Help the Newbie

[DoTheDog]

Hi All,
This is my first post here although I have been reading posts for a while. I have a question and would like some advice from you all. As you can see in my sig., I have a TBird 1.33Ghz @ 1.517Ghz (137x11) on an A7V133. The issue I have is that my water cooling rig is not giving me temps that I am happy with. I don't have a good thermister to check actual slug temps so I have to rely on the POS ASUS temp probe under the CPU.

Here is the H20 setup, Danner Mag5, DD Maze 2, Black Ice Rad with Sanyo 120MM fan in a closed loop. The circuit is pump->Rad->WB->pump.

My issue is that although the MB temp is 33C and the Water temp is 31C (temp probe is between pump and rad so it should be at the hottest point), my CPU temp is showing 51C in MB Monitor 5.09 and in ASUS Probe is showing 61C. This seems waaaay high. I have cranked down the nylon bolts as far as I can by hand but no diff. BTW, I also lapped the WB to a mirror finish. The only thing I can think of as causing the high temps is either the contact is not great, or flow is not high enough.

The one thing I noticed is that after I installed the Black Ice my flow is down to a trickle. But because my water temps seem ok, that doesn't seem to be right.

What should the Delta be between water temp and CPU Temp? How can I tell if I have good contact with the WB on the CPU?

Help Anyone?

[GuyBFF]

The faster the flow the lower the Delta T will be.

With a Danner Mag-Drive 5 you should be flowing much more than required. The Black Ice usually is considered to flow very well.

What's your ambiant temperature, BIOS version, Vcore measeured while in Windows and what are you running for coolant?

[DoTheDog]

Thanks for the reply GuyBFF. In answer to your questions. I used to have the same setup with a 4 inch stacked plate rad. It was running about 5C higher than with the Black Ice but it seemed like the flow was better.

Ambient is pretty warm as the computer warms up the room. But I have the case cover off so it should be pretty close to 30C, probably around 26 or so. Bios version is 1003a. vcore fluctuates between 1.87-1.89v. Coolant is distilled water with water wetter in the prescribed amount (I think 1oz per quart).

Any help is appreciated.

[DoTheDog]

Oh yeah, other info. I have 1/2 inch ID tubing, but I used a brass fitting to decrease it to 3/8 inch to go into and out of the Black Ice. The Danner Mag 5 has a 3/4 inch intake so I have another fitting going from 1/2 to 3/4. Again, it doesn't look like the water temp is the problem as it is actually cooler than the MB is. Thanks for the help.

[krp]

DoTheDog:

Is your maze2 the original or upgraded version? The original had much smaller cross-section channels in the maze and restricted flow quite a bit.

Dangerden's website states that all maze2's shipped after the end of July are the upgraded one's, how long ago did you get yours?

[DoTheDog]

It is not the "upgrade" version. It does have the 1/2 inch fittings however. I don't know if this is an issue as the flow was much better with the 4 inch stacked plate rad I had on there before.

[DoTheDog]

*Bump* I am back. I just had a the rig flake out on me. Its back and running now. No idea what the problem was, USB Mouse decided it didn't want to work. Anyway, Any Help on this heat prob would be appreciated

[redleader]

Make sure that the Maze is tight on the block. And I mean tight. I took 3C off my CPU temps just by tightening the bolts a little harder with a pair of pliers. Don't be afraid to get those springs all the way compressed (or reasonably closed to it, don't smash you chip trying to get that last degree out), your WB will work much better that way.

[DoTheDog]

Redleader,
Thanks for the reply. I do have the nylon bolts tightened down as far as I can by hand. Do you think that an extra turn with pliers will help? I don't want to crush the core (I have killed a 1Ghz AXIA with a Thermoengine and don't want a repeat )

[GuyBFF]

I can't see any real problems with what you're running. I'm running nearly the same setup, other than a homemade Big Momma like heater core and a 350 Danner.

I did however achieve 4c lower temps by recently re-making my fan shroud to allow an inch of space rather than only a 1/2" adapter, and opening up the grill in front of the rad. Maybe consider your're flow threw the rad, and if the air is flowing consistently? I heard some people getting better results by modifying / adding to the current Black Ice shroud, I think this is even now listed on Watercool-Overclock!

Also is the rad mounted to draw in fresh air, or have case air expelled threw it.

I now run about 46c max by ASUS, and 39c max by DigiDoc 5 right on the slug in a room with about 25c ambiant. This is with BIOS 1005a (which increased my idle temps nearly 6-10 degrees, but adds good features). Notice DigiDoc 5 is 7c colder than ASUS. This is partly due to ASUS compensating for there thermister mount. The DigiDoc shows temps raising and lowering probably ten times faster! Consider this, or a $5.00 ASUS thermister (used on P2B boards I think). Your board accepts one thermister on a jumper, and it reads as accurate as a DigiDoc, just can't see during a game. ASUS thermister needs to get temps with an air space to the radiated temps threw the CPU. A Thermister is mounted right on the side of the DIE. Pretty simple why its more accurate!

