A loudspeaker for Anders

The two drivers may be combined to produce a good speaker with many different forms of crossover topologies. Again, there is no “best” solution. Various crossovers, all producing flat electrical response, may sound differently. Although highly acclaimed speakers usually have flat SPL response, the flat SPL response does not in itself guarantee good sound. Thus, we have to consider various solutions and, of course, perform listening tests to evaluate and verify that we are heading in the right direction.

Fig. 24 shows the far-field SPL response of the woofer in the box, but without any crossover. The response rises with about 6 dB from 30 to 150 Hz and then evens out. This slope is due to the EBS design, as evident from Fig. 8. From 400 and up to 1,000 Hz there is a 3 dB rise due to cabinet diffraction. Above 3,200 Hz the response steeply drops off to a sharp dip at 5,225 Hz. The cone resonance peak is really serious at 6,600 Hz.

The off-axis response shown by Fig. 20 suggests a crossover frequency of 2,000 Hz or lower. The very high cone resonance peak shown in Fig. 24 suggests we should select a crossover even lower, say 1,600 Hz (two octaves below the peak). Since this would require a much extended tweeter range, we have to find a compromise. Let us therefore examine a crossover frequency of 2,000 Hz.

The woofer crossover should compensate for the rising response above 100 Hz and then attenuate as fast as possible above 2,000 Hz. Thus, a fourth-order low-pass SPL target response would be necessary to sufficiently suppress the cone resonance peak.

A first cut third-order low-pass filter, using not too uncommon component values, was created, see Fig. 25. It is based on an ordinary Butterworth design, but the inductor values are adjusted to add a tilt between 200 and 1000 Hz. The first inductor is paralleled with a resistor to lower its Q. Otherwise there would be a peak at the crossover frequency. The amplitude response of this filter is shown in Fig. 26.

Fig.27 shows the simulated woofer response with the above low-pass filter. The figure also shows the fourth-order Linkwitz-Riley target response. Obviously, the downward tilt between 200 and 1000 Hz is somewhat exaggerated. The cone resonance peak is attenuated with 30dB. I would suggest this is too little, and that a notch filter should be considered. But let us leave that for the moment.

Now let’s see what a little optimization can do for us. The now optimized filter component values are shown in Fig. 28. Fig. 29 shows the fourth-order target response and the simulated woofer response. The downward tilt is now removed and the response is reasonably close to the target for frequencies above 150 Hz.

Even if the woofer response is close to the target above 150 Hz, it is sloping off with about 6 dB down to 30 Hz. This is a result of two factors; diffraction loss and the EBS design. Although I don’t consider it very wise to add a curve shaping section to the filter, the response below 150 Hz can be straightened out a bit. There is a pronounced risk, however, of introducing too large cone excursions at low frequencies. In addition to this there is a price to be paid; 6 dB of gain, which is quite a lot.

[ previous ] [ back to top ] [ next ]