Current review project (MCWCHILL) - Page 2

Since87 · 10-10-2003, 06:18 AM

Quote:

Originally posted by Les
2) Give "heat pumping" results similar to "Pefectly Insulated" Kryotherm sums for a 172W TEC.

Sean, I think it is both a neat and useful tool.. Guesses at the enigmatic "Insulation Losses/Gains" can be added using Kryotherm.

Glad you like it.

It would be fairly easy to add the Rins factor to the spreadsheet. Also, with the spreadsheet, Rins could be setup to use an 'alternate ambient' (e.g. motherboard temperature) enabling more accurate modeling in that respect than Kryotherm.

BillA · 10-10-2003, 08:39 AM

no hurry pHaestus, keep it as long as you can extract more data from it
but hey, insulate all of the coldside loop !

Joe · 10-10-2003, 08:43 AM

Quote:

Originally posted by pHaestus

Some pics of the monstrosity:

Theres entirely too few LED fans in that case, how can it be a good test rig?!

Poser!!!!

hehehe

BillA · 10-10-2003, 09:02 AM

re the monstrosity:

we had the GM of a rather large co in last week and after a half hour of looking over my 2 rooms of equipment with its rat's nest of cables and hoses going back and forth between the rooms and equipment racks and setup table, he said:
"Very good, very good. This is most practical. Just as when I visit Professors in universities, not new and fancy equipment but the correct equipment with wires running everywhere. Not like big companies with $250,000 in equipment form which they admit they have only 70% accuracy."

pHaestus' test setup is not so bad really
the MCW-CHILL is a bear to connect up

the room was warming up, lol

pHaestus · 10-10-2003, 10:10 AM

Les:

When I fire it up tonight I will get some rad in and out and hot side water temperatures too. No biggie. I have a couple more water and air probes for the DIgitec. I still want to measure the CPU backside temperature as well.

Bill I didn't insulate it because I was using straight water in the coldside loop. I did this because it was such a pita to fill and I otherwise would have bee sucking down lots of glycol :/

I was most impressed by the SPEED at which it drops in temp (if you have resolution to watch it). Speaking of which, I don't think KnightElite was suitably impressed by the fact that my two digitec probes read exactly the same temperature (0.01C res).

Since87 · 10-10-2003, 10:43 AM

Quote:

Originally posted by pHaestus
Speaking of which, I don't think KnightElite was suitably impressed by the fact that my two digitec probes read exactly the same temperature (0.01C res).

Shame on you KE. More 'oohs' and 'aahs' tonight please.

Joe · 10-10-2003, 11:48 AM

Quote:

Originally posted by pHaestus
Speaking of which, I don't think KnightElite was suitably impressed by the fact that my two digitec probes read exactly the same temperature (0.01C res).

Thats what the LED fans are for silly! I know I cant help but be impressed at a case loaded with lit up fans

hehehehee

KnightElite · 10-10-2003, 11:49 AM

Yes, the accuracy of the digitech probes is indeed impressive

.

Unfortunately, I won't be able to come back this weekend, since I'm going to visit my parents. Looking forward to seeing more numbers though

.

utabintarbo · 10-10-2003, 11:59 AM

Quote:

Originally posted by KnightElite
Yes, the accuracy of the digitech probes is indeed impressive .

Unfortunately, I won't be able to come back this weekend, since I'm going to visit my parents. Looking forward to seeing more numbers though .

Dude, where are your priorities? WTF?

Bob

Les · 10-10-2003, 06:54 PM

Quote:

Originally posted by Since87
Glad you like it.

It would be fairly easy to add the Rins factor to the spreadsheet. Also, with the spreadsheet, Rins could be setup to use an 'alternate ambient' (e.g. motherboard temperature) enabling more accurate modeling in that respect than Kryotherm.

Not sure how would solve for dT algebriacallly.
For limited no of calcs I resort to iteration.
My own interests moved on from TECs via Water to Air. I do not think I would be very active in the further developent of TEC simulations.

KnightElite · 10-14-2003, 12:42 AM

Post some new updates!

I know they are out there....

