MoDT Mismatch: AOpen i945GTt-VFA & Silverstone LC-12

Cases|Damping | CPUs|Motherboards
Viewing page 8 of 9 pages. Previous 1 2 3 4 5 6 7 8 9 Next

THERMAL & ACOUSTIC TESTING

Aside from the i945GTt-VFA and the LC-12, only three other components were needed to complete the system:

Windows XP Pro SP2 was installed, and our usual gamut of software tools installed:

  • SpeedFan 4.31 for CPU and other hardware monitoring.
  • Intel Thermal Analysis Tool for processor stress testing, thermal monitoring, and throttle monitoring.

Other tools:

TEST RESULTS

Ambient conditions were 19°C and 18 dBA. It is Winter, and the lab was snowed in under about 30 cm of snow when the testing was conducted. This made for unusually cold, but also unusually quiet ambient conditions.

AOpen i945GTt-VFA & Silverstone LC-12 Test Results
Load Condition
CPU Temperature
(via SpeedFan)
CPU Core
Temperature
(via Intel TAT)
System
Temperature
HDD
Temperature
Fan Speed
System
Power Draw
Noise Level
Idle
47°C
(49°C w/o
SpeedStep)
33°C
(38°C w/o
SpeedStep)
59°C
38°C
2700 RPM
25W
23 dBA@1m
Load (Intel TAT)
106°C*
98°C*
80°C
50°C
3600 RPM
63W
25 dBA@1m
Load (Cover Removed)
79°C
73°C
59°C
31°C
3600 RPM
57W
24 dBA@1m
HDD Seek
47°C
33°C
59°C
40°C
2700 RPM
27W
27 dBA@1m
*CPU throttled at these temperatures.

Testing the i945GTt-VFA and the LC-12 together was an exercise in steady nerves. This system ran uncomfortably, even painfully hot, despite the wintery conditions in our lab. If your gut instinct is to shut things down as soon as the processor approaches 50°C, this is not the board for you; according to Speedfan (which reported a different sensor than Intel's Thermal Analysis Tool), the CPU idled at 47°C. This isn't a P4 Prescott we're talking about. This is a mobile processor that idles at ~1.5W. The "system" temperature (which we suspect is actually the Northbridge) idled at 60°C!

And, the worst part of all this? The total power draw was just 25W. Folks, the only systems we've measured with less heat output than that are VIA's EPIA based systems. Even the iMac couldn't match such a low power draw. When you have a 25W system that runs at 50°C in idle, your thermal engineers have done something seriously, seriously wrong.

Not surprisingly, the system throttled to keep cool under full load, even with the CPU fan going full tilt. What was surprising was how long it took to throttle. The CPU remained stable and fully functional right up to 98°C. We'll have to rethink our ideas about what a safe CPU temperature is; Intel's new Core-series chips appear to be more heat tolerant (or perhaps just report higher temperatures) than the old Netburst-based ones. Whether or not the CPU is built to take such a thrashing, other components are not, and the 80°C "system" temperature and the 50°C hard drive were unacceptable.

After our nerve-wracking experience at 100°C, we were surprised to find that the aluminum casing of the LC-12 was hot to the touch. When we popped the cover off and felt the drive tray, it was too hot to touch for more than a second or two. Ordinarily, the case material plays very little role in exhausting system heat, but that was clearly not the case this time. Presumably, the lack of a system fan and the tiny exhaust vents meant that heat had to use whatever means necessary to dissipate, including conduction via the aluminum shell.

All this pointed to the case as the culprit for the high temperatures, so we ran the test again without the cover. This dropped the CPU temperature by about 25°C — still high, but no longer dangerous. More significantly, the system power draw dropped by 6W, indicating that the power circuitry was no longer headed towards thermal runaway.

Thermal performance nonwithstanding, the tiny amount of power sipped up by the i945GTt-VFA was impressive. When the board was running properly, the 57W load was probably lower than any other Intel or AMD-based system we've tested, though exact comparisons are difficult to make due to differences in stress software (vs. the Shuttle X100) and in components (the LCD in the iMacs).

From a noise perspective, things turned out well enough. Despite the small heatsink and the buzzy fan, AOpen surprised us with one of the best, most conservative fan controllers we've ever encountered on a motherboard. Changes in fan speed were very, very gradual, and could not be detected by ear unless listened for specifically.

There were hints that the fan controller wasn't as simple as a thermal threshold. When the system was initially powered up, the fan spun at full speed for about five minutes before dropping slowly to minimum speed, where it stayed until the system was pressed hard. We couldn't identify a positive threshold temperature when the fan speeded up, but it seemed to start increasing when the CPU Core was somewhere between 70~80°C.

The acoustic difference between minimum and full speed was not large, which probably helped contribute to our perception that changes in fan speed were difficult to hear. And, given how fast the fan was spinning (2,700~3,600 RPM), the 23~25 dBA@1m noise measurements were actually pretty good. The noise was acceptably quiet no matter what the speed, although it would never fall into the inaudible category that the best systems achieve. Our biggest complaint was that the noise character was unpleasant, with lots of buzz and whine that was far from smooth.

The worst noise from the system actually had nothing to do with the fan at all; seek noise from the hard drive was the most disruptive noise we heard. The 27 dBA@1m is about accurate for what we heard. The seeks were sharp and clearly audible above the rest of the system noise. That's very loud for a supposedly "quiet" notebook drive, and much louder than the 20 dBA@1m seek noise that we heard when we did our original review. It was the effect of the aluminum box amplifying the vibrations from the hard drive.



Previous 1 2 3 4 5 6 7 8 9 Next

Cases|Damping - Article Index
Help support this site, buy the AOpen i945GTt-VFA: mini-ITX Core Duo / Core 2 Duo Motherboard from one of our affiliate retailers!