RME ADI8 DS

German manufacturer RME maintain their high reputation for quality A-D/D-A interfacing with this eight-channel device, which may be run at 24-bit, 96kHz operation if desired.

RME have built up a good reputation and strong following for their range of converters over the last few years, and many of the company's products have appeared before in the pages of Sound On Sound. The latest addition to the product range is the ADI8 DS, which is part of the burgeoning ADI range and also a development of the ADI8 Pro. In short, this is an eight-channel bi-directional converter with a host of additional functionality thrown in, and all housed in a 1U rackmount case. The most obvious difference compared with the ADI8 Pro is the fact that the 8 DS supports 24-bit, 96kHz operation — the DS stands for Double Speed!

Like its predecessor, the new unit provides both ADAT- and TDIF-formatted outputs and inputs, with dual interfaces for each to allow bit-splitting as before, enabling 24-bit data to be recorded on 16-bit media. However, a further extension to this idea is the Double Speed option, enabling the transmission of high-sample-rate data via the ADAT lightpipe interface using the S/MUX format, which sends one channel's high sample-rate signals via two ADAT channels in much the same way as the original bit-splitting technique.

A full range of clocking modes is also incorporated, and there is a digital copy facility to enable data presented on one of the digital inputs to be routed directly to all the digital outputs, complete with any required format conversion along the way.

Round The Back

Usually, a quick look around the back of a product like this often saves hours of reading through translated manuals, because the main functionality becomes immediately obvious. In the case of the ADI8 DS, the top panel is printed with a large block diagram showing the various signal paths available within the unit, and the rear panel is well laid out and surprisingly uncluttered given the amount of interfacing provided. In fact, the interconnectivity of the ADI8 DS is identical to that of the ADI8 Pro.

To the extreme left we have the ubiquitous IEC mains inlet for a switched-mode power supply accepting voltages from 100 to 240V. The analogue I/O is all arranged along the bottom row of connectors on the unit, while the digital I/O is arranged across the top.

Analogue inputs to the A-D are catered for in two formats: a 25-pin 'D'-sub wired to the familiar Tascam standard in the centre of the unit and an octet of TRS quarter-inch jack sockets — both are electronically balanced, adjusts for balanced and unbalanced inputs automatically, and can be switched for three different input sensitivities from the front panel. The balanced analogue outputs from the D-A converters are similarly equipped with a 25-pin 'D'-sub and eight TRS sockets and both may be used simultaneously if required. Again, the servo-balancing circuitry adjusts the level automatically for balanced or unbalanced destinations, and once again the maximum output level can be determined by a front-panel switch.

The digital I/O comprises a further pair of 25-pin 'D'-sub sockets to provide a main and auxiliary TDIF I/O, plus four TOSlink ADAT optical ports, each fitted with the usual, removable dust cap. Of these, two provide the main and aux inputs, and two the main and aux outputs. The slightly odd thing, to my way of thinking anyway, is that the digital inputs are grouped with the analogue inputs, and the digital outputs with the analogue outputs, rather than in a logical A-D section (analogue ins, digital outs) and D-A section (digital ins, analogue outs). It's a small point and one which will be largely irrelevant when the unit is plugged up and bolted in a rack, but it confused me when I first started to hook the box up.

The ADAT optical inputs employ RME's bespoke 'Bitclock PLL' to extract the embedded word clock, while the TDIF boasts a 'low jitter PLL' circuit to help extract its word clock signal. The TDIF interface incorporates support for automatic decoding of pre-emphasised tapes, but the emphasis flag is not passed through to the output in copy mode.

Word-clock input and output are provided on BNC connectors, and near the input socket is a recessed button which activates a 75Ω termination. Normally, the word-clock input presents a high impedance, suitable for bridging or daisy-chain clock termination. The sophisticated clock recovery circuitry can cope with a wide range of signal distortions including overshoots, and asymmetrical or misshaped clock pulses. The word-clock output can be used to distribute clock signal to other units, to act as the clock master for an entire system. This output clock is derived from that used by the A-D, and if an external input is selected, the output clock is phase-locked with a zero degree offset to the input clock. It is also required to synchronise slave DTRS machines such as the DA88.

