Dual H20 cooled rig running hot, help...

[wmandra]

Well, after running my w/c setup on my dual rig for a few days now the temps are getting way too hot. Right now the system has been running for a llittle over an hour and the temps are:
case/chipset: 42'c
CPU0 : 54'c
CPU1: 54'c
HDD: 40'c
Ambient: 30'c (est.)

I am using the Swiftech MCW5000 blocks on the CPUs, pump is MCP6000 with the Hayden 676 radiator.

I talked to BillA at Swiftech earlier this week and he suggested trying reversing the fans, so instead of blowing the air from inside the case over the rad, suck air over the rad from outside the case. This didn't help, and if anything made it worse. The biggest problem is the absolutely horrible airflow in the YY2201 case.

So here are my current thoughts for lowering my temps.
1. replace Swiftech 120mm fans with Panaflos

2. have rad fans blowing air from inside case over rad.
the drive bay side of the case stays fairly cool, usually 1-2' within room temp. right now all I am doing is heating up the inside of the case more and more.

3. I need to figure out a better way of ventilating the motherboard side of the case. The only place that I can mount a fan on this side is the stock 120mm opening in the front of the case. I am really not sure what works better though, trying to pull cool air into the case, or suck the hot air out....
The other problem with the MB side is that 120mm fan is on the bottom corner of the front of the case and most of the heat is being generated in the upper back corner.

The current loop is:

RAD -> F&B kit -> CPU1 -> CPU0 -> NB -> PUMP - RAD

all of the WB are running in series and the hoses are all warm to the touch even the one coming out of the rad and since I have decent flow I have to beleive that the problem is either:
A. Rad can't handle heat load of 2 CPUs and NB
B. Rad fans not pushing/pulling enough air
C. F&B kit being betwen rad and WBs might not be a good idea?? might be beter between pump and rad
D. Might get beter cooling with CPU hoses parallel rather than series. Can the MCP600 handle this???

One other thins I am considering right now is to make a simple water chiller using a pelt and an extra WB. I know it wont give me any massive temp drops, and I am expecting maybe a 2-5'c decrease in temps but that might be enough to get my temps down to ambient once I get all of the other problems ironed out.

**EDIT*** As I was sitting here typing this I decided to open up my case and what did I find??? one of the 120mm fans on the rad wasn't spinning...... easy fix, but still didn't solve the problem. Temps are now reading:
case/chipset: 41'c
CPU0 : 47'c
CPU1: 48'c
HDD: 39'c
Ambient: 30'c (est.)

This is definately an improvement, but still not good enough my goal is to run at low 30's idle @ 133 Mhz and high 30's idle @ 155mhz

[bozo]

add another rad maybe?

[wmandra]

Well, I changed the direction of the rad fans (now they are exhausting case air over the rad. Temps are:
System: 39'c - 41'c
CPU0: 46'c - 49'c
CPU1: 49'c - 51'c
HDD: 31'c - 34'c
Ambient: 30'c (est.)

So things have cooled off a bit, still not good enough (need at least 15-20'c cooler) So tommorow I will go out and pick up some 120mm Panaflos and see what happens. I am also going to start construction of the chiller this weekend, hopefully I will be able to run some tests by Monday. If all goes well I will install the chiller Monday night or Tuesday and change the flow at the same time. New flow will be:

RAD -> CHILLER -> CPU1 -> CPU0 -> NB -> PUMP -> F&B -> RAD

Above temps are with the system pretty much at idle @ 133mhz FSB. I have noticed that even when I push the FSB to 150 the system still runs stable without raising vcore and the temps go up maybe 1-3'c.

As I am typing this I have been monitoring the temps and the MB temp has been rising fairly steadily. Started at 38'c now it's at 41'c in only 20 minutes. I really need to figure out how to get rid of all the hot air on the MB side of the case!!!

I really don't think another rad s necessary. My HT(max) is around 126W and the 676 should be able to handle that, but I would like some hard data. Another problem is MBM only reads the MB temp sensor on my MB. Since the sensor is located under the CPU and there is very little air flow there my guess is that a pocket of hot air is being created causing the MB to report slightly higher temps. I can install a thermister and correct for this , but right now it's not worth it. I am very sure that the CPUs are hotter than I would like them just by grabbing the rad input and output hoses I can't feel much of a temp difference between them.

[#Rotor]

those oil coolers are nice and all.... I have one here meself... BUT They are very limited in there performance at these small Dt's we are running at. You will want to get your hands on a good ol Heater-Core. The fact that the liquid's warm even when it exits the rad, tells me exactly where the problem is....

the fact thet the liquid is warm, tells me the blocks are doing there job pretty well....so nothing wrong there...But a block can't cool, with warm water...