Apogee's new converter builds on the reputation of the original PSX100 and Rosetta units, upgrading the sonics and adding a pile of new facilities, including sophisticated jitter-reduction processing and sample-rate conversion.
The Rosetta 200 is the newest two-channel converter in Apogee's impressive range, and it replaces both the highly regarded PSX100 and its simpler sibling, the original Rosetta converter, both of which I reviewed back in SOS November 1999. The converter and clock technology employed in the Rosetta 200 is essentially the same as that in the Rosetta 800 (reviewed in SOS March 2004), with full support for sampling rates up to 192kHz (including SMux facilities for the ADAT port) and comprehensive I/O connectivity. There is also a slot for an optional interface module, with cards available to couple directly with Pro Tools (either new HD or older Mix systems) and for a Firewire interface for native DAWs.
The PSX100 was a superb unit, and the Rosetta 200 is designed to build on that, offering excellent signal routing and format-conversion functions, but perhaps the real icing on the cake is the new Coda 'Audio Finishing Module'. This provides (for the first time in an Apogee product) high-quality sample-rate conversion, along with an automatic level-setting facility for the A-D converter called the Aptomizer. The familiar UV22HR dithering algorithm is also included, as is Apogee's Soft Limiter — an fast-response analogue input limiter.
The rear panel of the Rosetta 200 carries balanced line-level analogue inputs and outputs on XLRs, adjustable for peak levels anywhere between +2dBu and +26dBu. AES3 inputs and outputs are also on XLRs, dual sockets being provided to accommodate double-wire interfaces for high sample rates.
Phono (RCA) sockets are used for coaxial S/PDIF in and out, along with a pair of Toslink optical sockets (configurable for ADAT or S/PDIF formats). Word-clock input and output are on BNCs (with selectable 75Ω input termination), and there is also a pair of MIDI sockets (in and out), a removable plate over the interface card slot, and an IEC mains inlet (accepting anything from 90-250V AC). The MIDI sockets are provided for software updates, but if the Firewire card is installed then MIDI data can also be passed via these sockets to the host computer.
The Rosetta is configured almost entirely through various front-panel button combinations, and this makes it very easy to change settings, as well as allowing a complete reset to default factory conditions if required. The only configuration option set 'mechanically', by an internal jumper link, is the power-on mode, with a choice between permanent powering, or power switchable from the front-panel button).
Apogee have always been praised for their stable, low-jitter clocks, and the latest Intelliclock design employs a dual-stage jitter-reduction technique that uses a FIFO buffer to isolate external, unstable clocks from the internal reference clock — a FIFO buffer (first-in, first-out) is a form of short-term memory. Incoming audio samples are loaded into the FIFO buffer using a 'write' clock derived from the incoming digital signal, and a high level of jitter and instability can be tolerated here, since stable timing isn't required. The audio data is then retrieved from the FIFO buffer using a very precise 'read' clock, which permits jitter-free D-A conversion. Of course, the FIFO buffer introduces a short storage delay, but this only amounts to a few samples, and is irrelevant compared to the normal conversion-filter delays. A similar approach is used to ensure that the A-D conversion is jitter free, even when using an external reference clock.
All the input sources are effectively connected using a buss arrangement, feeding two independent source selectors controlled by front-panel buttons. One selector feeds all of the digital outputs together, while the other feeds the D-A converter, and hence the analogue outputs. Two front-panel buttons determine the clocking. There are six standard sample rates derived from the internal crystal between 44.1kHz and 192kHz, plus an external input. If the external-input mode is selected, then the next button selects which source to use: S/PDIF, ADAT (or SMux), AES, word clock, or the option-card input. A pair of LEDs arranged to resemble an exclamation mark indicate the lock status: both lights mean a stable tight lock has been achieved, whereas just the upper LED means that the reference is unstable.
The Rosetta 200's card slot can accommodate a growing range of interfaces. Pro Tools users will be interested in the two cards which link directly with the Pro Tools core processing cards. The X-HD card is designed for the current HD hardware, while the X-Digi-Mix card interfaces with 'legacy' Mix-system hardware — both allowing users to bypass Digidesign's own interface hardware.
Another useful card is the X-Firewire module. This provides a generic interface to any native audio software running on Mac OS X or Windows XP platforms equipped with a Firewire socket. The card uses Apple's Core Audio Firewire driver on the Mac platform, or WDM/ASIO drivers for Windows. The Rev-E version of the Firewire card is equipped with two FW400 connectors, and the sockets are configured to pass power when connected with suitable computers.
Most of the front-panel buttons have a second function, activated by holding the button down. For example, the sample-rate selection button also determines the AES format (single or dual wire) when a dual- or quad-speed sample rate is selected, and also the optical output format (S/PDIF or ADAT). The external-clock selector button also sets the word-clock I/O ratio, which allows the unit to operate at a different sample rate to the external reference, or to generate a different clock output rate from the internal rate. The digital output source button also engages a sample-rate converter, allowing variable or asynchronous sources to be connected.
