Second rad in a loop - advice please

[Cyprio]

Looking for a bit of advice on expanding my watercooling loop guys.

I currently have the following:

Pump (Aquastream 50Z-DC12)>PA160 rad>Storm CPU block>Polar FLo TT VGA block>T-piece (Fillport)>Pump

This was fine for my 3700+ SD and 850XT (both idling around 30 and hitting 36/7 on load with fans on full). However i have just swapped my 3700 for an Opteron 146 and want to overclock this baby to the FX heights it can reach. Unfortunately its not one of the early steppings that will do 3.0Ghz on stock volts so its temps are going up.

Basically i was wanting to add another rad between the storm and the VGA block to take more heat out of the loop. My PA160 is sitting in the bottom of my V1100 like this:



With the loop looking like this:



The idea would be to mount the second rad to the top of the case above the gfx.

The questions i have are:

1. Would a 120.1 be enough (bearing in mind that i want the fans on as low as poss) or would a 120.2 be better?

2. Is my pump (Aquastream 50Z-DC12) powerful enough for the new loop with two rads?

Thanks in advance,

Cyp.

[gmat]

No problem for the pump. But if you get the 120.2, you can actually run it alone, it will keep it simple Could you lay the 120.2 along the bottom ?

[Cyprio]

Originally Posted by gmat

No problem for the pump. But if you get the 120.2, you can actually run it alone, it will keep it simple Could you lay the 120.2 along the bottom ?

It would be tight.

If the pump will manage it then i would rather have two rads with the fans running very low voltage rather than a 120.2 at higher speeds. Am i correct in thinking that this would give me better cooling at lower noise?

[BGP Spook]

Let me see...

(Regarding rad placement in OP)
Guess work:
CPU: 200 watts (89 watts standard peak?, 3.0 ghz @ 1.8 V = 192 watts?See OC clac)
VGA: 100 watts (92 watts standard peak, allow for slight overclock)
Pump: 10 watts (unless overvolt?)
Flow: ~3 LMP (conservative)

More Guess work:
4,187 J*L*C/s for H2O
3 L/60 s = 0.05 L/s
0.05 L/s * 4,187 J*L*C/s = 209.35 J*C/s (J/s=watts) so
209.35 watts*C

Anecdotal summary:
So if your Storm is passing ~200 watts a second from your Opy to the water then you should see ~1 C difference when comparing the "in" water temperature with the "out" water temperature. (This is a pretty BIG assumption, in reality there is probably much less change in temperature than that, too many variable ignored to list.)

Anecdotal conclusion:
With the above in mind...
It is doubtful you would notice any difference by putting the rad immediately after the CPU vs. in any other part of your loop.

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

(Regarding last post made by "Cyprio")

Just guessing...

If you have your PA160 + 120.2 in the loop, then you would be able to run your rad fans at a lower voltage for the same or better cooling. About how much better and at what voltage....
....is beyond my ability to form conjecture.


Good luck!

Edit: I am execting and hoping someone with more knowledge on the subject will come along and crucify my post. Thus re-filling my leaky head.

[Long Haired Git]

No C/W graph for the PA160.1
No C/W graph for the Polar FLo TT VGA block
Approximator doesn't yet have the Aquastream 50Z-DC12 (I've got the figures though)
So I can't approximate the loop you describe.

Instead, here's an approximation of a similar loop to show the effects of a second rad when the fan speeds are low (5v).

Personally I think the wattage's suggest are over the top, but if someone can point to something credible to show them I will re-evaluate. Until then, I am using 110W for the CPU (highest my approximator handles) and 70W for the GPU.

MCP655 @ 5
Swiftech Storm on 110W
MCW6002-64 as a GPU block on 70W
1.5m of 1/2" tubing
Single MCR220QP with dual 5v fans

***********
EDITED: BUG AS FOUND BY BILLA IN MCW6002-64 PQ FORMULAE.
THERE WERE HEAPS OF DATA HERE OUT OF APPROXIMATOR I HAVE RIPPED OUT AS THEY WERE WRONG.
SEE BELOW MY THANK YOU POST TO BILLA FOR AN UPDATE
***********

[Cyprio]

MCP655 @ 5
Swiftech Storm on 110W
MCW6002-64 as a GPU block on 70W
1.5m of 1/2" tubing
Single MCR220QP with dual 5v fans

Flow rate = 3.47 LPM
Water temp at CPU entry = 12.82°Caa (above ambient)
CPU temp: 27.24°Caa
GPU temp: 24.84°Caa

I understand the theory behind this (just!) and the 27/28 Celcius for the CPU is almost spot on. However my GFX is always 6/7 degrees higher than my CPU. How is this explained?

By the way thanks for workin all that out - it looks like i will add a second rad. Even just do so that i can turn my fans right down.

Cyprio

Edit: does 27 above ambient really mean something like 47ish?

