Buzz Audio DBC-20

Buzz update a characterful concept to create a compressor that's capable of both vintage and modern sounds.

Of all the different forms of dynamic processor on the market today, the diode-bridge compressor is generally considered to be one of the most characterful, While these became popular in the late 1960s and early 70s with the rise of solid-state electronics (the vintage Neve 2254 model is probably the most famous example, a design that lives on in the current Neve 33609), the very first diode-bridge compressor is claimed to be the Telefunken U13, which was introduced in the 1940s and used a valve rectifier. One of the reasons that the diode-bridge compressor became so popular was its very fast response — much faster than optical compressors, for example — and that made it very attractive in broadcasting applications, in particular, but its intrinsic sound character is also usefully different from its peers.

Now, New Zealand-based manufacturers Buzz Audio have produced their own take on the theme, in the form of the DBC-20 dual‑channel diode-bridge compressor.

The Bridge

At the heart of every compressor is some active device that provides variable attenuation, the amount being determined by a control signal of some sort. Fundamentally, a diode allows current to flow in one direction only, but there is a small region of its 'transfer curve' where its conductance effectively varies in proportion to the voltage applied across it, and that allows its use as a variable voltage-controlled attenuator in a diode-bridge compressor. Unfortunately, this operating mode isn't very linear and results in a fair amount of distortion, but by using pairs of closely matched diodes in a 'balanced bridge' configuration, a lot of that distortion can be cancelled out.

In the case of the DBC-20, Schottky barrier signal diodes are employed, as these were deemed to not only deliver the right sound character — more 'punchy' than from standard silicon types — but their tight manufacturing tolerances mean it isn't actually necessary to match them in pairs, and this helps to save costs and alignment time during production.

That 'bridge configuration' I mentioned typically places the diodes in a diamond array, with an audio signal applied across two opposite corners, and the control signal across the other two. However, this arrangement requires the audio signal to be in a balanced format (so that the control signal doesn't create a hefty DC offset at the output), and so diode-bridge compressors typically have fully balanced signal paths throughout, often with transformers to couple the signal in and out of the bridge. Typically, the remaining distortion from this arrangement is predominantly the odd harmonics, and that creates an harmonically rich sound character which tends to make the processed sounds appear to move towards the front of the mix — often a useful effect!

Another notable aspect of the diode-bridge design is that the audio signal has to be very small to remain within the appropriate part of the diode's transfer curve, with typical working signals levels around -40dBu or lower. Consequently, a considerable amount of gain is required after the diode-bridge attenuator stage to restore the signal to normal line levels — and in early designs that inevitably resulted in a high noise floor relative to other types of compressor. While this issue is much less relevant with modern electronics, the overall noise floor is still likely to be a little higher than, say, a VCA or opto compressor.

A shot of the internals. A simple linear power supply in the bottom‑left corner feeds the local regulator circuitry on each channel's circuit board. The yellow-banded 'radio steel' transformers can be seen on the right of each board along with the four modular plug-in discrete Class-A amplifier sub-boards. The compressor side-chain appears to be built using standard ICs.

Overview

Although it's fair to say that the Buzz Audio DBC-20 was broadly inspired by the Neve 2254, it isn't a copy or clone in any way. The amplifier stages in this elegant 1U rackmounting box (which extends around 250mm or 10 inches behind the rack ears) are all built using Buzz Audio's established BE41 discrete Class-A modules in a fully balanced configuration, and a 'radio steel' laminated transformer is used to feed the signal into the diode bridge. This transformer was selected specifically because it introduces some attractive saturation at higher drive levels, and this helps to lend the DBC a more vintage character.

The rear-panel audio I/O comprises a quartet of XLR connectors for the electronically balanced line-level inputs and outputs, along with an IEC socket for the mains inlet. A rotary selector is provided for 230 or 115 Volts AC mains sources. A simple linear power supply with a toroidal transformer is built into the rear of the rack unit, with the individual channel circuit boards carrying their own local voltage regulation.

The electronically balanced I/O are presented on XLR sockets on the rear panel. Though the two channels are independent, there's a stereo-link switch on the front which blends the control voltage for the two compressors.

Apparently the DBC-20's side-chain circuitry was inspired by classic compressors like the Fairchild 670, in that it has an adaptive ratio design. What this means is that the compressor has a very wide soft-knee response, so in effect the ratio moves smoothly and progressively from 1:1 below the threshold onto 2:1 and on up towards 10:1 (or so) as the input signal level increases. As a result, loud transients are clamped effectively, while lower‑level elements are controlled less assertively, and the transition into compression is quite gentle.

Most of the distortion comes from the third‑ and other odd-order harmonics, which is why it is typically perceived as adding 'focus' and 'clarity' rather than 'grit'.So instead of the familiar and traditional ratio and threshold controls, the DBC-20 features a single 'Compress' knob, which simply adjusts the input drive level, forcing more signal above the threshold and thus instigating more compression. This is partnered by a gain control, which provides up to 15dB of make-up gain to restore peak levels, if required. These two functions are set using the two largest knobs on the front panel, while two smaller knobs are used to select six attack and six release time constants.

The attack options start with 'Fast' (specified only as less than 0.5ms), followed by 5, 10, 15, 25, and 50 ms. The six release options are 50, 100, 200, 400, and 600 ms, plus an auto-release mode, which gives an initially fast response followed by a much slower recovery curve so that transients are handled effectively without causing audible background pumping. The range of both attack and release time constants is wider than most vintage units and some modern homages, too, allowing more modern-sounding dynamic processing effects to be dialled in when desired.

