Measurements that matter

In yesterday's postI suggested you can't hear the difference between distortion levels that are already below audibility. That sparked a few questions about what you can hear and why. Measurements sometimes help us understand an underlying change that is not exactly relevant to the actual measurement. For example, measuring the THD on the power line is rather meaningless unless you use that measurement to understand the underlying cause of the change you are measuring. So this begs the question we get asked a lot "why does our company place a THD analyzer on the front panel of your power products if the measurement it provides is meaningless?" The answer is simple: it helps us market the underlying change in the power we deliver that IS relevant. In this case, the added harmonics we measure on the power line are there as a result of the AC power wave being corrupted. Once the equipment corrects the corruption we measure fewer harmonics. The improvement you hear, however, isn't because there's fewer harmonics (which is what the analyzer is measuring) but because the wave is no longer corrupted. We are interested in the cause, not the effect we measure. In a similar vein, think about THD levels in an audio circuit. The measurements are relevant (below a certain point) not because there are fewer harmonics present (as in our last example), but rather as an indication of what the designer DID to lower them. Perhaps he used more or less feedback, made the devices more linear, rolled off the upper harmonics, etc. The only things relevant here are WHAT the designer did - the measurements themselves only a crude indication loosely related to the change. We use these measurements on our equipment as a marketing tool to differentiate ours from theirs. Manufacturers have been doing this in spec sheets and one upmanship for years. For measurements to matter, you really need a broad understanding of the underlying changes being measured.