.

pHaestus · 10-14-2003, 11:03 AM

Not TOO much to report. Some issues with digital pressure gauge; still ironing that out. I added antifreeze and insulation and can get -7 to -8C water without PC running. I epoxied a DOW sensor onto the base of the TBredB last night; I'll reinstall and see if I can get some higher resolution numbers tonight. Here's a couple of pictures:

http://www.procooling.com/users/phae.../coldwater.jpg

http://www.procooling.com/users/phae.../insulated.jpg

I have to get more insulation on the res still. I was having problems with water freezing in the chiller (hence the antifreeze) and was using the res to see the movement of water to the pump.

It takes a WHILE for the system to reach steady state; no hotside water or rad air in and out temps yet. Hopefully with CPU installed tonight I can get at least a single "open flow" set of temps to post.

pHaestus · 10-16-2003, 02:27 PM

Some preliminary temperatures from last night:

CPU diode is taken from motherboard (1C increments

)
CPU back is a DOW sensor epoxied under the center of the CPU. I noticed that this temperature was MUCH higher than the diode temp and so I also collected temp from a DOW sensor taped onto the back of the motherboard behind the socket area. Mobo back is this temp. Note that it is higher than the temp of the probe on the CPU; this suggests to me that all the mosfets and circuitry powering the system results in a seriously bad reading for those epoxying probes onto chips or (worse) using the in socket thermistor).
WB top is a DOW sensor on the top of the waterblock itself.

Rad air in and out are obviously the intake and exhaust sides of the radiator, and hotside water is the water temp of the hot loop.

Full flow set of data is with the chiller loop "wide open" while the 1/2 flow set are with a ball valve turned to a 45 degree angle. Sorry no flow rates yet; issues with the digital pressure gauge preclude them. Still fiddling with that. Based on die - ambient numbers I would estimate a TBredB at 2.0V 2200MHz to be putting out ~ 60 real W. That's using Bill's numbers here:

http://www.swiftnets.com/products/MCWCHILL-452.asp

I will set up a range of Mhz tonight hopefully.

Joe · 10-16-2003, 02:36 PM

Wow pH

DOW interference from the surrounding hardware? Thats wild.

pHaestus · 10-16-2003, 02:39 PM

no no no. Not interference. HEAT. The motherboard is substantially hotter than the CPU, and so the probe epoxied on backside is not very representative.

Joe · 10-16-2003, 02:57 PM

Hmmm could be the whole... insulation properties of glass epoxy wafers, and a plastic socket. Since the core is actively cooled all the secondary heat just collects and heats up the mobo and socket... but still... weird...

Les · 10-16-2003, 03:02 PM

Excellent.
Initial thoughts:
Delta T chiller(Hotwater - Coldwater)= 31.5c
For DeltaTchiller = 31.5c , Bill had an imput of (80w + Pump + Insulation gains).
So suspect your Total Imput would be similar.
Guess would be 80w from CPU (Includes Insulation(Secondary Heat) gains from motherboard)*

Will bore you further as I digest this and subsequent data.

* This is not far removed from a CPU Heat of 50w and Ri=1c/w (to MB at 31c) ....... EDIT

pHaestus · 10-16-2003, 03:18 PM

Les that number seems a bit closer to what I would expect based upon aircooling and water cooling temperatures with same chip (known C/W values then).

Les · 10-16-2003, 03:24 PM

Quote:

Originally posted by pHaestus
Les that number seems a bit closer to what I would expect based upon aircooling and water cooling temperatures with same chip (known C/W values then).

Maybe(but only maybe) my belated edit illuminates.

BillA · 10-16-2003, 04:50 PM

correcting for the inlet air temp diff, it would seem that the lower flow rate is more effective
but the discrepancies in some number sets preclude that conclusion

comments pHaestus ?

Les · 10-16-2003, 05:14 PM

Quote:

Originally posted by unregistered
correcting for the inlet air temp diff, it would seem that the lower flow rate is more effective
but the discrepancies in some number sets preclude that conclusion

No.
Unless have my silly head on:-
High flow gives more cooling : delta T( Air in - cpu) is 0.53c larger.
Relative to chilled water it also gives better cooling(smaller "C/W") : delta T(CPU- water) is 0.7c smaller.
However all are within the 1c resolution of the Diode measuring.

comments pHaestus ?

pHaestus · 10-16-2003, 05:43 PM

I was operating under the assumption that a lower cpu to ambient or cpu to chilled water differential meant more cooling (CPU is closer to room temp or water temp when the delta T decreases). That presents a problem though: How can the delta T for CPU-air respond in a different direction to flow rates than the delta T for CPU-water?

