I know deltas in WC or Air are totally different.
Thank you very much for your kindness bachus, I will update when I get my system finally tunned.
Printable View
I know deltas in WC or Air are totally different.
Thank you very much for your kindness bachus, I will update when I get my system finally tunned.
do you know where I can buy one DT200 Thermometre like yours please? :)
Very nice unit that you have here bachus :)
I'll probably buy one on July. :)
No problem... I'll be following how your cooperation with Chilly1 turns out ;)Quote:
Originally Posted by Ssilencer
I got mine from Chilly1 (PM him for pricing when interested). You can alos find some good deal on Ebay sometimes, like the one HERE :) Also, I believe you can find it online in other stores. That's as far as shopping in US is concerned ;)Quote:
Originally Posted by eviLRuLeZ
I think It'd be better to order directly one single phase change unit to chilly1 :D
Thanks a lot bachus anonym again :)
have a nice day
Your unit is a beauty bro :toast:
Yea nice write up!
The very first pic is great, especially after all the damaged unit's we've seen lately.
@ k|ngp|n, GoodOmens and G H Z
Thanks a lot guys :toast: I'm really glad you like that cooler.
--------------------------------
You all probably thought that would be all about the cooler, right? WROOOONG!!! :lol:
Chilly1 asked me If I could perform a test, measuring temperatures using temp probes instead of software (MBM, in that case). After some brainstorming (mostly Chilly1 threw in ideas :D) I tried to do my best mounting two additional temp probes that I had, totaling three measuring points:
(a) first probe, that Chilly1 put inside the block, about 1/2 inch up from the bottom of the block (that is where he puts temp probe as a standard),
(b) second probe, I mounted on IHS of my Winchester,
(c) third one, was attached to the side of the block, just below "O ring" as far on the bottom as I could safely do it.
Then I measured IDLE temps and LOAD temps, running Prime95 SmallFFT (for maximum heat) for slightly longer than 60mins @ my max Prime95 stable clocks/volts - 2860MHz / 1.84v. The results are as follows ;)
Legend, going in ascending order of photos attached
1. This is how I managed to attach temp probe to the IHS. As I know this is not perfect placement, this is the best I could do, as block doesn't leave much space on the IHS after it's mounted. So placing it in the corner was as good as it gets, without any problems leveling the block during mounting.
2. Here's how I did temp probe for bottom of the block. Same as above, If I could I would have mounted it directly to the bottom of the block (not it's side) but it's impossible. So this is how I did it :D
3. This is the photo showing how I punctured the gasket to allow the "Block" probe inside.
4. And that shows same, but for "CPU" probe :D
Probes were mounted on oposite sides.
5. This photo shows IDLE temps: TEMP1 is CPU and TEMP2 is BLOCK.
1. This is a screenshot I made at the same time when I took photo of IDLE temps.
2. This one, shows LOAD temps (during Prime95 SmallFFT, after 60mins, still going): TEMP1 is CPU and TEMP2 is BLOCK
3. Next, I swapped probe (TEMP1) and put Chilly1's probe, to measure evap temp at LOAD: TEMP1 is EVAP and TEMP2 is BLOCK
4. This is a screenshot of made at the same time when I took photo of LOAD temps (during Prime95 SmallFFT, after 60mins, still going).
Don't ask me why HIGHEST temp for CPU is 22C. I have absolutely no idea, might have been some odd reading at some point :shrug:
5. Last picture shows IDLE temps, again. But this time I wanted to show EVAP temp while IDLE: TEMP1 is EVAP and TEMP2 is BLOCK
----------------------
SUMMARY
2860MHz @ 1.84v
IDLE - EVAP -43.1C / BLOCK -34.1C / CPU IHS -32.4C
LOAD -EVAP -38.8C / BLOCK -23.4C / CPU IHS -23.6C
IDLE-LOAD DELTA (CPU IHS) ~ 8.8C
IDLE-LOAD DELTA (BLOCK) ~ 10.7C
IDLE-LOAD DELTA (EVAP) ~ 4.3C
----------------------
I understand that results may not be 100% accurate, considering placement of probes. It's just very difficult to do it any other way, without removing IHS altogether. Then also, IHS temperature is not CPU's core temperature. I'm trying to understand what temps would I get If I could place the probe closer to the center of the IHS. It might be possible that it would be even lower by 2-3C. But that test should give at least some info regarding DELTA values.
I might remove IHS of that CPU, but that will have to be done later. Then placement of CPU probe would be almost perfect and I'm pretty sure it would shave a bit of Cs off, too :)
I'm very happy with performance that this cooler gives me. 3027MHz benchable in SPi16M, 32M and all other benchmarks, including 3D + 3034MHz in Pifast and 3065MHz in SPi 1M is really good for my "not-the-best" Winchester.
Any feedback (comments or ideas how to improve the method) will be VERY appreciated.
If you ever saw an ihs out of a proc, you know that at that point is thicker than the center and the a64 ihs is is overall thicker than the ihs in p4, you can not take any real proc temp from the ihs.
How your delta pressures betwin idle and load?
Hehe... That I know even too well. That's why I was not trying to say that -24C LOAD is CPU temp, but CPU IHS temp. And still, not the best measurement there is as It's impossible to get it attached closer to the middle of the IHS, where actual core is. If you have better idea how to get as close as possible to real CPU die temp measurement, then let me know :D Without removing a heatspreader, that is ;)Quote:
Originally Posted by Ssilencer
I still believe that even after the IHS is removed, temps would be much lower than what MBM or Speedfan are showing. I just can't imagine that there would be -24C on IHS and +5C on a DIE. It just looks way to much of a difference.
