Ask the Doctors: Signal Detection in the dbx 160
By Dave Berners

In the December, 2007 “Ask the Doctors” column, we looked at the UA Precision Buss Compressor, which was created with the intent of capturing the characteristic vintage VCA sound. Benefits of VCA-based compression were enumerated, and typical properties of VCA compressors were discussed.

dbx 160

This month, rather than covering VCA compressors as a group, we will focus on the dbx 160, which stands out among VCA compressors for its distinctive sound. Although the 160 is a VCA-based compressor, it has some qualities that set it apart from most other VCA compressors. The 160 is distinctive enough that UA will release a separate plug-in created for the express purpose of modeling its intricacies.

In our December, 2007 column, we looked at the benefits of the generic VCA (Voltage-Controlled Amplifier) as a compression element: High bandwidth, low distortion, large dynamic range, and precise control over gain make the VCA suitable for most any dynamic-range control application. The VCA is, by some measures, the ideal gain control element.

The 160 is distinctive enough that UA will release a separate plug-in created for the express purpose of modeling its intricacies.

Most of the valuable properties of the VCA can be expressed as the absence of flaws. In other words, the VCA provides a clean, well-behaved, gain reduction element. Use of the VCA can simplify the design of a compressor, and can improve unit-to-unit consistency and fidelity as compared to other gain reduction technologies. However, for a typical compressor, much of the interesting behavior is contained in the signal detection subcircuit, rather than the gain reduction circuit. Usually, attack and release trajectories, compression ratio, and threshold are all determined by signal detection methods. For some technologies, compression knee shapes are in part determined by the gain reduction technology, but for VCA compressors, even a soft knee can be implemented as part of signal detection.

As it turns out, many of the interesting qualities of the dbx 160 are due to the design of its signal detection module. Known as the RMS 208 module, this was a subcircuit mounted within a metal can on top of the compressor's circuit board. The RMS 208 is thought of as an RMS detector, but actually performs a peak detection on a full-wave-rectified, log-encoded signal. This module, along with some post scaling and level adjustments, provides the control signal for the dbx 160's VCA module. Because the signal is transformed into log-space before attack/release filtering is performed, the output of the detector cannot be viewed, strictly speaking, as a true RMS estimate. Also, the attack being faster than the release for this detector would make it atypical for an RMS detector. However, the RMS 208, in conjunction with any logarithmic VCA, will provide selection of any desired constant compression ratio, which would also be the case with true RMS detection.

The RMS 208 uses several diode-connected transistors to perform log encoding and rectification on the input signal. Using a pair of op-amps for impedance buffering, the processed signal is used to affect the state of an attack-release filter. The attack is designed to be as fast as possible, with no resistors included to limit the rate of attack. As a result, the attack trajectory depends on the internal impedances of some transistors. This means that the attack curve for the dbx 160 is highly nonlinear, with an attack time that increases as the attack progresses. A current source provides the bias for the release out of compression; in conjunction with the VCA, this gives a release that can be roughly measured in dB/ms. As the compressor comes fully out of compression, the release time is increased, again due to internal impedances within the circuit's transistors. Nonlinearities in the RMS 208 module do not provide program-dependent compression per se, but they do give the dbx 160 a distinctive behavior.

The 160's popularity continues because of the eccentricities of its RMS 208 detection circuit.

The use of a VCA within the dbx 160 allows the signal-detection output to be adjusted in a simple way to provide a continuously variable ratio of compression. Also because of the use of the VCA, there is no soft knee on the dbx 160. However, the nonlinear nature of the release causes an effect similar to a soft knee, since the release slows down when coming out of compression. During attack, the nonlinear trajectory of the signal estimate causes behavior similar to the attack of a feedback compressor, even though the detected signal is the compressor's input rather than its output. So, the RMS 208 produces multiple desirable compression characteristics as a result of its inherent nonlinearities. When combined with the use of a VCA, the RMS 208 provides a flexible compression scheme that has many attractive sonic qualities.

At the time of its release, the dbx 160 was on the forefront of technology. Its integrated design and continuously adjustable ratio made it an advanced product. Initially, its popularity was partly because its design was cutting-edge in terms of technology. Today, the 160's popularity continues, at least in part, because of the eccentricities of its RMS 208 detection circuit. Its distinctive sonic signature promises to remain unmatched by other hardware products, because of the subtleties of this very interesting detector.

Questions or comments on this article?