Pro/Forums

Pro/Forums (http://forums.procooling.com/vbb/index.php)
-   Testing and Benchmarking (http://forums.procooling.com/vbb/forumdisplay.php?f=61)
-   -   pH results compared to mine on the Maze 4 (http://forums.procooling.com/vbb/showthread.php?t=11508)

jaydee 04-02-2005 07:33 PM

pH results compared to mine on the Maze 4
 
1 Attachment(s)
I am a little disapointed that my numbers are so far off on this first attempt. Not to sure why mine flattens out so much.

Note the numbers I used for pH are pulled off the water blocks graphs so they might be off a .1 or so.

Any way I am going to remount and start over.

pHaestus 04-02-2005 08:40 PM

It's probably because you're measuring temps under the CPU core Jaydee.

jaydee 04-02-2005 09:00 PM

Quote:

Originally Posted by pHaestus
It's probably because you're measuring temps under the CPU core Jaydee.

Probably so. I got it remounted with a little more pressure but expecting similar results. I am going to look into setting up the on chip diod... Is there anyway that can be run through the CF633?

jaydee 04-02-2005 11:55 PM

1 Attachment(s)
Here are the second round results. The Dt is a little lower because of the increased mounting pressure. As predicted the results are the same which is good as that means things are running failry consistant. I got my King Pin block on now and warming up the system

jaydee 04-07-2005 03:34 PM

1 Attachment(s)
Another with the Abit KD7 with the XP1700+ @ 1667mhz 1.75V with my KingPin block and pH's maze 4 numbers. I am still running flatter curve. Has to from the temp under the CPU.

I am replumbing the system now with the Cole Pamer YSI RTD's for water temps.

BillA 04-08-2005 09:37 AM

interesting example of the non-correlation between resolution and accuracy
lol
get the thermal mass out of your system, all of it

jaydee 04-08-2005 09:48 AM

Quote:

Originally Posted by unregistered
interesting example of the non-correlation between resolution and accuracy
lol
get the thermal mass out of your system, all of it

I am working on that right now. I am not entirly sure why the temps are so much higher. The Mobo isn't even in a case. My best guess is heat trapped under the CPU and heat from other sources adding to it. Only thing I can think of doing is putting airflow around the socket or under it. Probably give false numbers though?

BillA 04-08-2005 09:52 AM

you are looking at the compression of the temps

jaydee 04-08-2005 11:46 AM

Quote:

Originally Posted by unregistered
you are looking at the compression of the temps

I think I see what your getting at. I am not to sure how to work that out. I found this on google http://www.arcticsilver.com/diode_calibration.htm but not entirely sure that is the way to go.

jaydee 04-08-2005 02:10 PM

Following the Artic Silver article above for reference;

1667mhz @ 1.75V

CPU: 42.43C Water In: 26.94C DT: 15.49C
-----------------------

667mhz @1.75V

U: 33.50 Water In: 25.52C DT: 7.97C
-----------------------

1667/667 = 2.5

15.49/7.97 = 1.94

Shave 3C off;
CPU 39.42C Water In: 26.94C DT: 12.49C

CPU 30.50C Water In : 25.52C DT: 4.98C

12.49/4.98 = 2.5


So can I conclude I need to subtract 3C from my CPU measurments?

BillA 04-08-2005 02:20 PM

different issues
cal is one thing (you may, or not, be correct; does not matter unless you are after absolute values)
compression another

jaydee 04-08-2005 02:29 PM

1 Attachment(s)
Quote:

Originally Posted by unregistered
different issues
cal is one thing (you may, or not, be correct; does not matter unless you are after absolute values)
compression another

Revised graph -3C attached. I have searched for temperature compression but have come up with nothing. Not even sure what to search for. Is it how the reader computes the information from the thermometer to give a temperature? Continuing my search....

pHaestus 04-08-2005 02:42 PM

Take a stab at the "goodness" of your results by calculating a W for the CPU using the dT across wb. It should be mostly flat because the wattage doesnt change. It wont be perfectly flat because as you lower the C/W then more of the CPU's heat goes into the wb and less into the traces of the board. You'll also find with a "below CPU" sensor that there's another problem. There's a baseline temperature that is difficult to go below because the air near the mobo is heated up by the mosfets and other bits all around it. This tends to make good wbs all perform even closer to one another than you'd expect. It's basically the same problem as with using an insocket thermistor, it's just slightly less pronounced. I had all kinds of trouble trying to use an undercore probe with the MCWCHILL tests if you'll recall.

If you want Jaydee I'll solder you up a spare 6657 and CPU. I'm not sure exactly what chips I still have around the house as spares (maybe a 2500+ Barton and a 1200MHz Morgan Duron or something like that?)

jaydee 04-08-2005 03:06 PM

Quote:

Originally Posted by pHaestus
Take a stab at the "goodness" of your results by calculating a W for the CPU using the dT across wb. It should be mostly flat because the wattage doesnt change. It wont be perfectly flat because as you lower the C/W then more of the CPU's heat goes into the wb and less into the traces of the board. You'll also find with a "below CPU" sensor that there's another problem. There's a baseline temperature that is difficult to go below because the air near the mobo is heated up by the mosfets and other bits all around it. This tends to make good wbs all perform even closer to one another than you'd expect. It's basically the same problem as with using an insocket thermistor, it's just slightly less pronounced. I had all kinds of trouble trying to use an undercore probe with the MCWCHILL tests if you'll recall.