What exactly you're asking me about?
I just figured what you asked me about... :D Well, I can't measure that. I'm just a OCer, unfortunately I have no skills/tools/gases/instruments to built my own phase change. So I'm not able to answer that question for ya :(Quote:
Originally Posted by Ssilencer
EDIT: Can you run EXACTLY same test, with same method at your MAX Prime95 SmallFFT stable config?
EDIT: I'm leaving in few minutes, so I can answer to any other questions later today...
I'm asking about heat transfer, any changes in heat load will be reflected in presures I think.
I edited my post above... In other words, I'm not able to give you any answer to that question. I have absolutely nothing to do with building phase change units. I just use 'em :lol:Quote:
Originally Posted by Ssilencer
I cant run the test you need to see, cause I dont have my unit here, it's in the cosmetologist :D (Profesional case, G-froster X integration, etc)
But I can give you an idle and load screenshot, running at 3ghz with superpi 32mb running.
Cool!
It would be so nice if more people were trying to test their single stage cooling same way, skipping software and just going straight for probing CPU temps ( if possible, without IHS :D ). That would give builders some nice, useful info about real-life, real-load testing without wondering if onboard sensors are showing right temps and behave normally. Also, not only going by evap temps but cpu temps, measured with probes.
Anyway, this whole test made be wanna remove IHS and try again without it. Maybe that Winnie would get a bit higher ;) 3.1GHz SuperPi 16M/32M would be awesome :D
Dude! Either show better pic or sth, because that totally doesn't look like -6C to me! It more like +6C. I zoooooooomed in and compared both, and it seems like "-" sign is missing on that LOAD pic :confused:
Dude! IT IS +6c PIC!!Quote:
Originally Posted by bachus_anonym
That's why I'm interested to see where is the prob, when my other systems do -11c full load at 3100mhz or more and with the same motherboard and micro..
Ooohhhh! I thought that you said (you did in that one post above) that you got -6C. And I was like "Damn, something must be horribly wrong" :lol:Quote:
Originally Posted by Ssilencer
I even just took a screenie with SpeedFan @-8C and -26C to compare with yours :lol:
That was what I went with :) Something is wrong then... At -39C you got +6 CPU temp on that system you showed pic of above. And then u say that on other system (I assumed, you mean different one) you do -37C and get -11C on CPU. Doesn't seem right, might be some error.Quote:
Originally Posted by Ssilencer
speedfan is not accurate guys
the best is a thermo probe on cpu ihs, you did it isn't it?
bachus_anonym
Here is a comparison with my FX-55 with a MachII 404a moded by "Doctor" Ricky, running prime at around 3.350mhz with core at around 1.8V.
http://pedrorocha.planetaclix.pt/Prime3350.jpg
So, as you can see, DFI Cpu sensor (read by MBM, speedfan..) is not very trustable or accurate.
To check the good contact, better go to Bios and check the CPU temp - with the FX-55 at 1.8v I see around -11ºC with this MachII-404a.
Also note:
- The FX-55 at 1.8v produces a LOT more of heat than the A64 3000+ Winnie, thats why he only feels good at "Doctor" Ricky cascade running at 4.000Mhz :)
- Even runnig the fans at very low speed (almost no noise for 24/7) the MachII evap temp readings are very optimistic;), with digital thermometer should be more like -50ºC idle and -40ºC load ..
So here it is my advise : enjoy your new 402 XTREME, as usual Chilly1 have done a very very good work, after all he is a XTreme Legend of Cooling ;)
:toast:
I will have access to a Test board , similar to the one used in overclockers with wb,etc, two dataloggers, calibrated thermal resistance and 4 pt100, in a matter of weeks.
Then, I will test diferent evaps in the same system and we will have a better picture of reality.
Pedro, I think the heat load generated in bios setup is diferent than the heat load generated with prime 95 in windows, am I wrong?
Forget about crappy MachII probe :D
Of course it is!!Quote:
Originally Posted by Ssilencer
You missunderstand me - check the temp in Bios it s only a good way to quick check the good contact with evaporator afer mounting and also it is a rasonable way to compare phase-change sys performace ..just that :cool:
And now, we agree :D
Actually... that isn't out of the realm of reality... with your Vcore and Mhz used I figure something on the order of a 0.21 C/W thermal impedance... considering that your measurement is on the corner of the IHS and not at the center, it is actually lower. That said, those IHSs and the paste between them and the die stink ;)Quote:
Originally Posted by bachus_anonym
What surprises me to a degree [ pun intended :p: ] is the difference from the embedded probe ("evap") to the side of the block... I might suggest that you do two things... get yourself some copper tape to mount the probes with and put some very good paste on your probe to help get better contact.
I guess I need to take a closer look at Chilly1's blocks...
Peace :toast:
Ok, this is gonna be the last one, I promise :lol: I just thought that since I'm on it I will show you some more testing I've done tonight :)
I wanted to see, what difference is between temps shown in BIOS / Hardware Monitor in comparison to temps I got from all 3 probes. Sitting in BIOS gives CPU a little bit of excercise. Then I run same LOAD test in Windows, as I did last time: PRime95 SmallFFT.
1.This is what BIOS shows for temps.
2. Then we see temps from probes: TEMP1=CPU IHS TEMP and TEMP2 is BLOCK TEMP
3. The I switched probes to show evap temp: TEMP1=EVAP IHS TEMP and TEMP2 is BLOCK TEMP
4. This is sitting IDLE in Windows: TEMP1=EVAP IHS TEMP and TEMP2 is BLOCK TEMP
5. Again, with probes switched: TEMP1=CPU IHS TEMP and TEMP2 is BLOCK TEMP