Dtek whitewater temperature problems.

[tpenguin]

I recently assembled my WC setup with a johnson pump, dtek procore with shroud, 114 cfm panaflo fan, Dtek Whitewater (poly topped), dtek NB block (poly topped), dtek bayres and 3/8" Clearflex tubing (maybe the source of my problem?). I am currently idling at 35C cpu and 28C NB in a 21C room. My CPU is a 2100+ running at 2.43GHz with 1.825vcore. I have everything hooked up properly, and I have reseated the block several times, and I currently have it attached quite tightly with AS Ceramique, but my temps are still really high. My water temperature leaving my radiator, measured by a thermal probe on the brass barb, is 24C. As this is my first attempt at water cooling, I don't know if that is normal or not. My water mixture is about 10% zerex in distilled water. I was running a 5% mixture before, but I added more to see if it helped. It didn't.

I'd just like to know what I can do to lower my temperatures. If the 3/8" tubing is the problem, the best I can possibly do is replace the tubing from my pump to my rad and to my whitewater inlet with 1/2 ID because of the tight bends necessary and limited space. But before I start taking things appart, I thought I'd see if you guys had any other suggestions for me, and I wanted to know if just replacing part of the tubing would have any affect at all. I'd appreciate any ideas.

[sn_85]

i dont really see anything wrong w/ 35C. what are your load temps. especially cuz ur 1.73ghz is doing 2.4ghz, i dont see anything wrong with ur temps.

[tpenguin]

Originally posted by sn_85
i dont really see anything wrong w/ 35C. what are your load temps. especially cuz ur 1.73ghz is doing 2.4ghz, i dont see anything wrong with ur temps.

About 39C after an hour or so of UT2K3. I don't remember my prime95 temps, but someone I know with a whitewater setup and a similar overclock has 32C max load temps, and he's only using a via aqua 1300 and a less powerful fan, so I would expect better temps.

[sn_85]

yeah but you cant really trust what others say because there are so many different variables that effect what temperatures you get. like what kind of radiator or what kind of temp probe he's using. if theres one thing i know, its that temp probes arent accurate. i use two of them. one from the in socket and one from my compunurse. and i still dont trust them because i know that there are more accurate temp probes out there. even still, i think ur temps are great because i have a 1700+ non o/c w/ a WhiteWater, chevette heatercore, and eheim 1250 and i get a/b 35C idle and 40C load. w/ a 70cfm fan.

[tpenguin]

Originally posted by sn_85
yeah but you cant really trust what others say because there are so many different variables that effect what temperatures you get. like what kind of radiator or what kind of temp probe he's using. if theres one thing i know, its that temp probes arent accurate. i use two of them. one from the in socket and one from my compunurse. and i still dont trust them because i know that there are more accurate temp probes out there. even still, i think ur temps are great because i have a 1700+ non o/c w/ a WhiteWater, chevette heatercore, and eheim 1250 and i get a/b 35C idle and 40C load. w/ a 70cfm fan.

What's your room temp, though? mine is really low.

[sn_85]

mines a/b 26C. running 1/2" tubing.

[brucoman]

adding higher percantage of anything to your water will only hurt, not improve temps

if I remember right that board read odd temps, so don't take it at face value. compare it instead to your air cooled temps with same hardware.

[tpenguin]

Originally posted by brucoman
adding higher percantage of anything to your water will only hurt, not improve temps

if I remember right that board read odd temps, so don't take it at face value. compare it instead to your air cooled temps with same hardware.

Well, according to my 8RDA3+, my CPU is about 2&degC cooler with my SLK-900U and 92mm Tornado. I can't place a probe on the back side of the CPU core because the wires get crushed under the mounting pressure.

[tpenguin]

Originally posted by tpenguin
Well, according to my 8RDA3+, my CPU is about 2&degC cooler than with my SLK-900U and 92mm Tornado. I can't place a probe on the back side of the CPU core because the wires get crushed under the mounting pressure.

[nikhsub1]

Originally posted by tpenguin
Well, according to my 8RDA3+, my CPU is about 2&degC cooler with my SLK-900U and 92mm Tornado. I can't place a probe on the back side of the CPU core because the wires get crushed under the mounting pressure.

