Water temp high.

[CheeseBall]

The readings I'm listing are from this winter. Just now posting my question. Here's the info:

The waterblock I machined myself
1800+ DLT3C clocked at 208x11.5=2392mhz (1.8vcore)
Chevette heatercore and a 172mm (rated for 141CFMs; 36dB). The fan is shrouded away from the heatercore about 1.6" (fan is pulling air through).
Eheim 1250
Danderden waterblock on my 9800 Pro (445/370)
The air entering the heatercore is about 23.05C (73.5F)
Air exiting is 28.61C (83.5F)
I have 5/8" barbs on my CPU block and heatercore to not restrict flow (1/2" tubing yes)
I use a T fitting to fill/bleed the system
Pump-->RAD-->T-fitting-->CPU block-->Y-fitting-->GPU block-->Pump
Water temp that was was 32.2C (90.5F).

My NF7-S rev2 reports CPU temps as 49C running prime95. But I have a thermal probe tapped right next to the die of the CPU, w/ Ceramique between the die and probe. That probe reports temps to be around 39C.

Now my main concern is the water temp. The temp was taken from the bleed line, which is after the rad. The chevette heatercore can handle more watts than what I'm throwing at it right? From doing the benchtest.com calculations, the CPU is using 115.2W and C/W is about .23. Then dunno what the C/W would be for the GPU. I would like to figure out if I need a larger heatercore (might do vmod on the 9800 Pro). Or if not, why my water temps are high. Also, I'm wondering which temp reading you think is more accurate.

Thanks for any help.

[AngryAlpaca]

10F difference between the air going in and out?! I don't think that's physically possible with 32C water and 141 actual CFM. It seems like airflow might be the problem. Is the radiator dirty or clogged with anything? Is the air intake restrictive?

[CheeseBall]

You mean 10F being too small of a difference? I'm sure the actual CFM is closer to 120, and I don't THINK the rad was very dirty at the time of testing. But it could have been. w/ that much air going through, I wouldn't think 10C would be too small, but I wouldn't know. What's the norm?

I dunno if this info would be helpful or not. W/ my AMD 1.4ghz tbird at 1600mhz w/ 2.0vcore my temps were around 50C w/ the same mobo. That was WITHOUT the GPU block.

[Groth]

Based on the specific heat of air, your air temperatures, and some guesses for CPU and GPU heat output, I'd put your actual airflow closer to the 70 CFM range.

More air would help.

[CheeseBall]

That is less than half of the rated CFM, and this fan has very good staticpressure as I recall. One thing I was thinking about is that the heatercore isn't working as well as it could. I JB welded the 5/8" barbs on it. And when I did it some JB weld fell down into the heatercore. However, I had the heatercore on it's back, and from what I could see it was just dropped on the the back of the tank, and didn't get in the way of any "channels." Maybe I'm wrong.

Here's the link to my 172mm fan specs! (4.2 m3/min = 148.323 ft3/min).

Also, I have the 5W rheostat from allelectronics.com, but I run it on full blast for the most part. Could the rheostat be dropping the voltage to the fan a little even though it's turned up all the way? Or do you think it's just a dust thing. I know right now there is some dust in there, but I thought I had cleaned it out resently when I took the readings. Also, I have the Antec (Chenming) Dragon case, so the front panel isn't resticting air flow much. And I did dremel out a complete opening for the heatercore in the sheet metal.

Maybe I will clean out the heatercore's fins and retake some readings. Good idea? Thanks for the help.

[Groth]

Looking at the static pressure vs. flow for your fan and at some of Bill's old results at Overclockers, the airflow seem about right.

If read the whole article and compare your rad, you can get a sense of how well yours is performing. For the airflow I'd say it's decent.

I doubt that your rheostat is a problem. Cleaning the rad inside and out might improve the air-water temperature difference a bit, but more air will definitely help.

[CheeseBall]

There's not many fans that are rated for more than 148CFMs. There are some 12V Sanyo Denkis that are rated for 300CFMs, but I believe they are hard to find, and I don't want to live w/ 50+dB. I could get a 24v cormair rotron patriate. At 12v I think they are spec out at 160CFM. I was thinking about putting my current fan as a blow hole and getting a patriate for the rad anyways. But that isn't going to add too many CFMs.

And when you say "the airflow seem about right." You're talking about the 70CFMs you came up w/? Or what?

I'll check out the article. I think I started reading it in my newbie days, and it was over my head. I think another factor is, my waterblock is a flow killer. Those channels are 3/8" tall. SHoulda made them shorter.

Couple more things. You don't think a bigger rad is worth the trouble? And which temp probe do you think is more accurate?

[Groth]

Yeah, 70 CFM seems right as a rough estimate - that's the neighborhood where the static pressure vs airflow curves for the fan and rad would intersect. If you look at the graph for your fan, notice the spot where airflow drops dramatically with only a slight increase in backpressure. That's where you're operating. A push-pull arrangement could make a big difference. Look for fans with high static pressure (closely packed fins on heater cores need higher pressure), focus less on their CFM rating.

Both probes are equally unreliable - subject to varying placement, contact pressure, changes in local airflow. On some forum, somewhere (can't recall), I saw a mod to allow the NF7 to read the thermal diode. That, or a variation based on the readers MMZ_Timelord is making, would a better temperature indication.

I'd try more air before changing the rad.

[CheeseBall]

Here's the link on how to read the NF7-S's thermal diode probe. Thanks for mentioning it! I think I'll give it a shot.

[CheeseBall]

Ok, last question, maybe. You say more air flow will help, but I want to know as it stands now, how's my water temp. So which of the following is true:

My water temp is about average considering the heat of my components.

My water temp is higher than average because I have below average air flow (w/ a 172mm 141cfm fan?).

Or something else is the cause of high water temps.

[kronchev]

my vote:


screw the water temp, as long as the processor is cool and stable

[Groth]

Originally Posted by CheeseBall

Ok, last question, maybe. You say more air flow will help, but I want to know as it stands now, how's my water temp. So which of the following is true:

My water temp is about average considering the heat of my components.

My water temp is higher than average because I have below average air flow (w/ a 172mm 141cfm fan?).

Or something else is the cause of high water temps.

Your airflow is limiting how low your water temperatures can go. If it takes a 10 °F rise in air temps through the rad to carry the heat away, then your water will be at absolute best, 10 °F over ambient.

Your rad doesn't seem to be performing well. A difference of 7 °F between the water and the exiting air seems excessive. I wouldn't expect that to be more that a couple of °F -- the densely packed fins and flat tubes of heatcore are darned efficient heat exchangers. Dirty fins? A blocked tube or three? I don't know.

Don't screw the water temp'. Not only is it unhygenic, but water temps are a great indication of where to put time and money. For me, if the water is hotter than 3 °C over ambient, something's gotta change.