If you want Jaydee I'll solder you up a spare 6657 and CPU. I'm not sure exactly what chips I still have around the house as spares (maybe a 2500+ Barton and a 1200MHz Morgan Duron or something like that?)

That would be excellent. My soldering skills are not good at all on small stuff. I can send you a CPU if you want. I got 2 XP1700's, Duron Morgan 1gig and an older XP1600. I feel I am not going to get far with the probe under the CPU. I did expect that though.

jaydee 04-08-2005 03:31 PM

Quote:

Originally Posted by jaydee116
Following the Article Silver article above for reference;

1667mhz @ 1.75V

CPU: 42.43C Water In: 26.94C DT: 15.49C
-----------------------

667mhz @1.75V

U: 33.50 Water In: 25.52C DT: 7.97C
-----------------------

Using this information and the calculator at Benchtest (I know it isn't correct but it should be consistant) the C/W is off a bit. The Calculator shows 27watts at 88% load @ 667mhz and 67.49watts at 88% load @ 1667mhz 1.75V.

15.49/67.49w = .229
7.97/27w = .295

After the calibration above;

12.49/67.49w = .185
4.97/27w = .184

My math in the previous post seems to be .01C off a times for some reason though.

jaydee 04-08-2005 09:01 PM

1 Attachment(s)
Just for the hell of it I think I am going to hook the die sim up tonight like I have it. It isn't even close to being finished but I think I can get better results as it is than trying to get decent results under the CPU. Bad part is I only got about 36watts of power. Was enough to get the sim to 160C and melt through plastic though (doh, don't leave hot things unattended!). Once I finish another run with the maze 4 I will get to it.

jaydee 04-09-2005 10:01 AM

Heh, well I hooked up the die sime last night. I thought I could get 36watts but i forgot the heaters i have are 40VDC 150watt and 10.7ohm... With the supply I got it is running 13.8V. So that comes out to about 18watts of heat load.... My dT is like 6C at 1GPM..... So when Radio Shack opens up I am going to buy 2 12V (13.8V)@15a power supplies and hook them up in series to get 24V (well 13.8v x 2 actually...27.6V).

27.6V/10.7ohm = 2.58amp x 27.6V = 71.20watts. Right?

jaydee 04-09-2005 04:04 PM

I got the die sim hooked up with 71-73 watts of power. I tell you what I don't see it possible that the benchtest.com calculator is even remotely accurate (even less than I inticipated). When I fired up the power supplies the die sim temp started shooting up rapidly past any high temp I ever got even at .4gpm. Right now it is as 52.12C (dosn't count secondary losses either) and still warming up a bit with the Maze4 at 3.6GPM and water temp at 29.98C and climbing. I guess I better turn up the Delta 120 to full speed so the pump doesn't lock up. I will stop logging the temp and turn the fan up and show a graph of the rapid temp rise.

jaydee 04-09-2005 04:16 PM

I knew it was going to be hotter but I was not expecting that much. (graph software sucks)
http://www.customcooledpc.com/testbe...esimheatup.jpg

BillA 04-09-2005 04:24 PM

lol
diesel hp jd, got the real thing now

jaydee 04-09-2005 08:19 PM

1 Attachment(s)
This is lots better. The curve is actually resembling pH's. :) The higher temps are interesting though. We both are supposed to have 71ish watt heat load yet my temps with the die sim are considerably higher.

Anyway attached graph of mount one.

redleader 04-09-2005 09:05 PM

Could be that you're giving it more power then you think. How are you measuring the power dissipated? I assume you're using the warm voltage drop squared over warm resistance? What kind of meter?

jaydee 04-09-2005 09:19 PM

Quote:

Originally Posted by redleader
Could be that you're giving it more power then you think. How are you measuring the power dissipated? I assume you're using the warm voltage drop squared over warm resistance? What kind of meter?

Could be. I don't know all the math. My heater cartridge is rated at 40VDC 150watt and the ohm's is 10.7. I am using a DMM that shows 27.9V. The amps are not known as the amp setting is not working on the DMM. Theoretically the amps should be about 2.58 but maybe it is taking more power?

jaydee 04-09-2005 09:32 PM

On this second mount I am going to use more load pressure. I got my micrometer set at the spring height this time so I can reset the springs the same on each mount. I might not of had enough load pressure on that first mount. Springs were not to tight at all.

pHaestus 04-09-2005 09:47 PM

So are your milling capabilities good enough to make pH a die sim now JayDee?

Also are you planning on doing reviews fairly regularly? Someone asked me for a reference about you as a reviewer yesterday. It doesn't take long to become this site's premier wcing tester I think as long as I'm not doing anything and incoherent is quiet :)


All times are GMT -5. The time now is 01:24 AM.

Powered by vBulletin® Version 3.7.4
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.
(C) 2005 ProCooling.com
If we in some way offend you, insult you or your people, screw your mom, beat up your dad, or poop on your porch... we're sorry... we were probably really drunk...
Oh and dont steal our content bitches! Don't give us a reason to pee in your open car window this summer...