Home / Opinions / A Silent Solution / How We Did the Tests

How We Did the Tests

Because of the complexity and cost involved, measuring the sound power levels of the test components in this feature was out of the question. Instead, SPL measurements were made using a Bruel & Kjaer Type 2218 precision integrating sound level meter and Type 4165 condenser microphone. Although our noise measurements cannot be compared directly with any of the manufacturers’ quoted figures, the relative benefits achieved by using one or more of the noise-reducing PC components can be assessed reasonably accurately.

The high-quality B&K Type 2218 sound level meter

To verify its overall accuracy, the sound level meter was sent to an approved calibration agency in January 2004 to undergo a full calibration. In addition, an ‘on-site’ calibration check was performed on the meter immediately before testing using a Bruel & Kjaer 4231 acoustic calibrator, which produces a reference signal of 94dB at 1000Hz. A further calibration check was performed at the end of the tests to check for any instrument drift.

The B&K Type 4165 condenser microphone

Noise measurements were carried out with the PC case placed on the floor in the middle of a room, and at least 1.5m from the room boundaries in order to minimise reflections as much as possible. The microphone was orientated to point directly at the computer case and positioned approximately 30cm off the carpeted floor so that it corresponded closely with the heights of the CPU fan and the rear-mounted case fan.

To ensure a consistent noise level from the hard disk, the power-saving features in the OS were disabled so that the internal platters kept spinning throughout the tests. An audio CD was also played continuously in the CD-RW drive to keep any noise from the drive motor and bearings at a constant level.

For each set-up under investigation, the A-weighted Leq noise level measured over an averaging period of one minute was determined separately at the front, rear, left and right sides of the PC case at a nominal distance of 1m. The noise was measured at all four sides to check for directionality of the sound.

When interpreting these results, you should bear in mind that measurements taken under different conditions, for example with the case placed on a desk, next to a wall or on a hard floor, or with a different microphone orientation will almost certainly produce different answers. Each set of measured noise levels in this feature should therefore be viewed in a comparative sense rather than as absolute values.

Another important factor that can influence the accuracy of any measurement is the level of the background noise, which should be as low as possible so that it doesn’t drown out the sound you are interested in. The minimum difference that is acceptable is about 3dB(A) but ideally the background level should be at least 10dB(A) below the sound you are trying to measure.

Throughout the tests the background noise level in the room was so low that it could not be measured by our meter, which can only measure down to 25dB(A). Because we cannot be absolutely certain what the background level was during the tests, a maximum value of 24dB(A) has been assumed. This means that the accuracy of our noise measurements can only be guaranteed if they are above 27dB(A).

Finally, the LCD display on the sound level meter reports the measured noise level with a resolution of 0.1dB. However, even with high quality instrumentation like ours, that level of accuracy cannot be achieved in all but the most carefully controlled test conditions. For that reason, all noise measurements have been rounded to the nearest whole decibel.

comments powered by Disqus