the size of the Pacific.. would take an awful long time for a P4 to heat up that little lot.

Nobody uses the Pacific for a res for their PC.

Water takes about 4200J of energy per kilogram to raise it by 1 degree celcius, which happens to be bang on 1 litre of water.

A hot overclocked P4 can put out around 100W of power under full-load, possibly more.

100W is 100J/s.

This means that it would take a P4 around 4200/100 = 42 seconds to raise 1 litre of water by 1C.

Let's say you had a big bucket with 10 litres in it (around 2.7 US gallons).

That means every 7 minutes the water would rise by 1C. In an hour of gaming, the water would rise by around 9C (assuming no additional cooling).

Water is good at holding heat, but unless you have something like a bathtub full of it, it's still going to get just as hot through normal gaming use as a system with much less water.

 

[Diggrr]

Well, I'm glad you're happy with your settup, but how did you get your water temps below ambient using only a couple of radiators?

 

[jackal2513]

hmm.. im not sure how or why they are below ambient ....

one thing is for sure, the cpu temp speaks for itself... 43 C continous gaming load as opposed to 50 C so im not too bothered if something in my ambient/water readings is slightly askew.

The only reason i can think is that my ambient reading is off cos its being taken in a funny position which is not indicative of the temp of the air going into the rads ? One thing is for sure, when i move teh thermometer aroudn the room the temp does change a lot, especially at higher and lower heights.

 

[Junglebizz]

How much of a difference can you get from running radiators in parallel as opposed to series?

I am in the middle of making up a dual rad system and I was thinking about that.

 

[witchhunter]

I'm on the verge of completing my triple-chevette hcore parallel setup, I'll post some pics when its finished.

Its not that simple to judge whether parallel is better than series. First of all you'd need to know the specs of your hcore (or whathever you use), to be more precise its performance on different flowrates. Generally tests show that higher flows are better, BUT the difference is smaller the lower the airflow across the hetearcore... this happes in case of chevette.
There's also one important thing, flow across the whole setup, the higher the better and it's obvious that joining hcores in parallel lowers their total restrictions resulting in higher flows (thus lower temps) for the whole setup.

In my case I'll be having something like 20CFM of air across each of the 3 hcores, which makes them perform equally well on virtually every flowrate, thus i've chosen to connect them all in parallel to increase the total flow across the whole setup...

 

[jackal2513]

How much of a difference can you get from running radiators in parallel as opposed to series?

I am in the middle of making up a dual rad system and I was thinking about that.

Not sure yet .... hopefully it will make a difference. I'll let you know latest middle of next week on this thread. Just waiting for my Y splitters and some new tubing.

As its shortens overall distance the flow will increase a bit but my system is not really dependent on flow that much (innovatek) and i believe that with black ice there is a mjor pressure drop.. also 3/8 tubing.

But all rads will be working more efficiently cos water temps going into the BIMicros will be larger therefore more heat exchange. I did notice last night that changing my BIM fans to 12v from 7v only lowered idle water temps by maybe a further 0.3 degrees C.. so obviously the water going into them is quite cool already... they are just squeezing the last bit of heat out of the water.

Bear in mind that my 3 rad setup is for silence. If i put x2 100cfm fighter jet fans in push/pull conifg on each rad then im sure load temps would again drop substantially.

 

[E_Man]

HIGHSCHOOL CHEM ROXXORS!

 

[Junglebizz]

I am also using 2 chev cores in my dual setup. The 2 cores that I have, though, are different.. One has U shaped fins that are close together, while the other has V shaped fins that have bigger airgaps between them.

I am working on a cabinet design that will use a push/pull system using one 120mm high CFM (not sure how high..) fan in a wind tunnel configuration (push through the tighter heater core, pull through the spaceous one). I figured that while I was at it, I would also perform tests on this config between one radiator and 2, and possibly, now that you brought it up, the difference between series and parallel.

 

[FIZZ3]

Maybe temperatures would be even better with a single heatercore and an effective shrouded fan(s) setup? :>

The reason is that it would seem a better idea to increase the heat transfer possible at the site meant for it (the radiator), instead of adding multiple poorer performing sites. Also, you'd save on tubing length and bends and that's always a good thing.

 

[jackal2513]

fizz, possibly (although a heater core would have to be pretty spectacular to beat 3 rads in parallel) but I dont really want to ditch the innovatek stuff that I have paid for and its an aesthetic thing as well... the BIMicros fill out teh bottom space rather well i think and also make good use of the 5v intake fans there which kind of have to be there in any case (otherwsie i'd have zero case air flow)

 

[FIZZ3]

fizz, possibly (although a heater core would have to be pretty spectacular to beat 3 rads in parallel) but I dont really want to ditch the innovatek stuff that I have paid for and its an aesthetic thing as well... the BIMicros fill out teh bottom space rather well i think and also make good use of the 5v intake fans there which kind of have to be there in any case (otherwsie i'd have zero case air flow)

Hehe ok, just playing the devil's advocate.