Current review project (MCWCHILL) - Page 5

KnightElite · 10-21-2003, 11:39 AM

Quote:

Originally posted by Joe
So the info I need to know:

is it UV "reflective"? And how does it work the CCFLs? Also those are some nice abstract pictures and drawings in the previous posts, very post modern...

Oh and if its going to win an editors choice award, could you just remove all the text and just put pictures up with teh first one being "WE CAN JUST TELL ITS FREAKING AMAZING! - EDITORS AWARD FOR BEST COOLING BASED ON VISUAL OBSERVATIONS"

hehe

And here's Joe, summing up the classic enthusiast's view....

.

Yo-DUH_87 · 10-21-2003, 11:43 AM

Quote:

Originally posted by Joe
So the info I need to know:

is it UV "reflective"? And how does it work the CCFLs? Also those are some nice abstract pictures and drawings in the previous posts, very post modern...

Oh and if its going to win an editors choice award, could you just remove all the text and just put pictures up with teh first one being "WE CAN JUST TELL ITS FREAKING AMAZING! - EDITORS AWARD FOR BEST COOLING BASED ON VISUAL OBSERVATIONS"

Any BL00 LEDS? CAN'T HAVE ANY GREEN YELLOW OR RED LEDS, THEY HAVE TO BE BL00!!!

Nice results pH, it is good to see someone actualy taking up the task to provide "real" results instead of someone pasting a "Editors choice" on just any product.

pHaestus · 10-21-2003, 11:53 AM

Quote:

Originally posted by Les
Also most interested in variation with MHz : From June 3,2002
"Let me get my 6655 set up with 0.125C res and I will see what I can come up with regarding data. That may not help you directly though Les" http://forums.overclockers.com.au/sh...5&pagenumber=3 . The scaling with Mhz and my inadequate Laboratory Craft was what stopped me in my tracks.

Looks like we are back full circle with that again eh Les? Would the same MHz as last night repeated at a lower voltage help? That shouldn't be too much trouble to collect.

Les · 10-21-2003, 02:35 PM

Quote:

Originally posted by pHaestus
Looks like we are back full circle with that again eh Les? Would the same MHz as last night repeated at a lower voltage help? That shouldn't be too much trouble to collect.

Yes,it is difficult to get away from the complexity of even the Chiller-less system.
I would not be so bold as to say varying MHz at a lower V-core would help, but certainly it would be most interesting.
Possibly as good a variable to consider until some progress with flow-rate becomes feasable.

Aardil · 10-21-2003, 04:34 PM

Personally I dont put much stock in Editors choice awards,
Maybe we should have our own award :
The I Survived the "Procooling Test of Endurance" Award
Only offered to Products that actually pass the test of being usefull and do what they promise.
Would make for quite a mile stone, I came close with my PWM controller, just needed a few more outputs and some LCD display, oh and bring the cost down to free.

Dang I almost had one

pH,
You have done well from what I can understand, let no one say it useless info. Its only Useless to the people concerned with looks and not how it performs.
Aardil

pHaestus · 10-21-2003, 06:13 PM

haha I was just making a smartass comment. No need for all the praise. All I do is the grunt work on this; the interpretation is clearly a group effort here. And I would rather do an article that I WANT to do than what is popular or good marketing any day. That's why these technical pieces are pretty few and far between on Procooling (takes so much time).

Heading home now. I'll fire up the rig after dinner and report in at midnight as usual

jaydee · 10-21-2003, 06:33 PM

Quote:

Originally posted by pHaestus
So you'd like to see several MHz at same voltage then?

Wondering what use these numbers could be for? Seems to me any results would only apply to that particular CPU?

What I would like to see is these results done on the current mobo and then re-run on a different mobo and see what the differences would be. I suspect each mobo has a unique power curve which might greatly effect any CPU wattage guesstimator program/chart (along with many other reasons of course).

Not suggesting you do this pH, but it would be interesting (to me at least).

bigben2k · 10-21-2003, 07:22 PM

Quote:

Originally posted by Aardil
...
Maybe we should have our own award :
The I Survived the "Procooling Test of Endurance" Award
...

Yeah, one of those...

Since87 · 10-21-2003, 08:57 PM

Quote:

Originally posted by Les
The step if it occurs is at a Coolant temp of -4C(ish) and a CPU temp of 9(ish).

Yes, thanks for pointing that out. I'm totally swamped with work right now, and am not able to devote the time to keep up very well with this very interesting discussion.

Quote:

Originally posted by Les

That appears to indicate there is little dropoff in leakage current near 9C when the CPU is running at 2V. (A significant change in leakage would result in a more curved line.) Perhaps similar testing at a voltage closer the Vcore spec would be more revealing.

