80MM fan instead of 92MM on SLK900A (Mack is right!)
I saw Mackanz say something about suspecting that 92MM fans were not effective on SLK-900 series heatsinks. So I decided to try it out.
I was running my Primary rig with a 64CFM 92MM fan since I built it in this incarnation. It ran stable and fine with the Barton XP-M 2400+ @ 2.3GHz and 1.75VCore for weeks. Load temp was 43.5C this morning before the switch. (Don't even ask me what idle temps are, this processor never idles)
I switched out for a 53CFM Mechatronics 80MM fan. Runs about the same noise level, maybe a little quieter even. After another few hours running, max load temp was 40.5C.
So I dropped 12MM, 11CFM, and 3C in a matter of minutes. My guess is that the smaller spindle of the 80MM fan, coupled with more focused airflow through the SLK-900A fins, allowed for a temperature drop. Size might matter, but it seems that efficiency matters more!
Guess which fan is staying on my heatsink and which fan is going to be 10 dollars wasted in my spare parts bin?
I was running my Primary rig with a 64CFM 92MM fan since I built it in this incarnation. It ran stable and fine with the Barton XP-M 2400+ @ 2.3GHz and 1.75VCore for weeks. Load temp was 43.5C this morning before the switch. (Don't even ask me what idle temps are, this processor never idles)
I switched out for a 53CFM Mechatronics 80MM fan. Runs about the same noise level, maybe a little quieter even. After another few hours running, max load temp was 40.5C.
So I dropped 12MM, 11CFM, and 3C in a matter of minutes. My guess is that the smaller spindle of the 80MM fan, coupled with more focused airflow through the SLK-900A fins, allowed for a temperature drop. Size might matter, but it seems that efficiency matters more!
Guess which fan is staying on my heatsink and which fan is going to be 10 dollars wasted in my spare parts bin?
0
Comments
Yeah, I am running 80 mm fans, one on the SLK-900A and one on the Swiftec HS on the P4 running beside me so I can scope out temps before I load an O\S back on it. The Swiftec kicks, like the spindled heat disipators, and the YS Tech fan works fine at about 80% of standard RPM to keep the CPU a couple degrees below where the CPU hung under load previously-- I stuck the rheostat that came wired to the YSTech through a spare floppy plate, it sits above floppy drive where it is easy to reach. The IC7-Max3 is a sweet board, too.
The AS 5 is curing in, it is a tib better than the Ceramique now, we will see if it improves over the next month or two like the Ceramique did. It was actually EASIER to spread for me than Ceramique, but I warmed it up first in a pocket and air temp was in the high 70's to start with. Someday I will post a pic of the nylon pot scraper I use for this, it makes the job hyper-easy. today, I went to get the razor blade out, all rusty after one month of non-use when it had ZERO rust last time I used it-- it gets HUMID down south.
John D., who was not really blathering much, he realizes that HS, FAN, and compund all team to cool a CPU. But Ghoosdum is perfectly right about an 80 mm fan vs a 92 on the SLK-900's.
I was referring to using a larger fan reduced down to focus its flow on a smaller area. Thus getting all or most of its volume where it counts.
The only problem is, that it is quite noisy.
But it sure moves some air at the max setting, and the speed can be adjusted if you are annoyed by the noise.
76.7 CFM at max rpm's that's pretty good for an 80mm fan.
Basicly, if you have a fan where the working area (blade area inside case) sticks out beyond the fan, you are cooling the outer fins more than the inner fins(open air is pulled through the fan faster than air is sucked up between HS fins, you end up cooling the colder part of HS most with too big a spindle).
If you instead tunnel the flow to HS width, you are cooling more fins that are more toward the middle. Part of the reason the 80 works better is it also has a bit smaller spindle usually, more area toward center of HS where core of CPU is (essentially, it is not offset enough to make the core be outside the spindle at all for either fan) is more effectively cooled as the spindle and bearings and "motor" area is not blocking center as much. So, if fan is efficiently built, and blades are shaped to suck or blow air and allow for some backpressure, than the 80 will be better. If you want a fan that is more resistant to backpressure, you need one that does not let the bearing assembly "settle" when backpressure is present-- if you have a pin and spindle bearing, the fan working parts get pushed back with backpressure, friction at bearing increases, and fan actually slows down more and more over time as bearings wear and lubrication degrades with heat. Motor has to work more and more to pump same volume due to bearing friction. So, something that does not get hot when pressure exits against flow is better. Ball bearings actually have a smaller friction surface than spindles and pins do, rubbing. A bit less heat results until the race the bearings are running in wears and becomes mis-shaped. Eventually, the race gets enlarged, and the blade wobbles in fan.