I really need an old throwaway motherboard so I can get tolerable diode readings again. Will see if it is possible to resurrect my KT133A board again tonight.

BillA · 10-16-2003, 06:24 PM

would suspect equilibrium

pHaestus · 10-16-2003, 06:28 PM

A likely culprit. The dual pelts are dumping a LOT of heat into my room; I ran the system for several hours last night but perhaps it still wasn't enough. It's getting to cold to open a window, but I can scrounge up some box fans and point em at the setup tonight at least.

Les · 10-16-2003, 06:39 PM

Quote:

Originally posted by pHaestus
....
That presents a problem though: How can the delta T for CPU-air respond in a different direction to flow rates than the delta T for CPU-water?
.........

Does it respond in "a different direction"?

There is a "minus sign"(-) for the differential between CPU and "air in".

10-10-2003, 11:49 AM	#33
KnightElite Cooling Savant Join Date: Sep 2002 Location: Saskatoon, Saskatchewan Posts: 294	Yes, the accuracy of the digitech probes is indeed impressive . Unfortunately, I won't be able to come back this weekend, since I'm going to visit my parents. Looking forward to seeing more numbers though . __________________ Can anyone else here say that they have a watercooled monster that's 45" tall?

10-14-2003, 12:42 AM	#36
KnightElite Cooling Savant Join Date: Sep 2002 Location: Saskatoon, Saskatchewan Posts: 294	Post some new updates! I know they are out there.... . __________________ Can anyone else here say that they have a watercooled monster that's 45" tall?

10-16-2003, 02:36 PM	#39
Joe The Pro/Life Support System Join Date: Dec 1999 Location: Denver, CO Posts: 4,041	Wow pH DOW interference from the surrounding hardware? Thats wild. __________________ Joe - I only take this hat off for one thing... ProCooling archive curator and dusty skeleton.

10-16-2003, 02:57 PM	#41
Joe The Pro/Life Support System Join Date: Dec 1999 Location: Denver, CO Posts: 4,041	Hmmm could be the whole... insulation properties of glass epoxy wafers, and a plastic socket. Since the core is actively cooled all the secondary heat just collects and heats up the mobo and socket... but still... weird... __________________ Joe - I only take this hat off for one thing... ProCooling archive curator and dusty skeleton.

10-16-2003, 03:02 PM	#42
Les Cooling Savant Join Date: Oct 2001 Location: Wigan UK Posts: 929	Excellent. Initial thoughts: Delta T chiller(Hotwater - Coldwater)= 31.5c For DeltaTchiller = 31.5c , Bill had an imput of (80w + Pump + Insulation gains). So suspect your Total Imput would be similar. Guess would be 80w from CPU (Includes Insulation(Secondary Heat) gains from motherboard)* Will bore you further as I digest this and subsequent data. * This is not far removed from a CPU Heat of 50w and Ri=1c/w (to MB at 31c) ....... EDIT Last edited by Les; 10-16-2003 at 03:15 PM.

10-10-2003, 08:39 AM	#27
BillA CoolingWorks Tech Guy Formerly "Unregistered" Join Date: Dec 2000 Location: Posts: 2,371.493,106 Posts: 4,440	no hurry pHaestus, keep it as long as you can extract more data from it but hey, insulate all of the coldside loop !

10-10-2003, 09:02 AM	#29
BillA CoolingWorks Tech Guy Formerly "Unregistered" Join Date: Dec 2000 Location: Posts: 2,371.493,106 Posts: 4,440	re the monstrosity: we had the GM of a rather large co in last week and after a half hour of looking over my 2 rooms of equipment with its rat's nest of cables and hoses going back and forth between the rooms and equipment racks and setup table, he said: "Very good, very good. This is most practical. Just as when I visit Professors in universities, not new and fancy equipment but the correct equipment with wires running everywhere. Not like big companies with $250,000 in equipment form which they admit they have only 70% accuracy." pHaestus' test setup is not so bad really the MCW-CHILL is a bear to connect up the room was warming up, lol

10-10-2003, 10:10 AM	#30
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	Les: When I fire it up tonight I will get some rad in and out and hot side water temperatures too. No biggie. I have a couple more water and air probes for the DIgitec. I still want to measure the CPU backside temperature as well. Bill I didn't insulate it because I was using straight water in the coldside loop. I did this because it was such a pita to fill and I otherwise would have bee sucking down lots of glycol :/ I was most impressed by the SPEED at which it drops in temp (if you have resolution to watch it). Speaking of which, I don't think KnightElite was suitably impressed by the fact that my two digitec probes read exactly the same temperature (0.01C res).