The sample-rate converter is a very handy new addition, and reflects the significant advance in this technology in recent years. The best of the current crop of SRC hardware chips can now process digital signals with a resolution and quality that exceeds the performance of the analogue converters, and so the process is essentially transparent regardless of sample rates.
The Rosetta 200 features two sets of stereo bar-graph meters, both spanning a 50dB range, with the 0dBFS LEDs quaintly being labelled Ouch. There is also a calibration mode in which the meters are re-purposed to show 1dB increments between -20dBFS and -10dBFS (with an accuracy of 0.1dB) to allow accurate alignment of analogue I/O levels. In the calibration mode the A-D and D-A levels can be adjusted in 0.1dB increments, with independent or correlated adjustments for each channel.
I must say that I found the meter labelling rather confusing — the Analogue meter at the top actually shows the level feeding the digital outputs, while the Digital meter below shows the level feeding the analogue outputs through the D-A converter. This seems backwards to my simple mind!
By default, all the digital outputs are at 24-bit resolution, but pressing the Coda Process button introduces the UV22HR dither algorithm to transcode all outputs to 16 bits. However, the process only operates if the sample rate is 44.1kHz or 48kHz, the assumption being that all high-sample-rate recordings will always be at 24-bit resolution. There is also apparently provision here to engage an automatic SRC mode, the idea being that if any asynchronous digital source is selected the SRC will be engaged automatically. However, this function is not enabled in the current firmware.
The other part of the Coda processing is the Aptomizer, controlled with its own button. The Aptomizer automatically adjusts the input level of the A-D, based on an analysis of the peak levels during a 'learning' period (activated by holding the Aptomizer button down). The system aims to get peaks within half a decibel of clipping — which is closer than I would normally aim for when manually setting up peak levels. With controlled and predictable material (for example commercial CDs or broadcast radio signals) the system works well in practice, giving a very 'hot' output level. However, it can be caught out by material with unpredictable transient peaks if the learning period doesn't include any. Consequently, if you use the Aptomizer it might also be sensible to employ the Soft Limit function — although it is possible to adjust the Aptomizer level through an associated Trim mode, should you feel the need for a tad more headroom.
The Aptomizer does more than just alter the A-D input level. It also adjusts the output levels from the D-A to maintain unity gain through the unit. In some circumstances this is very convenient — for example, if you are trying to patch an external analogue signal processor into a digital signal path, or connecting an external analogue recorder to a DAW system for mastering. However, there are probably just as many situations where this automatic realignment of the D-A output level would be rather frustrating!
The one that got me was while using the D-A to feed a monitoring chain at the same time as using the A-D to feed different analogue sources into a DAW. This is likely to be a very common working arrangement, and one in which the D-A alignment must not change. However, if you use the Aptomizer to set the A-D conversion level automatically, it will mess up your monitoring reference level in the process. Currently there is no way of isolating the Aptomizer's control over the D-A, which I feel is a significant shortcoming. Hopefully a future software upgrade will address this problem, but until then you'll have to ignore the Aptomizer function and rely on manual adjustment of the A-D input levels instead in such circumstances.
The Rosetta 200 represents a significant operational improvement on the PSX100, and the sound quality is better too, although by a smaller margin than I had expected. Comparing the Rosetta directly against the Benchmark DAC1 and ADC1, as well as a PSX100, the more expensive Benchmarks revealed only a very slight veiling of the extreme high end, and a touch of 'dryness' at very low frequencies. Image width and depth seemed comparable, as was the general resolution and neutrality.
I have some reservations about the Aptomizer feature — specifically the inability to restrict its action to the A-D only — but the sample-rate and format conversion facilities are excellent, and the Rosetta turns in a very creditable performance indeed. If you are looking for a simple A-D and D-A, there are more cost-effective solutions around, but if you value the ability to interface multiple sources, feed multiple destinations, and convert digital signal formats, resolutions, and sample rates, the Rosetta 200 is probably the investment you have been looking for.
- Very good converter quality.
- Superb I/O flexibility and routing.
- Sample-rate conversion facilities.
- UV22HR dithering.
- Aptomizer automatic level-setting function.
- Front-panel configuration.
- Full output metering.
- Aptomizer leaves negligible headroom.
- Not possible to restrict Aptomizer to A-D only.
- Automatic SRC function not yet available.
A superbly flexible stereo converter, combining excellent A-D and D-A performance with comprehensive I/O routing, format conversion, sample-rate conversion, and 16-bit UV22HR dithering options.
Rosetta 200, £1639.13; X-HD, X-Digi-Mix, and X-Firewire option cards, £499.38 each. Prices include VAT.
Sonic Distribution +44 (0)1582 470260.
+44 (0)1582 470269.