Hi, great forum I found here! Lots of great information to build silent yet powerful rigs!
Here is the problem, and I must begin by apologizing for the long post...
Well, I got this weird fan behavior when I installed my new Opteron 185 with the boxed cooler. With the old AMD64 300+ everything worked perfectly, a very silent system.
Problem can be described shortly like this: the linear increase of cpu fan rpm vs temp that I expected from ASUS Q-fan bios fan control option (A8N-SLI dlx) somehow is disrupted with this cooler, a rather nice heatpipe-based unit with copper base btw.
Longer explanation, here I have to use a few example scenarios with some data:
*** Temps and rpms at idle - vCore 1.10 Volts (PowerNow activated)
a. Low room temp, 22C (72F)
- cpu temp 31C (88F) --- this is apparently as low as it is allowed to go by asus Q-fan (?)
- core temp 29C (84F)
- cpu fan: continuously varying rpm < 3000: like 1000, 1500, 2500, 500, even zero rpm (too low to detect, the fan is still kept in motion by short weak bursts/pulses)
- MB temp 31C (88F)
*Note. When the cpu reaches 32C (90F), the rpm goes up to 3000 rpm for like five seconds. Very irritating fan behavior, the fan is almost inaudible below 2000 so it sounds like it is turned on and off all the time.
b. Slightly higher room temp, 24C (75F)
- cpu temp 32C (90F)
- core temp 29C (84F)
- cpu fan: 3000 rpm solid
- MB temp 34C (93F)
c. Higher room temps, up to 28C
- temps like above, only a few degrees higher
- cpu fan: 3000 rpm solid
----
*** Temps and rpms at ~50% load, room temp at ~26C (79F)
- cpu temp 40 - 45C (104 – 113F)
- cpu fan: ~4000 rpm
- MB temp: 35C
----
*** Temps and rpms at full load both cores, room temp at ~26C (79F)
- cpu temp: 58C (136F)
- cpu fan: 3000 rpm (yes, back to rpm @ idle!)
- MB temp: 36C
--- --- ---
*** Case and ventilation ***
- AOpen H600B, comes with an air duct that supplies room-temped air directly on top of the cpu cooler
- a 120mm outblowing fan regulated by the MB to 1200 rpm typical, slightly more under heavy load, down to 900 when starting cold
- PSU with 120mm fan, I guess at ~1200 rpm typical load (very silent)
- no intake fans, except for a very slow spinning large fan on the side air intake over the pci slots to produce a slight breeze over my heatpipe-based chipset and graphics coolers (Thermalright HR-05-SLI, and Thermaltake Schooner)
When I remove the side cover under full load (and the cpu air duct), the cpu fan responds within seconds, producing like 4000 – 4300 rpm, cpu temp stabilizes at 60C (140F) or slightly above. This is good of course, also natural since the cpu fan now draws hotter air.
Can someone please explain why the CPU fan seems so uncontrollable? No matter what I do, turning down target temp from default 54C to 51C (124F) in bios or attempting to use Speedfan, nothing helps. It is like that fan is living a life of its own.
An interesting clue perhaps: if I adjust the fanspeed manually in Speedfan, I get a temporary effect that lasts a second or two before it adjusts itself back to where it was (software PWM modes are set). The chassis fan responds like it should though…
Another clue maybe: the stock cooler came with a 4-pin wire, my MB just have 3-pin headers. Could this have something to do with anything?
Yet another clue: when I plug the fan directly to an unregulated chassis fan header, the fan behaves much calmer, staying at 3000 rpm until beginning to spin up at around 60C. This is still too much noise for no reason at lower loads, also not satisfactory considering the lower target temp I want. It is a temporary solution right now.
In summary then:
- at low cpu loads and low temps, I’d like significantly lower rpm since I have no use of such low cpu temps
- at high loads and temps, I’d like the cpu fan to spin up when the target temp is reached
The cpu fan is by far the noisiest part in this system but I feel this should be possible to do something about without changing cooler, it seems very efficient in this system.
Sorry for this longwinded post but it is a little difficult to explain this odd problem in just a few words… thanks for reading and for any clues you can provide!
A8N SLI // Opteron185 - Weird cpu fan behavior!
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Sorry I forgot...
I currently have the boxed cooler connected to an unregulated chassis fan header that supplies 12V, hence I expected the fan to spin up to at least 4000 rpm which was observed earlier under 50% load and regulated by Asus Q-Fan.
Since it just spins at 3000 rpm when connected to 12V, and spins up slightly "by itself" at cpu temps over 60C, it is evident that the stock Opteron cooler has some kind of built-in temp control.