Three toggle switches select between hard/soft modes (the soft mode eases the maximum ratio to retain more programme dynamics), engage a high-pass filter in the side-chain (6dB per octave below 200Hz) to reduce the compressor's sensitivity to low-end signal energy, and provide a relay bypass.

This AP plot (H2) shows how the level of second‑harmonic distortion (green) varies with input level over a relatively small range. (The purple line is the residual level of the test set itself.) Compare this with the next plot (H3) which shows the level of third‑harmonic distortion rising dramatically as the input drive gets stronger and the compressor introduces more gain reduction.Each of the two channels has identical controls and operates entirely independently, but on the far right of the control panel two additional switches are provided to turn the power on or off, and to activate a stereo link mode. This last option combines the side-chain control voltages of both channels to ensure identical amounts of gain reduction in both channels, to prevent sideways image shifts when working with stereo material (it's important that the controls of both channels are set identically to prevent one channel's signal dominating the gain reduction response).

Above the attack and release switches for each channel are lengthy 15‑LED gain-reduction meters, covering a 20dB range with single‑decibel increments up to 10dB. Given the very gentle compression knee curve, though, I wonder whether 0.5dB increments might have been more useful over the first few steps, simply because I found that when applying gentle mastering-style bus compression, often only one or two LEDs lit up, wasting much of the potential display range and reducing the usefulness of the visual feedback. Apparently, the GR meter on the very first production units showed increasing gain reduction with the LEDs lighting towards the right, but designer Tim Farrant bowed to customer preferences and changed that in subsequent production units to the more conventional right-to-left scale for increasing GR, which is how the review model was set up. So perhaps some further product evolution might be possible?

Enhancing the gain-reduction scale in this way would be particularly relevant for the mastering world, and in fact there is a dedicated mastering version of the DBC-20 called the DBC-M. The differences between this and the standard version are minimal, though, with only the two standard potentiometers used for the gain and compression controls being replaced with 24-step Elma rotary switches (the make-up gain control providing 0.5dB increments). The idea, of course, is to allow precise repeatability of settings, which is often deemed important in a mastering context.

A plot of the overall frequency response with an extended frequency scale. The purple line is the response of the AP test set itself, while the green line shows that of the DBC compressor with its slightly tilted response (when connected to a very high-impedance destination).As usual, I checked out the review unit's specifications on the test bench using an Audio Precision test set, and found the maximum input and output levels to be a generous +27dBu, with the noise floor (ref 0dBu) around -85dBu (A-weighted) — very respectable for a device of this type. I also measured a -3dB point at 9Hz, which is slightly higher than the published specification of 6Hz, but that difference is easily attributable to the additive effect of the Audio Precision's own band-limiting filters.

A comparison of the compressor's ratio slopes in Hard (green) and Soft (purple) modes. The Soft option clearly reduces the maximum ratio that can be reached while also raising the overall gain slightly, giving a generally more dynamic sound.Looking at the frequency response, I noted a slight but distinct slope, which was -0.5dB at 50Hz, shifting to an even smaller rise above 5kHz. Although this is clearly visible in the frequency response plot, it may well be due to the minimal output loading (200kΩ) provided by the test set, and it isn't audible in normal applications. The overall bandwidth extends far beyond the AP's limit of 80kHz, and Buzz Audio claim the high end reaches 400kHz. This seems entirely plausible, although the absolute bandwidth does appear to vary a little with the amount of compression being applied (again, not in any audible way).

This plot shows the very soft-knee compression curve (green) compared to the linear bypass condition (purple).Harmonic distortion is, inevitably with this circuit topology, higher than some ultra-clean types of compressor, and it varies considerably with both the input level and the amount of compression being applied (as well as the Hard/Soft switch setting). The lowest THD figures I saw were around 0.02 percent, although THD of at least 10 times that was more typical in normal applications. When deliberately being driven hard for the most colourful and characterful effects, the THD reached as high as three percent or more. But that term 'characterful' is massively important here — the predominantly odd-harmonic distortion is generally perceived as adding focus and clarity, or bringing the sound forward, rather than making it sound dirty or crunchy.

In Use

I plumbed the DBC-20 into one of the analogue inserts of my Crookwood mastering console, and was able to compare it with both the digital processing of my Drawmer DC2476 Masterflow, and the emulated processing of UAD's Neve 33609 plug-in (I also used the hardware 33609 a lot in my former career, so know it fairly well).

The DBC-20's controls are simple and straightforward to set up, and all work over sensible ranges. In particular, I found the attack and release times to be well chosen, going from extremely quick for tight control, to nicely relaxed, with the auto-release option working extremely well. Engaging the side-chain high-pass filter with the Flat/Bass switch was useful with bass-heavy material, although even in the Flat mode there wasn't as much pumping as I expected. Careful choice of attack times allows drum transients, in particular, to be shaped in useful ways, and the overall sound character gains a sense of presence and clarity with even quite modest levels of compression, which I found very attractive without being overpowering.

When processing individual instruments, the amount of gain reduction and the input drive can both be cranked up considerably to really bring out the tonal coloration of the DBC, and I found that particularly useful with electric bass and male vocals.

Overall, I found the DBC to be a genuinely useful compressor. It works well as a bus compressor to add glue or cohesion to a final mix, while also bringing a gentle sense of presence and clarity. It's both cleaner and crisper-sounding than some diode-bridge compressors I've used, and while it can be pressed to develop a distinctly vintage character, it can also sound very modern and fresh. If you're looking for a distinguished compressor with a useful and interesting range of tonality, the Buzz Audio DBC-20 should definitely be on the shortlist.