A fluid bearing fan, if the liquid never boils or separates with heat, or takes avery long time to do so, dissipates heat through the volume of the liquid bath. It stays more effective longer if designed right, and is less likely to wobble-- wobbling creates turbulence in fan case area that can be uneven and blow air some against walls of case instead of straight out case opening, then that air has to go somewhere.
So, I used a Panaflo on the SLK 900A and my Barton, 80 mm after trying another fan. The fan I tried with a single-ball and spindle bearing structure began to give a high squeal and warble way too soon. That was from bearing rub due to pressure forcing bearing parts to rub more than was wanted. As fan squealed more, CPU temp rose, fan was inefficeint and getting more and more ineffective at doing its cooling job. Same model fan worked fine in a case input or output task, less uneven pressure to force bearing parts to rub more.
As to the empty fan under a working one-- if both same size, you get a duct effect, the air that would be blocked by the spindle close to fin tops is able to flow together and more even volume from all fin gaps happens. So fan bearings wear less unevenly, less heat, and less bearing wear.
John D.
I have a Panaflo 92 mm sitting on my Thermalright SP-97. I also have the exact same Panaflo fan, but as an 80mm, just laying around.
So I switched them, tried to run the CPU hard, and see what I got.
Couple of things; (using the Abit temp monitor), the CPU temp ran 2-4C hotter, both under load and at idle. And it took a couple of minutes longer to get back to idle temps.
In no way am I trying to dispute what you guys have found. Believe me, I'm pretty inexperienced at this. This is just my findings with my set up.
But the thing I was wondering. Both my fans have the same spindle size. As you guys have swithched fans, you have switched spindle sizes also. Is part of your extra cooling coming from that? Just a thought.
Thanks for giving me another reason to play in my computer.
Dragstk
Also, what are your case temps like, and is the heatsink nice and clean or has it been assimilated by the world's largest dust bunny?
According to the Abit monitor, case temps went from 24C (with the 92) to 28C (with the 80. CPU temps went from 40(w/92) to 44(w/80), at idle.
HS is very clean, as I just did case maintenance(general dusting and cleaning/filter replacement) this morning.
Eh no I haven't, I have heard the 60mm Delta fan though.
And boy that was loud.
I guess I am calling the SFII loud, is because I upgraded from a silent viking, it is indeed silent. (rated at 32,4 CFM/21 dBa)
You don't want to get in the way of a 60mm Delta fan.
I once saw a page where small cars were driven by Delta fans.
I have seen injuries from a delta too, it will cut you down to size if you are not careful.
Case temps went up too because the 92MM has a bit of overlap so the extra was cooling the NB as well. It's not 'case' temp per se in Abit monitor, it's actually NB temp, I'm fairly certain.
Although it's probably what you meant, 92mm's are only 'pwned' when it comes to heatsinks. They still make great intake or exhaust fans.
as for the 80mm vs 92mm fan, i've seen somewhere ppl say that it really depends on the location of the temp sensor for ur cpu (socket)........some is located around the socket i think....
a 92mm fan moves more heat away from the heatsink so there's more hot air surrounding the socket and thus getting a higher temp.....since a 80mm fan moves less heat, the air around the socket is less hot and so temp sensor would read a lower temp compared to a 92mm fan......
i'm not sure about this since i didn't do the experiment myself, but this is what i've seen from other forums talking about.....and i agree w/ the fact that the 92mm just cools the outer fins of the HS more rather than the middle cus of hub size.....
and i'm still wondering if i should get a TT SFII or 92mm Panaflow myself
I have a slk-948u and i figured i would experiment....btw, it's on a p4 2.8 @ 3.26 and AS3. First a 92mm no-name from an old psu and then an 80mm panaflo. I think both push about the same cfm (>50).
Monitored on Abit EQ...
cpu
80mm = 35 idle / 47 load
92mm = 36 idle / 45 load
sys
80mm = 36
92mm = 35
pwm
80mm = 37
92mm = 35
Also interesting to note that with the 92mm, my temps return to normal (idle temps) quicker than with the 80mm. Plus with the 80mm, temps seem to jump around a lot easier. For example I get a 1 degree increase when i use my tv tuner with the 80mm, but not with the 92mm.
Anyway, it's true the 80mm's cool the cpu more effectively, but i like how the 92mm keeps everything else around the cpu cool like the PWM, NB ,and even RAM. I am gonna stick with 92mm----for now.
I would like to see someone test with two SF2's in 80 and 90mm...