I was going to suggest more testing along these lines, but using temperatures to guesstimate current draw is just too backasswards. Maybe it's worth using the THG method of measuring current. Even 10% accuracy current measurements would be better than guesses based on temperature.

Quote:

Originally posted by Les
Harvard Thermal TAS (Thermal Analysis System) which I have, sells an add-on unit PC Trace - which I do not have
this permits a very accurate thermal modeling of the board in terms of heat inputs and dissipation
Since87 – it models with resistors so can also predict boardwide voltages, interesting stuff
I do not doubt that mobo mfgrs use this (in fact I know of several)

this 'variable' is why even a grothmeter will still fail to accurately characterize the heat load actually imparted to the cooling solution

Yes, a motherboard is such a fugly environment to try to determine anything with accuracy.

Pulling the Vcore voltage regulator off altogether and running a bench supply to the CPU would improve things. It gets a major heatsource off the board and makes measurement of CPU power consumption easier. That would be a pretty serious hack job though, and would require some research into the power supply start up requirements of the CPU/motherboard.

Then you'd want to provide the North Bridge with a TEC setup to maintain zero dT between the NB substrate and the mobo right 'below' it. (Ensuring 'all' NB heat flows through the TEC.)

Then thermal epoxy the back of the motherboard (in the region around the CPU) to a slab of copper.

Then...

Since87 · 10-21-2003, 09:18 PM

Quote:

Originally posted by jaydee116
Wondering what use these numbers could be for? Seems to me any results would only apply to that particular CPU?

Yes, the particular results would only be applicable to the particular CPU, but if any CPU showed a dropoff in leakage current at a certain operating point, it would be likely that all of the CPU's in that family would have a similar dropoff at some point.

Such a dropoff might indicate a point where a substantial gain in overclock could be achieved.

I don't know that many people would be able to make practical use of the information. (Of course anyone posting in this thread likely lost sight of what "practical" means, long ago.) Mostly, I'm just curious about this stuff.

pHaestus · 10-21-2003, 09:25 PM

Yup since87 I also remain curious. Almost all of my technical articles are fueled by interest in something similar. I disguise them as "cooling related" sometimes though

pHaestus · 10-22-2003, 12:14 AM

Les: This should look familiar to you:

It's good to see data that's reproducible right? Now if we could only understand why that behavior occurs...

And with today's data superimposed on the same old C/W vs coolant temp graph:

I think this picture would be a good bit different with a clearer estimate of true W. I am also starting to wonder if maybe noninsulated wb and imperfectly insulated res and hose become a factor at low coolant temps and cause higher apparent C/W.

Almost forgot; numbers for Les here:
http://phaestus.procooling.com/mcwchill/temps3.jpg

Les · 10-22-2003, 02:25 AM

Quote:

Originally posted by pHaestus

It's good to see data that's reproducible right? Now if we could only understand why that behavior occurs...
..
I think this picture would be a good bit different with a clearer estimate of true W. I am also starting to wonder if maybe noninsulated wb and imperfectly insulated res and hose become a factor at low coolant temps and cause higher apparent C/W.

Almost forgot; numbers for Les here:

Morn.
First coffee thoughts:
Is the V-Core =2.0 New data? - I not see in day's Numerical Display.
Yes, I do see some admirable reproducability in Diode Temp Readings, e.g MHz 2200, V-Core 1.6(Coolant Temp -5.7 and -6.07)
Is looking like a "..... Watts are worse" issue.
There does seem to be a transition Temp whichever Wattage "guess method" is used .
Perhaps even two transitions if including Bills data(as viewed in the larger 0.19 -0.26 "C/W" scale)
The obvious culprit would be a phase changes* in the insulation.
Will post later in the Morn with madatory Excels .
Ta for all effort.

Edit
* Both Condensation and Ice-formation in different sections of the system

Les · 10-22-2003, 06:12 AM

Shows no sign of anomaly suggested in my Morgan's DeltaT v MHz. In fact my results show a similar general trend.
For now will put my results down to poor experimental technique but not completely discount that enigma could be associated with change in FSB.
The inclusion of Mondays (220MHz, 1.81V) does qrestion the use of V and MHz data for guessing Wattage. Not sure how this data point stands up in "Coolant Temp based"guesses at Wattage .

Hopefully will return to main thrust(Chiller performance) in next post(still sorting out data):-
A comparison of Heat-source Wattage guesses, including the Flow-rate dependant DeltaT(chillin-chillout) guesses.
With good Cold-side Insulation this would =DeltaT(cpuout-cpuin) and as such is potentially the definitve number.
God pHaestus, I think we have entered Tester's Torture Chamber.

