smyslov is the Sun Blade 100 that I have been running as an OpenBSD firewall / router since last November.

Unfortunately, one of the bad things is that it’s loud. Inspired by methods such as this, I took apart smyslov and added sewing machine oil to the two primary fans in an attempt to quiet them.

The process is fairly straight forward:

Applying the sewing machine oil

Take the case off:

Locate and unscrew the power supply from the main case:

Unscrew the four screws in the top (including the one under the HIPOT OK):

Unscrew the power supply fan:

Take off the sticker, exposing the metal:

and drip in one drop of sewing machine oil.

The same technique applies to the case fan at the front of the chassis:

Measuring the results

I set up a USB microphone about 20 cm away from the back of smyslov, and recorded about 12 seconds of before and after noise to see whether or not this oil application would make a difference:

Data generated with sox:

sox ~/sun_blade_100_noise_before.wav sun_blade_100_noise_before.dat
sox ~/sun_blade_100_noise_after.wav sun_blade_100_noise_after.dat

Then gnuplot was brought in:

set terminal png nocrop size 600,600

set output "sun_blade_100_noise_comparison.png"

plot 'sun_blade_100_noise_before.dat' every 100 using 1:2 with impulses title 'Before',
'sun_blade_100_noise_after.dat' every 100 using 1:2 with impulses title 'After'

This comparison is unscientific and probably wrong but seems to show a decrease in amplitude in the ‘after’ waveform. Unfortunately the subjective noise is still mostly the same…