Experiences with the Behringer DCX-2496 digital crossovers



The DCX mounted on top of the quadruple power amplifier in a specially made MDF-frame.


Since the DCX-2496 was designed for pro-audio applications with symmetrical in- and outputs and much higher levels a need for a switched attenuator was required. I designed a passive attenuator with -10 and -30 dB attenuation. This switchbox looks as follows: :

The 5 XLR's on the left go to the 5 used outputs of the DCX's, the 5 pin DIN connector on the middle right is the link to the 4 amplifiers on top of the speakers and the XLR on the right is the output for a (future) subwoofer.

The inputs are connected to 5 outputs of the DCX-2496 : subwoofer, bass, midbass, mid and high. The outputs are used asymetrically (pin 1 and 3 shorted on the switchbox side).

Nearfield (10 cm) measurements of the drivers with appropriate highpass filters to protect the drivers from overexcursion :

Now with Butterworth filtering at 48 dB/octave and some equalising for the midbass unit :

The bass output is 6 dB lower because the measurement was taken on 1 woofer only. Since it's a nearfield measurement the baffle step dropoff is not present and the output of the bass reflex port is not taken into account. The adjustment of the influence of these must be done with a far field measurement, which is hopelessly inaccurate inside rooms because of all the reflections and standing waves.

The following graph shows a problem I discovered when measuring the DCX's with Audua Speaker Workshop :

The bass output shows a drop of over 1 dB towards the lower cutoff frequency, while the measurement setup was calibrated to less than 0.01 dB error. The subwoofer output shows a 1 dB too high level on the contrary. This "bug" was reported to Behringer more than 1 month ago, with no response up to now. The DCX's show a nearly flat frequency response when the outputs are configured for wideband, so the error seems to be from the digital signal processing and not from the analogue circuit parts. The calibration curve from Speaker Workshop :

This clearly shows that the error is from the DCX and not a measurement error. Both my DCX's showed exactly the same behaviour.