Let’s look at the role of the thermal engineer in determining the acoustic noise emission of an electronic system. By “acoustic noise” I mean the usual definition of unwanted sound. Readers can well imagine that the relationship between the thermal engineer and acoustic noise has to do with fans.

While it is no great mystery that fans contribute to a system’s noise emission, it is not necessarily a given that for a system to be quiet, it has to have no fans at all. During my time at InFocus, there were repeated calls to explore a “fan-less projector.” It was thought (for a time, anyway) that the only way to get a quiet projector was to eliminate fans from the design entirely. That is certainly one way to go to achieve a quiet system, but not necessarily the only way. There are other options, but of course they are not easy.

There are several key aspects to designing for low acoustic noise emission in fan-cooled products. These are:

• Agreement on a noise measurement standard
• A specific priority placed on low emissions from the beginning of the design process, and
• Knowledge about ways to keep emissions low.

Let’s look at these aspects in a bit more detail.

Noise Measurement Standard

The noise measurement standard is extremely important, because everyone involved should understand what measurements mean. Actually, this is true of thermal measurements too. A thermal target of “70” doesn’t mean much unless you know where, under what conditions, and how the temperature is to be measured. Similarly, just saying that “35 dB” is the acoustic-noise target doesn’t tell you anything. Is it sound pressure? Where is it measured? Is it sound power? There are many excellent sources of this kind of information on the web if you want to learn more. As far as I have been able to determine, there are some recommendations that most technical people in the field agree on:

• Sound power should be measured and specified in A-weighted bels.
• Sound pressure should be measured and specified in A-weighted decibels at one meter, according to ISO7779 or similar international specifications.

It is my understanding that the main reason to specify sound power in bels (instead of decibels) is to avoid confusion with the pressure. What’s the difference? Sound power is the total acoustic energy emission of the system. It’s completely independent of where the listener’s ear is, or what else is going on with the listener or the environment. (Maybe “listener” isn’t quite the right term – “hearer” describes it better, if awkwardly.) Sound power gives an objective comparison between one system and another, which is why it ought to be the measurement used for specifications.

On the other hand, most people don’t have a feel for magnitudes of sound power. In fact, human hearing doesn’t respond to the sound power scale. Actually, it doesn’t respond to the sound pressure scale either, as far as I have been able to tell. More on that later. But because people have been exposed to sound pressure numbers in dB, there is the tendency to specify acoustic noise emissions of equipment in those terms. But there are a lot of variations in the test conditions that can make data look good. For example, you can measure sound pressure at 1 meter away from the equipment, which is at least conforming to one of the measurement methods specified in an international standard. Or you could measure it at the operator or bystander positions. At least those positions relate to the equipment’s actual use. But I have seen data that related only to emissions from the quietest face of the equipment. That’s a strategy that clearly is intended to result in the lowest possible number in the “specsmanship” game, and doesn’t mean anything to the actual user of the equipment. So you can see that data sheets don’t necessarily give the conditions to which the sound pressure number corresponds, so you don’t know what you’re getting.

To make matters even worse, human hearing is highly variable and subjective. Is the hearer fatigued or stressed? Noise is more bothersome. Is the environment busy or noisy? System noise is less evident. Is the system going to be sitting right next to a microphone or telephone on a conference table? The other parties on a teleconference are going to hear it more than the people in the room. Are there specific tones apparent in the noise emission? Neither the power nor the pressure level gives any indication, and specific tones tend to make a system sound noisier that its instrument measurement would indicate, even if the standard test methods are used.

So the general consensus I have experienced is that although an objective measurement is essential for development and design work, it’s not quite enough to completely measure the goal of achieving a “quiet system.” People have to actually listen to it. The sound power is a necessary, but not sufficient, condition for determining whether a system is quiet. The customer base is ultimately going to decide whether the system is quiet enough, but it sure would be nice to have some advance warning before you put the product on the market!