Studio Update: The main Digital Audio Workstation at Studio J is presently a Gateway 700X. Ever since its arrival many years ago, its been a source of extra studio noise. After much research about quieting PCs (see
SilentPCReview and
EndPCNoise) and given that a new system is just a bit out of reach, we decided to try picking up a few tools to quiet it down. This computer has a Pentium 4 processor with a noisy little 60mm fan attached to a heatsink which Gateway recommends NOT removing. Well, for a mere $66, we were able to pick up a couple of 92mm fans, a 120mm fan, and a 60mm to 92mm adapter. The 92mm fans are
Noctua NF-B9
and the 120mm fan is a
Noctua NF-S12B ULN
. These are packaged with soft fan mounts and Ultra-Low-Noise Adapters (3-pin in-line resistors). The fans are rated to be very quiet and have self-stabilizing oil pressure bearings.
First to be removed were the old 60mm AVC (ball bearing) fan and the 92mm Sunon (sleeve bearing) fan on the power supply. Next, the 92mm Noctua fan was connected to the 60mm fan adapter (available through
SVC). The assembly was then attached to the heatsink, using a screwdriver carefully between the fan blades. The other new 92mm fan was attached to the old power supply, and the 120mm fan was placed on the inside back of the case (blowing out of course).
During the initial test of the system, there was a significant reduction in overall noise...notably, the elimination of the 60mm fan noise. After a little sleep, it became clear that there was one further avenue worth exploring - that extra fan controller on the motherboard (mobo) which Gateway says it wasn't using. Well, the CPU cooler was specified to be plugged into the fixed voltage pins next to the processor, but we discovered that the extra fan controller over by the memory sticks can indeed be controlled by
SpeedFan. BINGO! We installed the ULN adapters on the PSU and case fan which run at fixed speeds fairly quietly. Meanwhile, the processor cooling fan is now plugged directly into the extra fan controller. The lowest set point in SpeedFan is 20% to start the fan and keep it turning. On heavy load, the fan will speed up to 100%. According to SpeedFan, the sensor is indicating 570 RPM at idle and 600 RPM at full tilt but these seem low since you can hear the fan when running faster. (BTW, the
Terragen renderer is a great tool for loading your processors.) The CPU temperature runs between 37
°C (20% fan) at idle and 50
°C (100% fan) loaded. Prior to the modification, the CPU ran under 30
°C at idle.
A little rewiring, and behold, a nearly silent machine. Doing another ambient recording of the computer and comparing to the original spectrogram indicates about a 15dB reduction across the spectrum. The red line in the diagram is the original ambient noise from the unmodified machine. The blue line is the modified version. The yellow is the background noise measured outside of the studio and away from the unmodified machine.
A note about the disk. This Gateway 700X came with an EIDE 120GB Western Digital drive. After learning about how the computer case acts as a loudspeaker, amplifying the sound of the disk drive, the old 3-bay 3.5in containment structure was removed and replaced with a smaller, more compact 3-bay structure and placed on the floor of the case, seated on a bed of bubble-wrap and rubber. There is still an audible whine of the 7200 RPM drive, so the next upgrade will likely be a SATA controller card and quiet drive.
Photos:
before and
after.
Update: Turns out that the CPU Fan socket is providing the RPM reading, so this is the speed of the case fan which is plugged into that socket. The controllable Fan1 socket is apparently not providing the RPM reading. When the CPU fan speeds up, its proximity to the case fan affects the RPM slightly. Also, the motherboard is wired for additional fans, but Gateway or Intel did not install the extra sockets.
, hoping for a single platter and low noise. Used the Acronis True Image software from the Western Digital site to clone all of the data from the old 120GB disk after updating the BIOS first so that the full disk space could be recognized. Pulled the old disk out and made the new disk the primary and did another sound reading of the system from 1 meter. Fairly quiet now. Below is the spectrograph as it stands today. The blue line is the current machine noise level. The yellow line is a measure of noise outside of the studio, away from the original unmodified machine. The red line is the noise level of the pre-modified Gateway 700X. Overall, it appears that the changes have yielded about a 20dB drop across the spectrum. Much more tolerable, since now there is only a hint of moving air.