If you weren't certain from looking at the front panel, the back panel makes it clear just how versatile the ADI8 DS is, with all eight outputs available on the left (on jacks, analogue 25-pin 'D'-sub connector, Main and Aux ADAT optical connectors), all the inputs available on the right in the same formats, and the bi-directional TDIF-format digital Main and Aux connectors on 25-pin 'D'-sub connectors in the centre. A word-clock in and out (the input with recessed 75(omega) termination button) complete the picture.

The front panel is styled in the same way as most other RME equipment — a brushed metal panel with screen-printed grey boxes and white lettering to identify the various control sections and functions.

The mains power rocker switch is at the right-hand side, and the operational controls are arranged logically from left to right, with the A-D facilities on the left, the clocking and digital I/O format selection in the centre, and the D-A facilities on the right.

A-D Section

The converters are constructed using 128-times oversampling, 24-bit converters, which achieve a claimed 117dBA dynamic range, and the analogue circuitry is completely symmetrical and DC-coupled throughout. In fact, the published specifications are slightly better than those for the ADI8 Pro, and that was no slouch either.

The input level is adjusted for all eight channels simultaneously via a single push button which cycles around three input sensitivity options. The current mode is shown on one of three LEDs labelled 'Lo Gain, ' +4dBu, and -10dBV and these correspond to maximum input levels (ie. 0dBFS) of +19dBu, +13dBu and +2dBV (+4dBu).

Each channel's input level is metered with two LEDs, the first being a green 'OK' light, the brightness of which is proportional to signal level and this starts to illuminate when the input signal exceeds -40dBu. A red 'Over' LED illuminates at -2dBFS — so it's more a peak warning light than a true over LED — but I'd rather be warned when peak levels are close but below clipping than when they're over!

The A-D converter outputs are made available via both output formats simultaneously, and to all ports — both main and aux — unless one of the bit-splitting modes is in use (more on these in a moment).

The last facility in the A-D section is a button labelled 'Process' which is associated with three more LEDs. Again, pressing this button cycles through three options: dither, double-speed and bit-split. The dither mode reduces the word length of the output and re-dithers it using triangular dither (TPDF) at 16-bit resolution, for use with 16-bit ADAT or DTRS recorders, for example. The double speed mode causes the A-D converter to operate at double the selected sample rate, but because the ADAT lightpipe port does not support high sample rates, the unit automatically engages a sample-splitting mode, known as 'double wire' to Tascam owners and S/MUX to ADAT users.

It is worth noting that the word clock output from the BNC terminal remains at 48kHz (or 44.1) to provide a valid clock reference to DTRS recorders, even though the audio data contains 96kHz (or 88.2) samples — and this is the reason why the high sample-rate option is not included as part of the clock section on the front panel.

In the double speed mode, each channel occupies two tracks of the recording device, so channel 1 is dispatched to tracks 1 and 2 of the recorder, channel 2 to tracks 3 and 4, and so on. Channels 5-8 are output via the second 'aux' digital output: channel 8 being allocated to tracks 7 and 8 of this second machine.

The third option is the bit-split mode, and this is used to accommodate 24-bit data on a 16-bit medium. There are several different ways of doing this, and the method chosen by RME is the same as that provided as an option by Yamaha on some of its digital desks, for example the 02R. This arrangement requires two recorder tracks per channel, with the first channel being routed to tracks 1 and 5, the second to 2 and 6, and so on. To record all eight channels, two recorders are required, and channels 5-8 are output through the aux interface. Needless to say, it is not possible to use both double speed and bit-splitting modes simultaneously, as both require the aux output, but use it in different ways.

D-A Section

The D-A section is fractionally more complex than the A-D since there are three buttons instead of just two. The maximum output level can be selected using one of these buttons to cycle through the three options: 'Hi Gain', +4dBu and -10dBV. These three modes correspond precisely to the input sensitivities, providing peak output levels or +19, +13 and +4dBu respectively. Like the A-D, a row of eight green LEDs illuminates progressively as the signal level rises above -40dBFS.

The button to the left of the section provides one of two data recombining modes corresponding to the Double Speed, and Bit Split configurations of the A-D section. Clearly, it is vital that when more than four channels are involved, the two multi-channel digital inputs carry synchronous data, and a green LED labelled Sync is provided to indicate the clocking status. If data is not present on the second aux input port the LED is off, while present but unsynchronised data is shown by a flashing LED. If the data on the aux input is synchronous with the main input the LED remains lit.

