Should digital converters always be ultra‑transparent? Burl think they should offer a little more character...
Burl Audio are a small, niche pro‑audio manufacturer based in Santa Cruz, California, who grew out of the Paradise Recording facility. Essentially, the company's products are designed in a commercial recording studio, by professional recording engineers, for recording engineers, and using proprietary circuitry with custom transformers. The initial products were the B2 Bomber converters reviewed here, but the product line now extends to API Lunchbox-format mic preamps — and much more is planned.
The design of the B2 Bomber A‑D places a considerable emphasis — and rightly so, in my opinion — on the design and quality of the analogue signal path, since this is the key element in the interface between the analogue and digital worlds and is often where quality and character is lost in budget converters. The B2 employs discrete class‑A circuitry, configured in a zero-feedback topology and without any capacitors in the signal path. It also uses large, custom 'BX1' transformers to help create the intended dynamic and tonal balance. Being a two-channel converter, the B2 was designed principally as a mixdown and mastering unit, although it can be used for tracking too, of course. The review model was not supplied with a manual, and none is currently available on the company web site either.
The rear panel of this simple steel 1U rackmount unit carries the standard IEC mains inlet feeding a universal power supply. Two AES3 outputs are provided on XLRs, with provision for dual‑wire operation (a largely defunct operating mode to accommodate 'quad' sample rates, but employed in some legacy Pro Tools systems), along with a single S/PDIF output on an RCA phono socket. Three BNC connectors cater for an external word clock input, and provide duplicated clock outputs. Finally, a pair of female XLRs accept balanced line-level analogue inputs. There appear to be internal jumper links for setting the loading impedance of the clock and digital I/O.
The front-panel controls are just as simple and straightforward. On the left-hand side, a six‑way rotary switch determines input level, with options to set the digital reference level (or headroom margin) to ‑12, ‑15, ‑18, ‑20, ‑22 and ‑24 dBFS. An adjacent rotary switch turns the dual-wire AES3 mode on or off. A large 30‑segment bar-graph meter dominates the centre of the device's control panel, the meter range spanning ‑72 to 0dBFS, with considerable scale expansion at the top end (the right‑hand half of the meter covers a 12dB range, while the left half covers 60dB). This is useful in a mastering context, where precise control of peak levels is important, but less useful in a tracking or mixing situation, where the usual headroom margin means that half the scale is under-used. Usefully, though, the metering shows signal peak levels as a floating single LED above the solid block of LEDs representing the average (RMS) levels. Indicators within the metering section illuminate to display the sample rate and external clock lock status. A small black push‑button resets the peak-hold lights.
Over on the right‑hand side of the panel are two more rotary switches. The first selects internal or external clock, while the second sets the internal sample rate at all the standards between 44.1 and 192kHz. A large black rocker‑switch powers the unit and completes the front-panel controls.
The internal construction is very neat, with a Powdec OEM universal power supply section on the right‑hand side, and a large main circuit board covering most of the rest of the floor area. The digital side of the circuitry uses entirely surface-mount components, while the analogue side uses a combination of conventional discrete transistors and surface-mount components. The input stage appears to be a hybrid arrangement employing a pair of LSK389 ultra low‑noise dual JFET op‑amps in concert with a six‑transistor, class‑A gain stage, and the input appears to saturate gracefully and in a benign and musically appealing way when driven with high levels. This end of the device also runs pretty warm, and good rack ventilation would be desirable. The A‑D converter is the familiar and widely used AKM 5394 and the digital outputs are from Burr Brown DIT4192 transmitters, via pulse transformers.
My technical tests produced figures that agreed very closely with the published specifications. Frequency response was typically within ±0.1dB from below 20Hz to just below half the sample rate. At 44.1kHz, the ‑3dB point at the low end was 6Hz, rising to about 10Hz for the 192kHz sample rate. The THD+N figure was ‑100dB, or 0.001 percent, for a +4dBu input level generating ‑12dBFS, and with a predominantly odd‑order harmonic disposition. Interestingly, the phase response remained very linear, despite the presence of an input transformer — and no doubt this has a beneficial effect on the sound character. The AES17 dynamic range measurement provided a figure of 115dB, which is very good. The Windt Hummer test revealed no ground-loop susceptibility, and the noise floor is very quiet and clean.
The partnering unit to the A‑D is constructed very similarly and features the same pattern of front-panel controls and rear-panel connectivity, albeit with different functions. The rear panel carries the same IEC universal mains inlet, a pair of AES3 inputs on XLRs and an S/PDIF input on an RCA phono socket, a trio of BNCs providing word clock in and duplicated outputs, plus a pair of male XLRs for the analogue line-level outputs.