OK, first of all the Epox's are the WORST at reporting temps, they use neither a diode or an in socket thermistor, rather one of the pins on the CPU, pure crap. Is your system stable? That is all that matters. If you got better temps aircooling it is because of the airflow generated by the fan on the HS around the socket of the CPU. You no longer have that airflow. Your temps are better than aircooling it is just that you board sucks (most boards suck too) at reporting even remotely accurate temps.

If you want to mount a probe, you mount the tip so it touches the side of the CORE and then tape it down on the CPU PCB. Then, add about 7C ontop of that reading to get a semi true core temp.

[Cathar]

If you want to "simulate" the sort of temperature reporting you got with your SLK-800, stick the SLK-800's fan on the back of the video card pointing at the CPU area and turn it on.

Then watch your reported temperature "drop" by up to 10C or so.

The temperatures that Epox boards report are heavily affected air-flow around the socket area. I had two Epox 8RDA+ boards myself and could never get anything resembling a consistent reported temperature because it was so impacted by air-flow around the socket area.

Your reported temperatures are probably somewhat accurate now, whereas before the thermistor would have been cooled heavily by the SLK-800's fan, because quite a fair bit of air does get under the CPU socket area.

[tpenguin]

Originally posted by Cathar
If you want to "simulate" the sort of temperature reporting you got with your SLK-800, stick the SLK-800's fan on the back of the video card pointing at the CPU area and turn it on.

Then watch your reported temperature "drop" by up to 10C or so.

The temperatures that Epox boards report are heavily affected air-flow around the socket area. I had two Epox 8RDA+ boards myself and could never get anything resembling a consistent reported temperature because it was so impacted by air-flow around the socket area.

Your reported temperatures are probably somewhat accurate now, whereas before the thermistor would have been cooled heavily by the SLK-800's fan, because quite a fair bit of air does get under the CPU socket area.

The 8RDA3+ has a compunurse-style thermistor/diode (I'm not sure which) sticking up out of the center of the socket, and that it touching the underside of the pcb directly beneat the core, so I'm fairly certain it is pretty accurate. So is a 15&degC rise above ambient at idle and an 18&degC rise above ambient at load normal for a whitewater connected to a heater core with this type of overclock? These are the highest temperatures I've seen with a room temperature near mine (average of ~ 20&degC). I just want to know if my whitewater is performing as well as it possibly can. Also, Cathar, could you tell me if there is a significant performance difference with the whitewater when 1/2" ID tubing is used instead of 3/8"? If there is a big difference, I'm just going to replace all my tubing with 1/2" and try to make it fit into my case.

[winewood]

A.) 1st of all your temperatures are great.
B.) 2nd of all, if it ain't broke don't fix it.
C.) You have a system to be proud of.

I see nothing that needs fixing. The only reason I would change to 1/2 is if you are attempting to overclock and you are unable to reach the desired OC. The 1/2 tubing will help waterflow, but it may not be worth it.. your call.

[Cathar]

You ask for advice, and then you dismiss it. Why, then, did you ask?

I have owned two 8RDA+'s. The thermal sensor is the same as you describe and it is crap, and it is affected by airflow, just as I said. Try the fan on the back of the video card like I suggested and tell us what you see.

I can't answer: "Is a 15C rise normal?", because the thermal sensors on all motherboards are basically trash.

If you want to know if the block is performing well, try to overclock more. You should be able to.

There won't be a significant difference between 3/8" and 1/2" ID tubing. The 1/2" ID tubing will allow for higher flow rates, and the flow rate difference may mean 0.5C better cooling.

[tpenguin]

Originally posted by Cathar
You ask for advice, and then you dismiss it. Why, then, did you ask?

I have owned two 8RDA+'s. The thermal sensor is the same as you describe and it is crap, and it is affected by airflow, just as I said. Try the fan on the back of the video card like I suggested and tell us what you see.

I can't answer: "Is a 15C rise normal?", because the thermal sensors on all motherboards are basically trash.