10-14-2003, 11:03 AM	#37
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	Not TOO much to report. Some issues with digital pressure gauge; still ironing that out. I added antifreeze and insulation and can get -7 to -8C water without PC running. I epoxied a DOW sensor onto the base of the TBredB last night; I'll reinstall and see if I can get some higher resolution numbers tonight. Here's a couple of pictures: http://www.procooling.com/users/phae.../coldwater.jpg http://www.procooling.com/users/phae.../insulated.jpg I have to get more insulation on the res still. I was having problems with water freezing in the chiller (hence the antifreeze) and was using the res to see the movement of water to the pump. It takes a WHILE for the system to reach steady state; no hotside water or rad air in and out temps yet. Hopefully with CPU installed tonight I can get at least a single "open flow" set of temps to post.

10-16-2003, 02:27 PM	#38
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	Some preliminary temperatures from last night: CPU diode is taken from motherboard (1C increments ) CPU back is a DOW sensor epoxied under the center of the CPU. I noticed that this temperature was MUCH higher than the diode temp and so I also collected temp from a DOW sensor taped onto the back of the motherboard behind the socket area. Mobo back is this temp. Note that it is higher than the temp of the probe on the CPU; this suggests to me that all the mosfets and circuitry powering the system results in a seriously bad reading for those epoxying probes onto chips or (worse) using the in socket thermistor). WB top is a DOW sensor on the top of the waterblock itself. Rad air in and out are obviously the intake and exhaust sides of the radiator, and hotside water is the water temp of the hot loop. Full flow set of data is with the chiller loop "wide open" while the 1/2 flow set are with a ball valve turned to a 45 degree angle. Sorry no flow rates yet; issues with the digital pressure gauge preclude them. Still fiddling with that. Based on die - ambient numbers I would estimate a TBredB at 2.0V 2200MHz to be putting out ~ 60 real W. That's using Bill's numbers here: http://www.swiftnets.com/products/MCWCHILL-452.asp I will set up a range of Mhz tonight hopefully.

10-16-2003, 02:39 PM	#40
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	no no no. Not interference. HEAT. The motherboard is substantially hotter than the CPU, and so the probe epoxied on backside is not very representative.

10-16-2003, 03:18 PM	#43
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	Les that number seems a bit closer to what I would expect based upon aircooling and water cooling temperatures with same chip (known C/W values then).

10-16-2003, 04:50 PM	#45
BillA CoolingWorks Tech Guy Formerly "Unregistered" Join Date: Dec 2000 Location: Posts: 2,371.493,106 Posts: 4,440	correcting for the inlet air temp diff, it would seem that the lower flow rate is more effective but the discrepancies in some number sets preclude that conclusion comments pHaestus ?

10-16-2003, 05:43 PM	#47
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	I was operating under the assumption that a lower cpu to ambient or cpu to chilled water differential meant more cooling (CPU is closer to room temp or water temp when the delta T decreases). That presents a problem though: How can the delta T for CPU-air respond in a different direction to flow rates than the delta T for CPU-water? I really need an old throwaway motherboard so I can get tolerable diode readings again. Will see if it is possible to resurrect my KT133A board again tonight.

10-16-2003, 06:24 PM	#48
BillA CoolingWorks Tech Guy Formerly "Unregistered" Join Date: Dec 2000 Location: Posts: 2,371.493,106 Posts: 4,440	would suspect equilibrium

10-16-2003, 06:28 PM	#49
pHaestus Big Player Making Big Money Join Date: Aug 2001 Location: irc.lostgeek.com #procooling.com Posts: 4,782	A likely culprit. The dual pelts are dumping a LOT of heat into my room; I ran the system for several hours last night but perhaps it still wasn't enough. It's getting to cold to open a window, but I can scrounge up some box fans and point em at the setup tonight at least.

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