Internal construction is very similar to the A‑D, with the same Powdec universal PSU module and a large main PCB. Digital inputs are handled by AKM 4115 transceivers, while the D‑A converter is the AKM 4399 chip. The analogue stages are transformerless and employ one of Burl's proprietary BOPA1 all‑discrete op‑amp modules. Again, the circuitry is all class‑A, with no capacitors in the signal path, and it runs surprisingly hot. Good ventilation above the Bomber D‑A is probably a must.
The front-panel controls start with the familiar output-level rotary switch, offering options from ‑18 to ‑8dBu. A 0dBFS digital input generates +21.3dBu at the analogue output when set to the 18dB position, and +12dBu when at the 8dB position, the other positions giving 2dB increments in peak level. The scale numbering is indicating headroom above a nominal +4dBu reference level — hence the 8dB position giving a peak output 8dB above +4dBu. The maximum output should be +22dBu, but there appears to be some minor attenuator misalignment at the very high end.
The adjacent rotary switch selects the input source, with options for the two AES3 inputs, the S/PDIF input, or the two AES connectors operating in dual‑wire mode. The metering display is identical to that in the A‑D converter, with the same peak/RMS display, scale range, sample rate and lock indicators, and peak reset button. The next rotary switch is unusual, because it determines the clock source. The options include each of the digital inputs, as you'd expect, but adds to that list the external clock input and an internal generator, the sample rate of which is established with the final rotary control, from any of the standard rates between 44.1 and 192kHz.
This is an unusual arrangement and should be used with considerable care, as inappropriate settings will badly damage the audio signal. The idea behind the facility is to enable the B2 DAC to serve as the master clock, by setting it to run on its internal clock. The rear-panel word-clock outputs can be used to distribute this master clock to all other digital devices in the system (eg. the DAW) that operate as clock slaves. In this configuration, the digital output from the slaved device should be synchronous with the DAC's internal clock and high-quality conversion will result. The advantage of this is that the DAC's converter chip is running on its own internal — and very stable — crystal clock, rather than trying to slave to an external or embedded clock, and as a result, jitter artifacts should be considerably reduced and the audio quality enhanced.
Several commercial DACs employ techniques for running the converter from a stable internal clock divorced from the source's embedded clock, but very few do so by acting as the system master clock, even though this is actually a very logical and practical solution.
The down side is the potential for user error, which is quite a serious risk. For example, with a free‑running 44.1kHz source the converter's output may appear satisfactory even when switched accidentally to its internal 44.1kHz clock. However, the source audio will be corrupted with regular clicks as the two clocks drift in and out of sync, and the audibility and rate of those clicks will depend on the nature of the audio material and the relative drift of the clocks. Unfortunately, there's no warning for this simple user error: no flashing lights warn of clock drift, and there's no automatic muting for mismatched clocks. The best case is when the internal and source clocks are mismatched (say 44.1kHz internal and 48kHz source), as this will produce very obvious aliasing artifacts, but if the internal and source clocks are related, the best case would be to notice regular but quiet clicks...
When the clocking arrangements are configured correctly, performance is very good indeed, achieving an AES17 dynamic range figure of 114.5dB, with crosstalk better than 103dB and a frequency response that's flat within ±0.1dB from 10Hz up to just under half the sample rate. Again, there's a pronounced tendency towards odd‑order harmonic distortion, especially when driven hard, but this is all part of the B2's sonic character.
The Burl B2 Bomber converters are attractive, well-built units, and capable of delivering excellent sound quality, but with a substantial side‑order of analogue character. Indeed, it's the analogue design that sets these devices apart, bestowing the subtle but often welcome sonic colorations associated with transformers and discrete class‑A circuitry, as well as the nice qualities those technologies lend to transients. The digital side of things is well specified and generally well implemented, although it's hard to overlook the risks of user error due to the unusual clocking options of the D‑A converter.
My overriding impression of the Bomber converters is that they really do sound a lot more 'analogue' than most. We're talking about the deliberate inclusion of some 'nice' and subtle distortion artifacts, but there's also something smoother and more natural‑sounding about the top end, while the bottom end is very tightly controlled but still larger than life, and the mid-range gives the impression of being slightly more expansive and dynamic than expected. None of these subtle characteristics were revealed in my bench tests, but they add up to a very nice style of presentation that addresses the 'sterile' nature of ultra‑clean digital systems rather well. First and foremost 'musical' rather than 'transparent' converters, they have a kind of tape‑like ability to handle transients in a very flattering way, and the power to make a mix sound glued together in a way that I usually associate with top‑flight analogue systems.
The closest-sounding alternative converter to the Bomber that I can think of is the lovely Universal Audio 2192, which I reviewed recently in SOS and which has a very similar audio character. I wasn't surprised to learn, therefore, that Rich Williams, the man behind Burl, also designed the UA 2192 and describes his B2 converters, logically enough, as the 'next generation' of converter.
During the course of this review, Hugh used the Audio Precision Analyser to take some test measurements of the B2 converters. We've placed the results on the SOS web site for all to see, at: /sos/nov10/articles/burlb2media.htm.