Les · 10-22-2003, 09:51 AM

After much pissing about.
The most consistent calculations do appear to be the "Billa Delta Hot-Cold" based that you are using.Whether they are meaningful is perhaps another matter.

BillA · 10-22-2003, 10:07 AM

Quote:

Originally posted by Les
[B. . . .
God pHaestus, I think we have entered Tester's Torture Chamber. [/b]

yes Les, an apt description

pHaestus, I have a spare calibrated dp xmtr I could loan you
it weighs ~20lbs but is pretty hard to break by shipping
you would need a psu and a good dmm with scan rate selection to truncate surplus digits (you want a 20ma scale to 3 decimal places)

send ?

pHaestus · 10-22-2003, 10:53 PM

Revisited the flow rate issue again today:

Looks like W are virtually unchanged (59.9-60 ain't bad) as would be expected considering voltage and MHz are unchanged. Performance indeed appears slightly worse when the flow is throttled. I am going to get Joe's Swissflow flowmeter this weekend and I will plumb it in and return to this issue on Monday.

Les · 10-22-2003, 11:52 PM

Quote:

Originally posted by pHaestus
....
Looks like W are virtually unchanged (59.9-60 ain't bad) ...

Corresponding to Flow rates of 0.6194 lpm and 0.5568 lpm *

* Using Heat Extracted by Chiller(Watts) = 69.5466 x flow-rate(lpm) x DeltaTchiller.

Will edit to include previous Flow-test result.

EDIT
Made Dog's Dinner of this.
This a complete REPOST of Edit.
Were 3 Flow Tests.

Comparison appears tobe the usual Ugh(1/2 Ugh anyway).This is notwithstanding the data mainipulation difficulties** and the effect of Fan blowing on socket in 1st Test.
Possibly illustrates the delicacy of the system

** pHaestus gives no W values for previous flow-test. My values of W differ slightly from pHaestus's using Delta(hot-chillH2O) v "Billa;sWatts" technique - .

Difference is probably no consequence but is bloody irksome .
Used my W values of 58.92 and 59..05 for 3rd Test
76.41 and 78.75 for 2nd Test
79.95 and 79.9 for 1st Test

pHaestus · 10-23-2003, 09:08 AM

Les:

I will have to look at my spreadsheet again as my number for W SHOULD agree with Bill's Swiftech data as they come from the same eqn. Perhaps an error in my goal seek?

The first set of full flow vs. half flow data was collected at 2200MHz and 2.0V whereas this is at 1.81V. There could be flow rate effects from this (higher W meaning higher flow rates seems reasonable)

Les · 10-23-2003, 09:34 AM

Quote:

Originally posted by pHaestus
Les:

I will have to look at my spreadsheet again as my number for W SHOULD agree with Bill's Swiftech data as they come from the same eqn. Perhaps an error in my goal seek?

The first set of full flow vs. half flow data was collected at 2200MHz and 2.0V whereas this is at 1.81V. There could be flow rate effects from this (higher W meaning higher flow rates seems reasonable)

Could well be me.Probably not worth locating the error unless is used for further sums and/or eror is compounded.
Yes,possibly shows the correct trend re Heat/CoolantTemp.However have to cross-check all the parameters before can be judged with any confidence. This is a chore especially if Flow-rate measurements are just round the corner.

pHaestus · 10-23-2003, 09:49 PM

No testing tonight. I have to get up at 4am to catch a flight. I'll be back on Sunday with Joe's swissflow. I am tempted to take you up on the dP xmitter Bill but I don't even have a working DMM at the moment. Gotta buy a couple for my die simulator though. Recommendations?

KnightElite · 10-24-2003, 02:54 AM

I have a cheap Canadian Tire one you could borrow

.

Les · 10-24-2003, 08:28 AM

BillA · 10-24-2003, 08:51 AM

Quote:

Originally posted by KnightElite
I have a cheap Canadian Tire one you could borrow .

lol, ahem

well, for current the Keithley 195A is quite nice, BIG leds, good features, AND GPIB
second would be a Philips/Fluke PM 2534, lcd, good features, AND GPIB
and of course a GPIB pci card, and cables, etc. etc.
= automation !

the more modern approach is to use a (pricey) pci card to take the measurements directly in your pc (shunt for current)
but I am a Cro-Magnon and prefer instruments

Les, delete that 20W value - it was extrapolated

Les · 10-24-2003, 09:09 AM

Quote:

Originally posted by unregistered
.....Les, delete that 20W value - it was extrapolated

Done.