The last button is labelled 'Copy' and this is used to route the digital inputs through to the corresponding digital outputs, thus enabling digital cloning of data between machines.

Clocking Section

The final, central control section of this unit is concerned with the clocking and digital source selections. Only one of the two digital formats can be accepted as the input to the D-A at a time, so a push button allows either the ADAT or TDIF input to be selected.

The clocking arrangements are quite sophisticated, with three more buttons to determine what is going on. The clocks for the A-D and D-A can be selected independently from one of three sources: either the selected digital input (ADAT or TDIF), the external word-clock input socket, or the internal crystal clock. If the latter, sample rates of 44.1 or 48kHz can be selected via a third button.

The logic of the clock-source switching initially seems slightly obscure. For instance, changing the D-A clock source also changes the A-D clock source under certain conditions, and sometimes the A-D clock can not be changed at all unless the D-A is changed first! For example, if both A-D and D-A are set to the external input, and then the D-A clock is changed to internal, the A-D also changes to internal, and if the D-A is set to internal, the A-D cannot be changed until the D-A is selected to something else. If you think about it, the automatic clock selections and barred modes are actually perfectly sensible and logical — it can just be a little frustrating when you want to change the A-D clock source and the machine won't let you!

In Use

The ADI8 DS is a very simple product to use, and I had the unit hooked up and running in just a few minutes. Configuration is simply a case of deciding on the best operating levels in and out, which doesn't require much thought really, and what the clock source(s) should be.

I started by hooking the ADI8 DS up to my usual Apogee PSX100 converter, using both the lightpipe and TDIF interfaces to pass just a stereo pair over two channels at first. I was then able to feed the output from the Apogee A-D, acting as a reference source, through either the RME or the Apogee D-A to compare the differences. Equally, I was able to reconfigure the system to compare the two A-Ds, this time using the Apogee D-A as the reference decoder.

In both cases, I was impressed with the sound quality provided by this unit. Stereo imaging seemed very stable, with good depth and width — indicating low jitter — and the resolution of both the A-D and D-A converter was to a high standard.

Hooking up the unit in a more conventional way, it proved to be very easy to use the ADI8 DS — it simply got on and did all that was asked of it. The 16-bit dither mode worked exactly as it should, with a smooth noise floor at -93dBs, entirely characteristic of the TPDF dither function used.

The difference in sound quality between standard and double sample rates was subtle, but noticeable — the latter being slightly more open-sounding when running from its internal crystals. However, connecting the word-clock input to an external 96kHz word clock source caused no end of confusion! It seems odd to provide an external word clock input on a 96kHz converter, but not include the ability to lock to an external 96kHz reference clock! Standard clock rates were accepted without problem, though.

Both the bit splitting (enabling 24-bit data to be recorded on 16-bit media), and the Double Speed modes (using the S/MUX format) worked well and without any problems at all, although it is obviously vital to ensure that if two eight-track recorders are being used they are correctly synchronised together to sample accuracy on replay. Any loss or drift in sync is highlighted by the flashing of the Sync LED, but even if the Aux side of the link is lost completely the audio continues with no interruption (only reduced quality, and the reduction is pretty subtle). Losing the main side of the link quickly causes both Combine modes to give up in disgust!

I liked the clocking facilities — in particular the ability to slave the D-A converter's clock directly to the same clock as the A-D — ideally the internal crystal. This arrangement can help to reduce jitter, noise and distortion considerably, since the D-A clock will be the same one as used for the original A-D conversion, and which will also usually be the clock master for the entire recording chain. The advantage as far as the D-A goes is that, rather than having to extract the word clock from the incoming digital audio signals — and thus potentially suffer from cable-induced jitter artefacts — the local word clock is guaranteed clean and accurate.

Overall, then, this is a fine eight-channel dual-converter offering a useful range of facilities and functionality at an attractive price, making it an ideal partner for a wide variety of project studio systems. It is ideal for use with 'legacy' ADAT and DTRS recorders — helping to extend their useful life by enabling them to become high-resolution recorders, albeit with only half the number of tracks per machine. It can also act as an interface for eight-channel soundcards equipped with ADAT interfaces, providing a no-fuss means of getting analogue signals in and out.