[gmat]

yes, if the radiator air inlet temp is 20°C

[Long Haired Git]

My values are derived from published C/W charts measured a certain way on a certain day with a certain sample etc etc.
Different temp probes from the same manufacturer vary by whole degrees, and then t, the sw reading them has errors.
Basically, comparing temps is a nightmare. Best avoided.
The trick is to take your own measurements using air etc and then tune to improve your temps.

Eg: by moving the probe so that it doesn't actually touch anything hot, you can get near ambient temps reported. W00t! Take photos and post to the nearest modding forum as a cooling god.

[BillA]

from the Swiftech site:
kit/configuration . . . . . . . . GPM
Apex (1) CPU water-block 1.07
Apex (2) CPU water-blocks .82
Apex (2) CPU water-blocks + (1) GPU water-block .75
Apex "Ultra" (1) CPU water-block 2.07
Apex "Ultra" (2) CPU water-blocks 1.77
Apex "Ultra" (2) CPU water-blocks + (1) GPU water-block 1.36

suggest you correct your flow rates

[Long Haired Git]

Whoah: right you are BillA.
My formulae for the MCW6002-64 was out by a factor of ten.
Apologies.

I'll edit my posting to say the figures were rubbing and then repost at the end.

[Long Haired Git]

[This post is a repeat of the original post but corrected}

No C/W graph for the PA160.1
No C/W graph for the Polar FLo TT VGA block
Approximator doesn't yet have the Aquastream 50Z-DC12 (I've got the figures though)
So I can't approximate the loop you describe.

Instead, here's an approximation of a similar loop to show the effects of a second rad when the fan speeds are low (5v).

Personally I think the wattage's suggest are over the top, but if someone can point to something credible to show them I will re-evaluate. Until then, I am using 110W for the CPU (highest my approximator handles) and 70W for the GPU.

MCP655 @ 5
Swiftech Storm on 110W
MCW6002-64 as a GPU block on 70W
1.5m of 1/2" tubing
Single MCR220QP with dual fans.

Flow rate = 5.5 LPM (that's better eh BillA!)

Fans @ 5v:
Water temp at CPU entry = 13.2°Caa (above ambient)
CPU temp: 28.3°Caa
GPU temp: 24.8°Caa

Fans @ 7v:
Water temp at CPU entry = 9.0°Caa (above ambient)
CPU temp: 24.1°Caa
GPU temp: 20.6°Caa

Fans @ 12v:
Water temp at CPU entry = 6.3°Caa (above ambient)
CPU temp: 21.4°Caa
GPU temp: 17.9°Caa


Analysis: Definately room for an extra radiator.


So, lets drop in a second MCR220QP radiator with dual fans and add in 50cm more tubing.

Flow rate = 5.2 LPM

Note that from now on, I'll just report the water temp at the entry to the CPU.
The CPU itself runs approx 15.1°C warmer than the water, and the GPU similarly runs 11.6°C warmer than the water temp.


All Fans @ 5v = 6.5°Caa
All Fans @ 7v = 4.4°Caa
All Fans @ 12v = 3.0°Caa

Analysis: Water temperature is halved, as radiator power is doubled. Simple really!

The law of diminishing returns then kicks in hard. For laughs, lets add in up to seven MCR220QPs with dual fans and 50cm of tubing each and see what happens.

FLOW RATES:

One = 5.5 LPM
Two = 5.2 LPM
Three = 4.9 LPM
Four = 4.7 LPM
Five = 4.5 LPM
Six = 4.4 LPM
Seven = 4.2 LPM


FANS AT 5V:

One = 13.2°Caa
Two = 6.5°Caa
Three = 4.2°Caa
Four = 3.1°Caa
Five = 2.4°Caa
Six = 2.0°Caa
Seven = 1.6°Caa


FANS AT 7V:

One = 9.0°Caa
Two = 4.4°Caa
Three = 2.8°Caa
Four = 2.0°Caa
Five = 1.6°Caa
Six = 1.3°Caa
Seven = 1.0°Caa


FANS AT 12V:

One = 6.3°Caa
Two = 3.0°Caa
Three = 1.9°Caa
Four = 1.4°Caa
Five = 1.0°Caa
Six = 0.8°Caa
Seven = 0.7°Caa


So a case can be made for a second radiator, and for the extreme cooler a third radiator, but a fourth radiator is kind of hard to justify.

Shame there are no PQ and C/W tests for 120.3 radiators....

As per another thread, let me supply one MCR220QP and one MCR120QP with fans @ 7v and no tubing between the two to "approximate" a 120x3 low flow optimised radiator:

Flow rate = 5.0 LPM (that's better eh BillA!)
Water temp at CPU entry = 5.7°Caa (above ambient)


Hmmm, tempting?

[Cyprio]

Wow, thanks for all this guys.

Second rad here i come... well when i have the cash and when i upgrade as it will mean taking all the hardware out to cut the case up.