If you want to know if the block is performing well, try to overclock more. You should be able to.

There won't be a significant difference between 3/8" and 1/2" ID tubing. The 1/2" ID tubing will allow for higher flow rates, and the flow rate difference may mean 0.5C better cooling.

I did not dismiss your advice, and I didn't know what type of thermal sensors 8RDA+'s used. My statement about the thermal sensor was mainly in response to the statement nikhsub1 made about Epox boards measuring the temperatures through the pins, which is not true in my case. I tried the fan suggestion, and I got a 1&degC temperature drop. I probably would have gotten a larger drop if I didn't have all the tubing in the way. I believed your statement about the motherboard reported temperature to begin with, and I was simply asking another question, not attempting to argue with you. The primary reason I posted this question here is because you post in these forums, and I was hoping you would be able to tell me if the tubing I used was an issue. Thank you for your help, and I'm sorry for the misunderstanding.

[pakman]

8rda+ temp readings are very suspect as cathar mentions. Recently I have added a 50mm fan to cool my mosfets (which are just left of the cpu socket) and are getting bloody hot (60+c). I have noticed a slight decrease in cpu temp readings since installing that fan. My guess is that 8rda+ have been calibrated to take into consideraton the cpu fan on a standard HSF. of course the cooler mosfets mean cooler surface temps around the cpu socket...

As cathar mentioned, try that little test by taking a fan and have it blowing on the cpu socket area and see how your temps change.

[ItsSoLARGE]

Try and run a program like Speedfan and check if you have a ondie diode reading (like ph does on his a7n8x-x)

[nikhsub1]

Originally posted by ItsSoLARGE
Try and run a program like Speedfan and check if you have a ondie diode reading (like ph does on his a7n8x-x)

NO epox does not use them anymore, not since the 8K3A because of too many compaining of high temps.

[pakman]

he's using a 8rda3+ which doesn't report cpu in-diode temps. it uses a in socket thermistor.

[nikhsub1]

Originally posted by pakman
he's using a 8rda3+ which doesn't report cpu in-diode temps. it uses a in socket thermistor.

I dont even think it uses an insocket thermistor! That was my point. I know for a fact the 8RDA/+ measured CPU temps via a pin on the CPU, the worst way possible. I'm pretty sure the 8RDA3+ uses the same method.

[pakman]

Originally posted by nikhsub1
I dont even think it uses an insocket thermistor! That was my point. I know for a fact the 8RDA/+ measured CPU temps via a pin on the CPU, the worst way possible. I'm pretty sure the 8RDA3+ uses the same method.

pls carefully look at that photo of the 8rda3+ above. There is a THERMISTOR sticking out from the socket which would touch the backside of the cpu.

[nikhsub1]

Originally posted by pakman
pls carefully look at that photo of the 8rda3+ above. There is a THERMISTOR sticking out from the socket which would touch the backside of the cpu.

You could be right... I know that the 8RDA had something similar, but, it was not an insocket thermistor.

[pHaestus]

nikhsub:

Pin from CPU is worst possible method? What is a diode? How would one access something like a diode on a CPU?

Hint: Check the tech specs for the processors, and you'll see the diode's positive and negative leads running out through 2 of the CPUs pins.

Do check speedfan bacause it will show you if two readings are on the same channel of the Winbond IC. That is a pretty good indicator of a diode reading (check the appropriate winbond spec sheet and you'll see that diode and thermistor share same channel on their designs...

[pakman]

Originally posted by nikhsub1
You could be right... I know that the 8RDA had something similar, but, it was not an insocket thermistor.

what is that in socket? I circled it in red just in case you keep missing it.

[pakman]

Originally posted by pHaestus
Do check speedfan bacause it will show you if two readings are on the same channel of the Winbond IC. That is a pretty good indicator of a diode reading (check the appropriate winbond spec sheet and you'll see that diode and thermistor share same channel on their designs...

phaestus, although the 8rda+ uses athlon xp in diode temp readings for its emergency shutdown, unfortunately it doesn't allow mbm to get readings from it unlike some other brand nforce2 boards.