Viewing page 1 of 4 pages. 1 2 3 4 Next
November 2, 2002 by Mike Chin
We have two modest thermal-control fan heatsinks for review: the Super Silent Pro TC by Arctic Cooling of Switzerland, courtesy of Silicon Valley Compucycle (SVC), a retailer located in San Jose, CA, and the Spire 5U213C1H3G (snappy name, isn't it?) marketed and submitted for review by the Fanner Group, heaquartered in the US. Both are inexpensive, and the level of workmanship and finish are fine, but at a level considerably below that of the recently reviewed heavyweight heatsinks.
Thermally controlled fans on CPU heatsinks are not highly visible but they are much more commonplace than most people think. The standard HSF that comes with the Intel P4 is thermally controlled. Yes, its fan speed varies 2000~3000+ RPM in accordance with temperature. If this comes as a surprise, rest assured that you're not alone. Most people are unaware of this fact, partly because Intel hasn't really made any noise about it, and partly because there is nothing on the outside of the fan to show that it is thermally controlled. The thermistor appears to be inside the fan rather than protruding anywhere visibly, as is usually the case.
So why is there a thermostat between the two heatsinks? Because that's kind of like what these heatsinks feature. The difference is that the temperature setting for the heatsinks is preset by the manufacturer.
The basic concept is sound: vary the fan's airflow in accordance with real need by tying its speed to a thermal sensor that monitors CPU heat. It is a feedback loop much like that in a home or office air conditioning system. But have you noticed how some air-conditioned environments get too hot before the cooling starts or get too cold before the cooling stops? Hysteresis is a word that could be used to describe this phenomenon. It's a lag between stimulus and response in a mechanical system. Those uncomfortable air-conditioning systems are too slow to change in response to stimulus, the rising or dropping temperature, and they overshoot their mark.
A similar phenomenon has been reported in a handful of web reviews on thermal fan CPUs: the fan speeds up too slowly, and it slows down too slowly. Whether this is a problem is worth considering.
The objective of thermal control is usually to keep the fan spinning at the lowest speed necessary in order to maintain an acceptable temperature, thus keep the fan noise at a minimum. Definitions vary about what is low noise and what is acceptable temperature. We'll see that the Arctic Cooling Super Silent Pro TC by and the Spire 5U213C1H3G take different approaches to thermal fan control, and get quite different results.
How to Test Them?
Before we get down down to the nitty gritty, there is the question of methodology. We detailed our standard approach in an article about our unique heatsink testing methodology. It calls for all HS to be tested with the same low power 80 mm Panaflo fan, thereby removing the fan as a variable that affects cooling results. The approach allows each heatsink's intrinsic cooling power with a low noise fan to be accurately determined.
With each of the HS in hand, the fan is part and parcel of the design. Presumably, the characteristics of each fan and its thermal control is matched to the heatsink. Neither are high-end products for which a separate fan would be purchased by the vast majority of users. They could easily be generic OEM heatsinks. For these reasons, the standard SPCR methodology is left on the shelf this time. Instead, each heatsink is examined as a complete HSF package. The questions asked are:
How much noise does this HSF make under normal and high load conditions?
Does the dynamically changing aspect of the fan noise make it more audible or obtrusive?
How effectively does it cool?
|Help support this site, buy from one of our affiliate retailers!|