I guess this is what screws things up for me, problem is how can I correct this design blunder as easily as possible?
The fan is not designed to come off easily for a fan swap to a known good fan, the fan is sort of integrated into a molded plastic bracket on top of the sink that probably will break if I attempt to remove it. I could of course strap on a normal 92mm fan, connect it to the regulated cpu fan header and see if that is improving things.
Does anyone know if the integrated fan on the AMD heatpipe-style coolers (as they use on most dual cores) is possible to remove in a non-destructive way, any tips?
*** Update ***
The fan turned out to be very easy to remove without damage to anything.
I replaced the temp controlled 70 x 70 x 15 mm fan with soething more suitable: a dirt cheap Sunon 2.3W / 0.2A 80 x 80 x 25 mm, $5.
Got stunning results with this excellent little heatsink. How about this, as measured by AsusProbe 2.
- 28C room temp, Asus QFan controls the cpu fan speed with a target temp of 60C:
Idle, Vcore 1.10 (PowerNow activated)
- cpu: 33 C
- cpu fan rpm: 1300
Full load both cores
- cpu: 57 C
- cpu fan rpm: 2300
* Note: this is a deliberate compromise, I could have chosen a lower target temp at the price of a higher cpu fan rpm, or a higher target temp which would make the fan spin slower.
This is very promising since it leaves some headroom for even higher room temps, this particular fan has a max rpm of 2700. It could all be fine-tuned by speedfan of course but this is definitely good enough for me and the machine for now!
The fan is attached to the sink with cable ties. I also hotglued a couple of plastic "curtains" on the sides of the fan body in order to guide the flow through the heatsink fins, this is since this larger fan rests higher on the sink and leaves some ineffective gaps otherwise.
All in all, a very quick and easy solution. Sound-wise, it is like a dream came true: the sound at full load is much less than at idle with the stock fan. At idle, the sound is now equal to what it was with the old 3000+ Winchester which means close to complete silence.
I must conclude that the stock heatpipe cooler is extremely efficient, provided it is paired with a suitable fan and control system.
A nice side effect is that this system is no longer at risk of overheating, ever.
Case closed.
I currently have the boxed cooler connected to an unregulated chassis fan header that supplies 12V, hence I expected the fan to spin up to at least 4000 rpm which was observed earlier under 50% load and regulated by Asus Q-Fan.
Since it just spins at 3000 rpm when connected to 12V, and spins up slightly "by itself" at cpu temps over 60C, it is evident that the stock Opteron cooler has some kind of built-in temp control.
I guess this is what screws things up for me, problem is how can I correct this design blunder as easily as possible?
The fan is not designed to come off easily for a fan swap to a known good fan, the fan is sort of integrated into a molded plastic bracket on top of the sink that probably will break if I attempt to remove it. I could of course strap on a normal 92mm fan, connect it to the regulated cpu fan header and see if that is improving things.
Does anyone know if the integrated fan on the AMD heatpipe-style coolers (as they use on most dual cores) is possible to remove in a non-destructive way, any tips?
*** Update ***
The fan turned out to be very easy to remove without damage to anything.
I replaced the temp controlled 70 x 70 x 15 mm fan with soething more suitable: a dirt cheap Sunon 2.3W / 0.2A 80 x 80 x 25 mm, $5.
Got stunning results with this excellent little heatsink. How about this, as measured by AsusProbe 2.
- 28C room temp, Asus QFan controls the cpu fan speed with a target temp of 60C:
Idle, Vcore 1.10 (PowerNow activated)
- cpu: 33 C
- cpu fan rpm: 1300
Full load both cores
- cpu: 57 C
- cpu fan rpm: 2300
* Note: this is a deliberate compromise, I could have chosen a lower target temp at the price of a higher cpu fan rpm, or a higher target temp which would make the fan spin slower.
This is very promising since it leaves some headroom for even higher room temps, this particular fan has a max rpm of 2700. It could all be fine-tuned by speedfan of course but this is definitely good enough for me and the machine for now!
The fan is attached to the sink with cable ties. I also hotglued a couple of plastic "curtains" on the sides of the fan body in order to guide the flow through the heatsink fins, this is since this larger fan rests higher on the sink and leaves some ineffective gaps otherwise.
All in all, a very quick and easy solution. Sound-wise, it is like a dream came true: the sound at full load is much less than at idle with the stock fan. At idle, the sound is now equal to what it was with the old 3000+ Winchester which means close to complete silence.
I must conclude that the stock heatpipe cooler is extremely efficient, provided it is paired with a suitable fan and control system.
A nice side effect is that this system is no longer at risk of overheating, ever.
Case closed.