Tweeter Testing - Part 3

Case study of alternate method.

A friend sent some tweeter test charts for my information which is useful to illustrate an alternate approach. He has kindly given consent for its post here. The tweeter is Indonesian branded with label erased to maintain anonymity.

The measurement is made by an audio sound analyzer. In essence, it is a frequency response plot except a bird sound profile is used here (vs a pure sine wave test signal as in lab test). In addition it provides measurement of sound level with respect to its frequency content.

This is an alternative in what I regard as a "next best" test. Such method are normally employed by sound engineers in the field to sample sound quality at different spots in auditoriums, stadiums, event stage etc.




















Reading the chart:

The horizontal axis of the graph is in logarithmic (log) scale. This is common in frequency response plot where incremental step (of equally spaced vertical scale lines) along the x-axis means 10 times increment in frequency. e.g. 20 Hz, 200 Hz, 2KHz, 20KHz etc.

The vertical axis is the measure of SPL (sound pressure level) in dB, which is also logarithmic in nature. 0 dB is about the threshold of human hearing, that of a mosquito flying 3m away! Can you hear that? I can't. Each 20 dB increment is 10 times louder (rather similar to Richter Magnitude scale for earthquakes), 40 dB is 100 times, 60 dB is 1000 times etc.

The vertical bars (in lighter shade of blue) are the frequency components of the sound profile.

Finally the solid line (brownish?) is the profile of the measured sound. Be aware it is the combination of the tweeter & amplifier responses together.

So this is a comparative test between the "envelop" of the sound source and the measured response - not truly of the tweeter alone.

The circled area on graph highlights gross mismatch in performances.

"Apparent" frequency response between 20Hz to 1KHz are relatively flat whereas there are peaks in mid-region of the sound profile of the same range.

As the lowest frequencies used by the swiftlets in echolocation are reportedly between 1KHz to 6KHz, you may question why are there frequencies lower than 1KHz in the bird sound profile? Do the birds create them?

No, in my opinion. This require some understanding of signal theory & Fourier Transform analysis to explain (beyond the scope of discussions here).

Let me try to put in lay terms. The time intervals between echolocation calls itself is a frequency component in the recorded signals. There are random variations between calls as one can expect in a sound profile. So these are not acoustic waves in swiftlets sound per se.

The frequency plot is an analysis of the time signals by Fast Fourier Transform computations. Any non-ideal sine waves will have its range of harmonics generated too.

The bird sound profile is a complex sound with it frequencies content changing all the time. So the analyzer only register the peaks - that forms the envelop. Hope it is evident here why a pure sine wave test signal give a much more objective measurement.

What's more - the tweeter seem to reproduce these frequencies too - note readings just below 60dB on graph! Not quite true. This is likely the lowest measurement limit (or sensitivity) of the instrument though part of it could be ambient noise level which may be disregarded. Note the higher sound levels are > 90dB, almost 40dB (100 times) higher.

In general this gives an overall good picture how "faithful" is the reproduction of the "original" sound. One has to interpret the result with discretion.

This is a still good method to match a sound profile for tweeter selection. On the other hand if you have a tweeter measured with adequate frequency response, it will match any sound profile within the frequency range.

Hope I provide here an understanding to the relative merits of the two methods.