44C WC on a PIII?!

[DeafinONEear]

So I just finished my first wc setup about 20 minutes ago and I'm having an issue. I was *really* happy with the way it looked, glad as hell it was over. It took me three days to finish-- all the cooling is remote in another room with the radiator and res in there with lines going through the drywall and I/O lines in the back of the comp, not to mention a switch on the front of the case for the remote setup and hard drive blocks that I mounted to the rails and had to cut into the case (don't ask me about that part-- I'll put it up in the page I'm doing on this.)
So anyway, I was proud of my work.
So I go to post and check my teps after idling for about 10 minutes, and, holy shit, it's at 53C!!!! Mind you this is on a PIII 700 @ 933 w/ 1.65 vcore. So it shouldn't be that high. I take off the block and take out the shim, put everything back and now my temperatures are down, BUT ONLY TO 44c!!!
WTF?
So, was I just expecting too much? am I doing something wrong? what gives?

[ECUPirate]

What cooling / WB are you using?
What did you use to take the temps?

44C sounds very high. Is the block on straight? (even) Is it flat on the bottom? I'm assuming you used artic silver or equivalent...

[DeafinONEear]

Waterblock= DD Maze (1)
DD chipset block
DD GeForce block
2 homemade HD blocks
all blocks are 3/8" fittings.
Radiator= 87 Escort w/ a 120mm fan on it
2 Gal res


circut goes like:

res-> (1/2" line) radiator-> computer-> Split to dual 3/8 lines
line 1= HD block 1-> HD block 2-> GF block
line 2= CPU-> chipset
then they both get spliced together into a 1/2" and go to the res (but first pass through a flow valve, which I have yet to screw with.)

pump is a 250 Mag Drive.

no kinks in the lines-- 1/4" thickness

[WebMasta33]

I would say it's because your lines are thin, and because you have alot of load in that system....

[ECUPirate]

well, there's your answer.


1st issue, you're pushing water through five block, a rad, and a res w/ one, not so great, pump.

2nd, and most important, your lines split. For all you know 90% of the flow could be going to the HDs rather than the CPU. If your homemade HD blocks have low resistance..... well, fluids tend to follow the path of least resistance. Use two pumps, or run the whole thing in series, starting with your CPU.
If there is a flow valve on the HD line, try closing it most of the way.. (don't burn up your vid card).

problem solved.

[ECUPirate]

p.s.... it's only OK to run rads in parallel. In fact, it's better to run rads in parallel, rather than series. (the higher the temp diff. b/n the water and ambient, the better they perform. they also like low flow rates. waterblocks like high flow rates, which is why you run those in series.)

[DeafinONEear]

but wait-- I don't have two radiators. so what's this about running radiators in parallel?
True, the blocks I made don't have much resistance, but the path that they have to go through involves much more lift than the CPU one. In fact, the CPU block is right at the level that the inlet is at, whereas the line to the HD block has to climb firs, then dip, then climb.
And to tell you the truth, I'm getting quite excellent out put flow, so I don't think the problem is not enough flow.....
hmmm...

[ECUPirate]

the bit about parallel rads was just FYI.

How do you know what your flow is? Where was it measured? Did you measure the aggregate output from the joint 1/2" line, or did you compare simultaneously the output of the two lines before they joined at the Y? I'm guessing that you really don't know how much water is flowing through your waterblock, relative to the HDs. The added resistance from the vertical lift could easily be negated by flow constriction in the WBs on the other line. I repeat, for WBs, run the lines in series, or use two pumps.

[DeafinONEear]

so there are two frongs sitting on a log-- Pete and Repete. Pete fell off, who was left?

[ECUPirate]

ha. is pete OK? hmmm.

...as long as we're all clear...