Also, I do as well crank the nylon nuts like RedLeader said all the way down. Be careful however to make sure to stop just before bottoming the springs. If you do bottom the springs you chance loading one side more than another, chipping your CPU or lifting an edge and killing your TBird, I know how fast this can happen.


Lastly, I get the same exact figures from MBM 5 as ASUS probe. Both are current. Maybe one has a software problem? I did notice if you try to run both at the same time readings become very unstable, is there possibly a conflict, or maybe other software interfering?


Hope one of these suggestions help, GuyBFF.

[DoTheDog]

GuyBFF,
Thanks for the reply. I tried cranking down the nylon bolts a couple of extra turns with pliers and it doesn't seem to make a difference, MBM5 still shows about 51C under load and about 48C idle.

I do have the Sanyo fan offset from the BI by 1 inch with a spacer and cardboard & duct tape shroud. It is sucking cool air in from the outside. I have cut a 4inch round hole in the front of the bezel to allow more air flow through the rad as well.

I just realized what may be causing the slowdown in the flow. I have two temp sensors in line (one right before the WB and one right before the Rad, so they are measuring the coldest and hottest water temps) and I have a water pressure sensor before the pump (which is really doing nothing because the pressure is 0 at that point). I am using the on board thermister jumper for the water temp probes. I wonder if these three 1/2 inch "T" connectors are killing my flow rate? Will increased flow rate lower my temps? I could easily remove the water pressure and 1 temp probe. The thing is that the water temp is still running a degree below the Mobo temp, and no matter what load I put on it the relative temp to the Mobo temp is always like 1C lower.

Yeah, I noticed that you cannot run both ASUS Probe and MBM at the same time too. ASUS Probe gets real flaky when you do that. I don't think I have the latest ASUS Probe version and that may be why I get a 10C difference with MBM. The MBM temps are very close to what it says in the Bios. ASUS Probe is high by 8-10C.

I guess at the end of the day, is 51C too high? I would like to shave about 5-6C off of that to get closer to what GuyBFF is getting as I am running at 100Mhz slower than he is.

At a 174Mhz overclock 1.33Ghz running at 1.507, I am not quite 100% stable. If I clock it down to 1.485Ghz it is rock solid. If I could keep the processor a little cooler...

[GuyBFF]

T Fittings in themselves do present a restriction to flow, but with your big pump it shouldn't be that bad unless theres something directly blocking the flow of water in them.

51c under load isn't too high, but if you want to overclock any further it would definetely help to get it lower (especially so there's headroom to Vmod).

Is there any air left in the rad? Try turning your computer so it can rise out (don't allow air in the T-fitting/fill tube to get back in however).

It could help to reduce your voltage. Some people have found success at lower overclocks at 1.6-1.7 volts on TBirds! Once you do go for higher mhz or if this doesn't work you will need however to V+. Try 1.6 to 1.7 in BIOS (which will likely be 1.7 - 1.8 once running). Lower voltage will definetely lower temps, sometimes this outweighs the voltage being lower, other times it just needs the juice.

[DoTheDog]

GuyBFF, Again thanks for the reply you are immensely helpful on this. I think I will try removing one of the temp probes and the pressure guage to see if that helps the flow any. I when I do it I will measure the flow rate see if it makes any difference.

I never learned how to solder so I don't think I will be doing any vmod'ing anytime soon however, I do want to be able to upgrade the CPU and hopefully get higher clock speeds. So yes, I would like some headroom in the cooling system.

I don't think there is any air in the rad as when I filled the system, I did it outside the case in a bucket, running it for 24 hours, turning the rad upside down until no more bubbles came out. But I can check it again. I will be pulling it out of the case to remove the "T" fittings anyway so I will re-bleed the system at that time.

I never thought to lower the vcore. I will give that a try when I get home.

[gmat]

Also beware of:
-sharp turns in your tubing
-pinched tubes
-total number of turns
-total water height. I made the experience with a small test closed-loop on the pump, and a 'free' 1 meter long tube. Raising the tube by hand did lower the flow down to zero at a certain height.
-Low water flow = high delta T = high core temps
-Thermal paste quantity. Specially this 'Arctic Silver' which is *very* thick. Use a razor blade to apply it or make sure there's barely a minimum. Simply pushing the WB down will *not* squeeze it at all.

Note: i have DD maze 2.1, i found it doesnt restrict flow that much. What brings the flow down is definitively water height and sharp turns.

[DoTheDog]

Surefoot,
Thanks for the reply. I do have a couple of 90 degree turns. 1 coming right out of the pump, the blackice rad connectors go straight into a plenum chamber, with 3 passes through, from there it is a straight shot to the DD Maze 2, out of the Maze 2 I have another 90 degree elbow to return it back to the pump. No sharp turns or kinks.

Water height is the pump at the bottom of the case and up to the water block (about 14 inches?) or the width of the Mobo.

I will try re-applying the ASII. I think I had too much on before but I removed most of it and spread it thinner. I may be able to get it a little thinner still. I will give it a shot after I take the water cooling system out.