10-21-2003, 04:34 PM	#105
Aardil Cooling Savant Join Date: Oct 2002 Location: Louisiana Posts: 174	Editor's choice Personally I dont put much stock in Editors choice awards, Maybe we should have our own award : The I Survived the "Procooling Test of Endurance" Award Only offered to Products that actually pass the test of being usefull and do what they promise. Would make for quite a mile stone, I came close with my PWM controller, just needed a few more outputs and some LCD display, oh and bring the cost down to free. Dang I almost had one pH, You have done well from what I can understand, let no one say it useless info. Its only Useless to the people concerned with looks and not how it performs. Aardil

10-22-2003, 12:14 AM	#112
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	Les: This should look familiar to you: It's good to see data that's reproducible right? Now if we could only understand why that behavior occurs... And with today's data superimposed on the same old C/W vs coolant temp graph: I think this picture would be a good bit different with a clearer estimate of true W. I am also starting to wonder if maybe noninsulated wb and imperfectly insulated res and hose become a factor at low coolant temps and cause higher apparent C/W. Almost forgot; numbers for Les here: http://phaestus.procooling.com/mcwchill/temps3.jpg Last edited by pHaestus; 10-22-2003 at 12:21 AM.

10-22-2003, 06:12 AM	#114
Les Cooling Savant Join Date: Oct 2001 Location: Wigan UK Posts: 929	Shows no sign of anomaly suggested in my Morgan's DeltaT v MHz. In fact my results show a similar general trend. For now will put my results down to poor experimental technique but not completely discount that enigma could be associated with change in FSB. The inclusion of Mondays (220MHz, 1.81V) does qrestion the use of V and MHz data for guessing Wattage. Not sure how this data point stands up in "Coolant Temp based"guesses at Wattage . Hopefully will return to main thrust(Chiller performance) in next post(still sorting out data):- A comparison of Heat-source Wattage guesses, including the Flow-rate dependant DeltaT(chillin-chillout) guesses. With good Cold-side Insulation this would =DeltaT(cpuout-cpuin) and as such is potentially the definitve number. God pHaestus, I think we have entered Tester's Torture Chamber. Last edited by Les; 10-22-2003 at 06:19 AM.

10-24-2003, 02:54 AM	#122
KnightElite Cooling Savant Join Date: Sep 2002 Location: Saskatoon, Saskatchewan Posts: 294	I have a cheap Canadian Tire one you could borrow . __________________ Can anyone else here say that they have a watercooled monster that's 45" tall?

10-21-2003, 06:13 PM	#106
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	haha I was just making a smartass comment. No need for all the praise. All I do is the grunt work on this; the interpretation is clearly a group effort here. And I would rather do an article that I WANT to do than what is popular or good marketing any day. That's why these technical pieces are pretty few and far between on Procooling (takes so much time). Heading home now. I'll fire up the rig after dinner and report in at midnight as usual

10-21-2003, 09:25 PM	#111
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	Yup since87 I also remain curious. Almost all of my technical articles are fueled by interest in something similar. I disguise them as "cooling related" sometimes though

10-22-2003, 09:51 AM	#115
Les Cooling Savant Join Date: Oct 2001 Location: Wigan UK Posts: 929	After much pissing about. The most consistent calculations do appear to be the "Billa Delta Hot-Cold" based that you are using.Whether they are meaningful is perhaps another matter.

10-22-2003, 10:53 PM	#117
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	Revisited the flow rate issue again today: Looks like W are virtually unchanged (59.9-60 ain't bad) as would be expected considering voltage and MHz are unchanged. Performance indeed appears slightly worse when the flow is throttled. I am going to get Joe's Swissflow flowmeter this weekend and I will plumb it in and return to this issue on Monday.

10-23-2003, 09:08 AM	#119
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	Les: I will have to look at my spreadsheet again as my number for W SHOULD agree with Bill's Swiftech data as they come from the same eqn. Perhaps an error in my goal seek? The first set of full flow vs. half flow data was collected at 2200MHz and 2.0V whereas this is at 1.81V. There could be flow rate effects from this (higher W meaning higher flow rates seems reasonable)

10-23-2003, 09:49 PM	#121
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	No testing tonight. I have to get up at 4am to catch a flight. I'll be back on Sunday with Joe's swissflow. I am tempted to take you up on the dP xmitter Bill but I don't even have a working DMM at the moment. Gotta buy a couple for my die simulator though. Recommendations?

10-24-2003, 08:28 AM	#123
Les Cooling Savant Join Date: Oct 2001 Location: Wigan UK Posts: 929

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)