I hope I helped

[Kevin]

deafinoneear - Flow could be your issue. If so, try just pinching the line going to your HDs/Vid card w/ your finger or a clamp for a while and watch if the CPU temp drops (give it about 30 mins to even out). If so, then flow is probably your problem. Another thing may just be the fact that you have a lot of components dumping heat into the water. 2 HDs, a northbridge, a CPU, and a video card. You must admit. That IS a lot. So you may not have a powerful enough heatercore/fan. Just something to think about all though your temp does seem higher than usual even with all those components.
-Kevin

[DeafinONEear]

yeah, I think I might try that pinching technique.
ECU-- the thing about running everything in series is that I really don't want the HD's to heat up-- they are both inside silent drive enclosures and the rails act as sinks. to tell you the truth, i don' think that those engineers thought one bit about thermal transfer because the insides of those things get phukin hot! so I need those things to get cooled as much as the CPU.
any ideas?

[ECUPirate]

yeah, you need two complete circuits. Unless you are running a great big pump w/ tons of flow, and you are certain that each path is receiving adequate flow, you want to avoid running you blocks in parallel. Also, as Kevin mentioned, you are dumping lots of heat into your system, and the Escort core isn't all that big. I've got the Ford Econoline van core. It's about twice the size of yours... it holds 2 120mm fans.

Here's what I'd do. ...

get a bigger, better pump...at least an Eheim 1048, maybe a 1250. mine's a 1250, and right now I've only got the CPU maze 3 w/ 1/2" hose. Use the new pump first feed the three chipset blocks, then to the rad, then to the res, and back to the new pump. Run the old pump from the pump to the HDs and back to the res. The problem here is that the HDs will heat up the res, and that water will have to flow through the CPU before it reaches the rad. You could position the rad between the new pump and the chipset blocks, but then those blocks would heat the res, thus heating your HDs. You could add a second rad to the HD loop, and have the two rads dump into the res. From the res to each pump. (Res should have two in's and two outs, or one in and two outs, depending on 1 or 2 rads.)

hmmm..?

[DeafinONEear]

My god you have a great imagination!
See, this is a great solution, but there are two things holding me back-- one, I just finished this project. Like, *just* finished. I got done sending the lines through the drywall, spakling up the holes, painting over it and getting the res/rad all situated in a shockmount enclosure. So sending yet another line in would add a whole lot of work to this that I am just not ready to do.
second, this project was for silence. I record bands and do folley and, due to space considerations (ie I live in San Francisco and *there is no space*) I record all the sounds in this room which acts as both a control room, a live room and a mixing room. The computer is housed in here. Because of this I need to keep as much computer noise out of the room as possible and right now I can just barely hear the sound of that one fan. Now think of two fans and another pump. You think your hearing is good? A Nueman U87 condenser giong through a NEVE pre will pick up everything you hear times two.
So because the main goal of this one is not overclocking, though I hoped I could push it a little more with this setup, my quest for silence outweighs everything else and I am forced to live with 44C, which is still an acceptable temp.
Which brings me to this-- I knew what ASUS PC probe CPU reading were always a little off, usually about 7-10 degrees high, so I primarily use the BIOS reading to check my CPU idle temp, whcih seems to be 44C. But then I just recently downloaded the newest version and I think there might have been something in there about temp accuarcy. So now when I read from the new PC Probe it shows 32C, which is less than what the BIOS says....
what should I believe?

EDIT: "Live" room, not love room!

[Kevin]

Ya know... the P3s have the internal temp diode in 'em right? So the bIOS is probably right. BUT hey here's the main issue. Is it stable? If so, you're fine......

