40mm Fan Shootout: Part 1 – Noise

Introduction:

Keeping electronics cool is an essential to their longevity. Most will agree that passive cooling is insufficient for the MiSTer, but many disagree on the best solution for fans. Some find generic 40mm 5V fans to be too noisy and opt for the premium Noctua NF-A4X10-FLX 5V fan. Others find that running a generic 40mm 12V fan at 5V is a nice balance between airflow and noise.

This shootout consists of four parts:

  • Part 1 – Noise
  • Part 2 – Cooling
  • Part 3 – Airflow
  • Part 4 – Conclusion

From the beginning, I’ve opted to use generic 5V fans as they’re cost effective and also provide good cooling, at least from what I could tell. They get a little noisy at times, but “noisy” is subjective. I decided to conduct an experiment to test my hypothesis.

Test Methodology

From left to right: (Top row) generic 5V, Noctua 5V, generic 12V, Scythe 12V / (Bottom row): M3 nuts, M3x14mm bolts, rubber washers, rubber standoff feet

40x40x10mm Fans:

Measurement Equipment:

  • Dayton Audio iMM-6 Calibrated Measurement Microphone
  • Decibel X Pro for iPad

Dampening configurations (see above image):

  • No rubber: DE10-Nano sitting on stock brass female standoffs
  • Rubber Washers: Rubber washers placed between fan and IO board
  • Rubber Feet: DE10-nano rubber feet placed on bottom standoffs
  • All Rubber: Both Rubber Washers and Rubber Feet

Setup:

Yes, I know the IO board isn’t fully populated.
  • Home alone (is an amazing movie)
  • Air conditioning and heat turned off
  • All other fans from computers, etc., turned off
  • Measured ambient (noise floor) at the beginning and end of the test
  • Folding plastic table placed on carpet with DE10-nano placed on top
  • No additional noise dampening
  • iPad with Dayton Audio microphone suspended 10 cm above the fan at roughly a 10 degree angle from perpendicular
  • One sample of each fan type tested over a period of 10 seconds, after fan speed had reached a stable level.

A Note About Sound Measurements:

The data collected by the Decibel X Pro app is stored as a CSV file of 200 ms average sound pressure level (SPL) in dB, specifically power ratio, not amplitude ratio. These were imported into Google Sheets and plotted.

SPL is measured using the logarithmic scale. Humans don’t perceive sound in a linear manner, meaning that if sound B is twice as loud as sound A, it will be 6 dB louder (SPL_B = SPL_A + 6).

Results:

A lot of data
An average of all four dampening configurations for each fan
An average of all four fans for each dampening configuration

Analysis and Conclusion

  • Any kind of dampening is better than none
  • Simply adding rubber feet to the bottom standoffs lowered the average from 40 dB to 37.3 dB, a reduction of 2.7 dB, or 53%.
  • The Noctua fan isn’t as silent as people make it out to be, but it does come stock with rubber dampening on all four corners of each side to minimize vibration. Anecdotally, it feels like it moves a lot of air. Time will tell.
  • Both 12V fans are very quiet, but how well do they cool?
  • The generic 5V fan is the loudest, as predicted.
  • Using multiple samples of each fan model would help provide more accurate results, but it’s outside the scope of this experiment.