Avtive bass filter

An active filter for bass amplifiers

This active filter was primarily designed to be used with the bass boxes presented on these pages. It can, of course, be used together with any suitable bass box.

The filter circuit encompasses an input buffer with gain, a high pass filter, a low pass filter, and a level control. Although this can be considered quite a number of functions, they can all be justified;

The buffer is there to give the filter input high impedance. Since the signal source for the bass is normally tapped from the input to the main amplifier, or at least close to it, it is vital that the bass filter does not put any load whatsoever onto the main amplifier input.

Bass drivers normally have lower efficiency than mid- or treble drivers. Thus, they require more power. Since I use power amplifiers with about the same low output power for both bass and main amplifiers, the buffer provides 15 dB of gain. If bass amplifiers with higher power output are used, this gain can of course be reduced.

The high pass filter is used to cut off frequencies below 20 Hz. In many cases this may not be desirable. For vented bass speaker boxes, however, it is mandatory. Since the driver in a vented box is not loaded below resonance, it really has to be protected from such low frequencies. Further, I also want to filter off low frequency wow and flutter from my turntable.

The low pass filter has the main task to filter off high frequencies in order to integrate the bass speaker with the main speaker. Since the main speaker low frequency cut-off to some extent depends on how the speakers are placed in the listening room, and also on other room effects, the high pass filter cut-off frequency may have to be adjusted.

The level control has proven very useful, since the bass level varies widely among recordings, productions and publishers. It seems like some kind of standardization is missing here…

Fig. 1 shows the schematic diagram. The buffer is an ordinary non-inverting stage with gain. It should be noted that stability problems can occur since the following high pass filter stage produces quite a demanding load. Thus a series resistor, Ri, in the diagram, is necessary to provide some isolation.

The value required depends on the op-amp used. If you have access to a scope the value can easily be determined by experimentation. If the value is too low, the buffer will oscillate at a very high frequency. I use the OPA 2134 in this circuit, and 22 ohm is required to keep the buffer stable. If you don’t have the opportunity to measure, set Ri to 200 ohms to play it safe.

For high frequency stability, I also added the capacitor Ci to reduce the gain. This may not be absolutely necessary, but is a good safety measure. In my case this capacitor tames a slight ringing at 220 kHz on the trailing edge of square-wave signals.

For filters I don’t use the common Sallen-Key filters, rather the third-order multiple-feedback filters (MFB). These filters have a few benefits when compared to the Sallen-Key filters, one of them being lower sensitivity to variation in the component values.

Further, I did not choose one of the usual filter alignments (Butterworth, Chebyshev, Bessel, etc.) because I wanted to control the pass-band response. By selection of appropriate Q-values, the response curve has been given a slight slope, and thus a careful equalization.

Capacitor C7 in the schematic diagram has been made variable. By the use of a small DIP-switch, one of three different values can be selected;
C7 = 15 nF gives ca 80 Hz roll-off frequency and about 2.6 dB eq
C7 = 25 nF gives ca 60 Hz roll-off frequency and about 2.0 dB eq
C7 = 40 nF gives ca 40 Hz roll-off frequency and about 1.5 dB eq

The proposed design was simulated on my computer. Fig. 2 shows the simulated response. The low pass filter cut-off is slightly gradual in the beginning and then rapidly increases to 18 dB/octave. This response is well suited to match the low end roll-off of my main speakers.

The high pass filter is given a small peak at 22 Hz. The peak gives the equalization and, at the same time, a steep roll-off of 22 dB/octave below 15 Hz.

After simulation and prototyping on the breadboard, it was time for printed circuit board layout. Fig. 3 shows the results. To the left is the component layout, to the right the film template, and in the middle the etched board ready for use.

And, after a few hours of mounting and soldering, all components and the completed board is depicted by Figs. 4 and 5. A nice box for the filter is also a requirement.

When the board is completed it is hooked up on the test bench for measurements. A lab twin power supply is attached, and the actual response measured. Fig. 6 shows the results. As is immediately obvious, the response agrees very well with the simulated design response. This is true for all three values of C7. Fig. 7 shows the response at somewhat higher resolution.

Most loudspeakers have a frequency range down to 100 Hz and often below that. Floor standers extend even lower, many are capable of reproduction down to and below 40 Hz.

The selectable upper cut-off frequency enables integration with main speakers in many different situations. My personal choice is not to use a high pass filter on the main speakers. Rather, I prefer to adapt the bass filter to integrate with the main speaker’s natural roll off. This does not only save on component count and phase change, but most of all I think the sonic results benefit from as few filters as possible.

Fig. 8 shows my current bass power amplifier. It is a “Gainclone” built specifically to drive bass boxes. Thus, it has a “heavy duty” power supply, as opposed to the more normal Gainclones.

Fig. 9 shows my bass boxes together with my main loudspeakers. The small bass box is the ULBOx presented on my Projects page. It is based the 8-inch Tangband W8-740C. The large box is my normal bass speaker (not yet published), a 72 liter vented box using an Audax HM210Z0 driver.

The active filter works just beautifully with both bass boxes and integrates very nicely with the main speakers.

The completed filter box is shown on Fig. 10. I did not include an integrated power supply, but an umbilical cable to provide power from the power amplifier. Rectified and filtered voltage is fed to two series regulators in the far end of the box. The regulators prevent any interaction between filter and power amplifier via the supply.

Signal connectors are mounted on the back panel, and the input signals are connected directly to the filter input buffer. The output signals are brought to the connectors via screened microphone cable (black on the picture). The front panel hosts the APLS audio grade potentiometer for convenient level control, see Fig 11.

[ previous ] [ back to top ] [ next ]