[ECUPirate]

which prog? I dunno. I use speedfan v4.02 ?? and it agrees w/ my bios. If you are comparing before and after temps, stick w/ the same source. Be consistent.
Noise is why I switched also. Use 120mm panaflo L1A fans. Get a rheostat... a 5W, 100 ohm should work nicely, or a 7/off/12v switch. At lower voltages, those fans will be completely inaudible, even in the same room. My eheim pump is almost completely silent, and it sits next to me. Any sound you hear will be a vibration it causes. In a separate enclosure in a different room, you should be able to dampen it adequately.
I'm assuming that the Y in your hose happens once the hose reaches the case. For now, run the hose in series from the CPU, Geforce, Nbridge, HDs, Rad, Res, Pump. Then, you may want to simply get a bigger radiator. Use a fan shroud to distance the fan from the rad... at least one inch. This will improve performance and reduce noise. A cardboard box works nicely... use your imagination. This doesn't have to be painful.

Well, it's almost midnight in SF, so that means it's almost 3AM here... goodnight... I'll check back tomorrow AM.

cheers

[ECUPirate]

RE kevin's post..

yeah, but I'd hate to risk it... the CPU line could constrict, or kink somehow, and the CPU would be toast, because the flow would divert.

[Kevin]

yeah but that doesn't have anything to do w/ CPU temp

[DeafinONEear]

yeah, I might try that. Actually I did raise the fan about 2 inches using sponge strips and then making a shroud of duct tape. Pretty ghettz, but it seems to work and I can feel air being sucked in from all points on the front of the radiator. rheostats don't reach into the next room, so I'm sticking with this. And what do you mean about kinking the line and what not? I don't see where that came into the conversation.

[Brad]

my P3 1100's give me 28C and 29C when at 1423mhz 2.05v, which is about 6-7C over ambient

so your temps are really quite poor. That whole flow going into the hdd blocks idea is probably most of it

[Kevin]

his temps are not poor considering the circumstances (poor flow rate, 5 items to be cooled, etc)
-Kev

[DeafinONEear]

ok, so I downloaded speed fan and the temps read as follows:

temp: -32C
temp1: 30C
temp: 11C

hmmm....

anyone know what the hell "-32C" means? this sure as hell aint no pelt setup!
and I don't know what 11C is for... board should be about 30.

Also! I checked the SMART monitoring and for both HDD's (one IBM 75GXP 45G, the other a 75 GXP 30G) and they BOTH showed values that corresponded to the "worst" value in the next colum! I know for a fact that these drives are not failing, so I for anyone that uses speedfan, could you please tell me about this?

EDIT: Ok, this is actually kind of funny-- I went into Sandra '01 and checked CPU temps and it gave me *10.0C*
HAHAHAHAHA!!!! Yeah right! What a cruel world-- I just don't know who to trust any more!

[Brad]

your temp probes are just all configured wrong

change which temp probes are being used

[EMC2]

You might want to try Motherboard Monitor - it has always read the probes correctly for me when setup properly. Another alternative that also has the SMART drive monitoring is CPUCool.

And a bigger pump would definitely help matters out with that setup. If it isn't already, you could put the pump out with the res/rad to eliminate the background noise concerns and not have to change your "thru-wall" plumbing at all

[ECUPirate]

Originally posted by Kevin
yeah but that doesn't have anything to do w/ CPU temp

yes, it does. I'm not sure we're on the same page....

He has a Y in his line. The water could go either way, or SOME combination of the two. The more water that goes to the HDs, the less that can go to the CPU... thus, the higher the CPU temps.

And what do you mean about kinking the line and what not? I don't see where that came into the conversation.

I brought it into the conversation. I can do that. I'm just saying it could happen, depending on how radically the pipes bend w/in your PC. If the water can easily divert to the open pipe, there will be nothing to hold the CPU line open... Normally, if you close off a line in the circuit, the pump will strain and make noise. Maybe you can correct the problem in time. With your setup, you'd never notice a problem. (absent an accurate temp guage)
whatever.. that's just the worst case scenario. Bottom line is this: Your temps are quite shittie, period. The fact that your system isn't that bad 'considering' is irrelevant.... kind of like saying.. "hey, my truck engine is only 500 degrees hotter than it should be. Not bad considering I'm cooling it with a bag of ice on the engine block." If you want appropriate temps, get the right equipment.