|
10-16-2003, 06:43 PM | #51 |
CoolingWorks Tech Guy Formerly "Unregistered"
Join Date: Dec 2000
Location: Posts: 2,371.493,106
Posts: 4,440
|
lol
my chiller was overwhelmed so I added a prototype product in series worked fine except the room temp rose at 1°C per 20 min (with all the instruments, etc) so then I put 8 ft hoses on it and ran it outside 2 big TECs plus a load will more than heat a room |
10-16-2003, 11:39 PM | #52 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
Another evening another set of numbers. The box fan seemed to help a good bit.
I HAVE to get a new motherboard; Asus A7V-133 isn't working and this 1C res crap is for the birds. I may just suck it up and go buy a board tomorrow. |
10-17-2003, 04:02 AM | #54 |
Cooling Savant
Join Date: Oct 2001
Location: Wigan UK
Posts: 929
|
Before I turn to the serious business of the day( the Newmarket Champion Stakes meeting) a little speculation :
Previously posted http://forums.procooling.com/vbb/sho...5&pagenumber=5 using Billa data from http://www.swiftnets.com/products/mc...sp#description Addendum. Dominantly Coolant Temp, Other or None? Last edited by Les; 10-17-2003 at 05:56 AM. |
10-17-2003, 08:31 AM | #55 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
I found a Leadtek NForce1 motherboard at otv KE. $59CAD. I would guess it lacks diode support (if not I'll return it and say it didn't suit my needs)
//edit: Crap it may not work: http://www.leadtek.com/motherboard/w...7n415da_1.html Overtemperature shutdown doesn't sound good at all. I'll call around and find a cheap ECS or something. |
10-17-2003, 09:00 AM | #56 |
CoolingWorks Tech Guy Formerly "Unregistered"
Join Date: Dec 2000
Location: Posts: 2,371.493,106
Posts: 4,440
|
heh Les - how did your Ascot card come out ?
looking like I may be proven wrong on the low flow I think what is skewing pHaestus' results is that while I used a very small pump (Siccie Nova), pHaestus has the same pump for both flow regimes (and throtteling the pump will put more heat into the coolant) pHaestus: How are you able to reset the valve to the same point ? ball valves are pretty twichy |
10-17-2003, 09:22 AM | #57 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
Highly scientific. I started out fully open on day one and then switched to as close to a 45 degree angle as I could get the valve. Then on day 2 I started at that point again and after collecting those points switched to wide open.
Don't ask me to reproduce the same points again though |
10-17-2003, 04:22 PM | #58 | |
Cooling Savant
Join Date: Oct 2001
Location: Wigan UK
Posts: 929
|
Quote:
Newmarket(July Meeting)_ won Goodwood - July/Aug- Lost - Impossible - going(sticky) draw Doncaster (St leger) - lost- going unseasonal. Oct 18th Newmarket(Champion Stakes/Ces)- dunno but going is Unseaonable Fast and most fast ground horses are past their bedtrime. |
|
10-17-2003, 06:19 PM | #59 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
New plan. I'll just remove the motherboard's connection to the diode pins. Then I can add my diode reader with no issue to the current board. As done here:
http://www.ocworkbench.com/hardware/...k7s5amodp6.htm I have a couple of dead boards on which to practice. Results soon hopefully. |
10-17-2003, 11:42 PM | #60 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
So far so good. I removed the diode connections from the socket, and at boot the CPU temperature is reported as -28C with a heatsink.
http://www.procooling.com/~phaestus/8k3aneutered.jpg This is because diode readers report -128C if there isn't a diode present, and I guess the motherboard doesn't have the extra digit to display the full temp. Anyway all I have to do now is solder wires onto the TBred without killing it I bet I am the first person ever to REMOVE internal diode readings from a motherboard! |
10-18-2003, 12:20 PM | #61 | |
Cooling Savant
Join Date: Oct 2001
Location: Wigan UK
Posts: 929
|
Quote:
Some pedictions/simulations: Simulated Chiller as "2m, 6mm ID tubing" using SF1.5 http://www.pressure-drop.com Gives good agreement with the Sicce Nova's flow of 2.417 lpm. * The only info I found http://www.sicce.co.uk/nova.htm |
|
10-18-2003, 04:48 PM | #62 | |
Cooling Savant
Join Date: Oct 2001
Location: Wigan UK
Posts: 929
|
More ramblings:
Quote:
A Full Flow rate of 2.417 lpm would suggest* that 163.0Watts and 218.5Watts are being Extracted in Tests 1 and 2. Extraction of ~ 200Watt should (according to Kryotherm/Since87) give coolant Temps 15 -18c.. Although I think that 2.417 lpm may be a reasonable estimate of Flow rate at Ambient, I suspect the Flow-rate falls quite rapidly as 0c is approached.This would alsoconveniently explain the apparent rise in "C/W" * Heat Extracted by Chiller(Watts)=Q = 69.5466 x flow-rate( lpm) x (Ti - To) ..... (Someone please Check sum) Full Flow Test1, Q =69.5466 x 2.4167 x 0.97 = 163.0 Watts, Full Flow Test2, Q= 69.5466 x 2.4167 x 1.3 = 218.5 Watts |
|
10-18-2003, 11:16 PM | #63 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
Today was a lot easier going than I expected. I went and picked up a new tip for my soldering iron, and managed to solder copper wire (from twisted pair) to the diode pins securely in a few seconds. First try even. Everything works as it should:
External #2 is the CPU diode. I have a pretty good technique for this modification now but I honestly hope this is the last time I do it. I guess with the socket destruction and this bit of soldering that one can add proper calibrated diode support to any motherboard regardless of COP or diode reading onboard. Dunno why anyone other than me would bother though I don't want to calibrate until tomorrow when I get some doublesided tape to hold the wires securely in place inside the center of CPU. |
10-19-2003, 02:10 AM | #64 | |
Cooling Savant
Join Date: Oct 2001
Location: Wigan UK
Posts: 929
|
Quote:
Otherwise will be quite a Heat Transfer conundrum. EDIT: IGNORE COMMENT I keep getting this wrong. You have soldered the "twisted pair" directtly to the CPU's pins and intend to calibrate as previously http://forums.procooling.com/vbb/sho...?threadid=6442. THEN install CPU in socket. Last edited by Les; 10-19-2003 at 02:44 AM. |
|
10-19-2003, 11:37 AM | #65 |
Cooling Savant
Join Date: Sep 2002
Location: Santiago, Chile
Posts: 403
|
Sure I'm late with this suggestion, but you didn't need to destroy the socket, maybe isolating the cpu pins with nail enamel will do the trick.
