Sub-$20 CPU Coolers: A Reader's Roundup

SPCR MEASUREMENTS & ANALYSIS

The use of the Zalman Fanmate fan speed controller imposed a limit on the top speed of all the tested fans. The maximum voltage/speed for each fan was not the standard 12V, but something close to 11V. We've measured 10.8V in the past. The Fanmate has a ~1.2V DC loss when its control is set to max. This means most of the coolers' fans can run a bit faster, make a little more noise while doing so, and cool slightly better at top speed.

The QUIET fan speeds listed here are somewhat arbitrary. They are the speeds at which the coolers became quiet enough to be masked by the hard drive and power supply noise in Andrew's system. This is a reasonable marker for "quiet" for non-obsessive PC users seeking to reduce noise. Most users would be pleased with the noise level of almost any of the coolers at the QUIET test RPM, and most would be unhappy with the MAX speed. It's possible that many would choose a heatsink/fan speed somewhere between the two tested points.

As a group, the SPL at minimum speed falls in a tight group between 12.5 and 15.5 [email protected], a spread of just 3 dBA, which vindicates Andrew's subjective "quiet" speed choices. Even the highest QUIET level (15.5 BA) is very quiet, generally below ambient acoustic conditions in most places. There's much higher noise variance at the MAX speeds, as the RPM itself varied much more, from a low of 1630 to a high of over 3000. All in all, Andrew's subjective comments on the sound are fairly well supported by the SPL measurements (though you can also judge for yourself by listening to the recordings described below).

Finally, recordings were made of all the coolers. These MP3s are linked in the table below. Each recording consists of 7 seconds of silence in SPCR's anechoic chamber (10~11 dBA), followed by 10 seconds at the QUIET setting, and 10 seconds at the MAX setting. This allows you to judge the noise qualitatively. Keep in mind that our hyper-sensitive microphone & recording system lets you hear every detail and nuance of the noise, and if you turn the volume of your player up, it will be very exaggerated. The best way to listen is to use the first 7 seconds to set the volume so that hiss becomes just audible, then back it down a touch to make it inaudible. Then leave the volume setting untouched while listening to all the MP3s.

There is no easy way to compare this data to SPCR CPU reviews, as we use considerably hotter, completely different platforms, and a different testing procedure. Even with different samples of the same core gear, there would be differences in temperature sensors and actual power/thermal characteristics. Having stated all that, Andrew's test platform is closest to our small heatsink test platform.