That's what I want to try, but your ¿5 to 1? score of dead cpu's in the soldering step is what's holding me. |
10-19-2003, 11:49 AM | #66 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
I'm not sure you could avoid having the pins come in contact with the wire in the socket (it's like an upside down "L" if you pop one out and look at it) just with polish. Maybe if you used really really thin wires but then you have resistance issues. After doing it, I would think that popping those pins out is actually easier than nail polishing the pins.
I have killed um 2 CPUs in this testing: one was killed by electricity when I plugged a pump into an AC outlet wired incorrectly and had voltage going through water with CPU connected to separate ground loop. Other one the wires weren't in between the pins properly when running out of the socket and as a result the ground wire got nicked in several places and was in contact with CPU pins. That mishap killed an 8K7A too. So really only one died directly from the modifications.I had a third 1600+ that had a diode pin fall off from the soldering, but the chip still works fine in my HTPC at 1575MHz. |
10-19-2003, 03:11 PM | #67 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
Yes I have the cart in front of the horse, but I decided to just test the chiller with diode prior to calibration. I'll adjust temperatures after the tests. It will hopefully make a big difference in test quality to have better resolution. It's running now so everything is at least functioning properly. Water -2.8C and dropping; CPU 11.875C and also dropping. I did lower the voltage a bit to 1.85V though. I also can run CPUBurn at high priority now that I dont rely on MBM readings.
|
10-19-2003, 09:40 PM | #68 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
//edit. I am pulling these numbers for now because I am not 100% confident yet in the setup. I am rerunning as many tests as I can tonight and will post the numbers I am confident in before bed.
Last edited by pHaestus; 10-20-2003 at 12:30 AM. |
10-19-2003, 09:55 PM | #69 |
Cooling Savant
Join Date: Oct 2002
Location: Louisiana
Posts: 174
|
GAWD I wish I was smart enough to sypher out what all them purty numbers mean
|
10-19-2003, 09:59 PM | #70 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
well the raw temps and the C/W should be fairly straightforward
|
10-19-2003, 10:12 PM | #71 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
Oh ONE interesting and useful observation today. In previous tests, I had a 120mm fan blowing onto the socket so that I could keep condensation at bay. This was recommended by several people, but when I tried it with and without fan today I noticed something interesting:
The performance of the chiller is noticeably worse when a fan is pointed at the socket/wb area. If you think about it, this makes perfect sense. The fan will cause heat transfer from the cold block surface to the case air to increase. This secondary loss will drive up temperatures of water and eventually CPU. So don't do that. Instead, the entire waterblock and the socket area really should be insulated. One other item of note from today. I am now testing with CPUBurn in high priority rather than "medium" since I am pulling all temperatures from external sources now. So today's numbers and previous ones aren't directly comparable. |
10-20-2003, 01:21 AM | #72 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
Ok was not a problem apparently as new data and old data all are in reasonable agreement. I tried to collect temps over a wide range of W today by testing @2200MHz and 1.6-2V and then for the last point I tested at 2084MHz and 1.6V. Here is the results of today/tonight's testing:
The W for C/W was estimated using BillA's hotside and chilled water temps to get a delta hot-cold from his data. A polynomial (2nd order) was fit to his numbers and then I just used goal seek to solve for W given my hot-cold delta T. My curve and bill's look similar but they don't seem to overlap at lower water temps. Not sure why off top of head; bad guess at W? I might as well put all the data up for your perusal: http://www.procooling.com/users/phae...l/newtemps.jpg |
10-20-2003, 01:37 AM | #73 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
Another useful graph:
Data starting to make a little more sense now. |
10-20-2003, 01:58 AM | #74 |
Big PlayerMaking Big Money
Join Date: Aug 2001
Location: irc.lostgeek.com #procooling.com
Posts: 4,782
|
Ok I just remembered that Bill's 20W data was extrapolated. If you ignore that point, our curves overlap exceedingly well:
I go to sleep now. |
10-20-2003, 02:01 AM | #75 |
Cooling Savant
Join Date: Sep 2002
Location: Saskatoon, Saskatchewan
Posts: 294
|
Good work pH. 23ºC below ambient for CPU is not bad...
Is this CPU running at 2.0V with CPUburn at high priority?
__________________
Can anyone else here say that they have a watercooled monster that's 45" tall? |
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
|
|