A unique problem-solver from Roland combines line mixing, sample rate conversion and a USB audio interface in a single rack space.
With digital audio de rigeur in most home studios these days, getting everything to interface and work together can be a bit of a problem sometimes. We often have to connect stand-alone analogue and digital audio equipment together, and frequently we have to find ways of interfacing both with computers too. This can preset a variety of complex problems, often with costly solutions. However, Roland have recently made life a little easier by releasing their new M1000 digital line mixer. This unique product is deceptively simple, but is actually very cleverly designed and superbly flexible, making it useful in — if not indispensable to — just about any installation that involves a mixture of analogue and digital audio equipment, or where signals need to be exchanged between these two domains and computers.
So, you'll now be wondering just what is it about the M1000 that makes it so indispensable. Essentially, what we have here is a straightforward stereo line mixer, but with both analogue and digital I/O, USB interfacing and some neat facilities. It has one unbalanced stereo analogue input connected via a pair of quarter-inch jack sockets, but without any cross-normalling to accept a single input as a dual-mono source. There are also four digital S/PDIF inputs connected on phono sockets, and the first can be switched to accept an input from a front-panel Toslink optical port instead.
Each of these five stereo input channels has a rotary level control and the four digital inputs also have Lock LEDs to indicate the presence of good data streams. The level control for each channel — including the analogue one — occurs in the digital domain with a range extending from unity gain all the way down to off. Given the ease of basic signal manipulation in the digital domain, it is a little disappointing that there are no other facilities provided here — no channel swapping or selection, phase inversion, balance adjustment, or M&S matrix conversion, for example. In fact, the only other input control is a slide switch to select the analogue input sensitivity.
The only way to adjust the analogue signal level prior to the A-D converter is to switch between -10dBu and +4dBu sensitivities. That's not a typing error by the way — this Roland unit really is offering a -10dBu sensitivity instead of the more usual -10dBV semi-pro reference level. This means there is a 14dB differential between the two settings instead of the (nearly) 12dB between -10dBV and +4dBu. I don't think anyone need be concerned about a couple of decibels in this case, but it is as well to be aware of the issue, because meters might not quite line up where you expect them to. In practice, then, optimising the analogue input level and headroom will be down to adjusting the output level of the source.
The M1000 provides both analogue and digital outputs, with a rotary output level control that enables up to 6dB of gain to be added at one end, and completely mutes the signal at the other. Unfortunately there is no detent at the unity gain calibration point (three o'clock in the rotation), but the concentric balance control does have a centre detent. A simple output level bar-graph meter is provided spanning a 40dB range, with the first five LEDS indicating headroom in 6dB steps. The analogue stereo outputs are presented on a pair of balanced XLR sockets with a nominal +4dBu operating level, while the digital outputs are dispatched via a rear-panel phono socket and a front-panel optical port. I was impressed that both the input and output optical ports include automatic flaps in the sockets to help prevent dust from collecting — a rather thoughtful feature.
A second unbalanced analogue output (via a pair of quarter-inch sockets with a nominal -10dBu operating level) is intended for monitoring purposes, and follows the main mixed output. It has an independent level control which is scaled simply from zero to 10, and this also determines the volume of the built-in headphone amplifier — a stereo quarter-inch socket mounted on the front panel allows standard headphones to be connected.
Enhancing this basic digital signal path, each of the four digital inputs has its own independent sample rate converter. The first input channel can accommodate source rates anywhere between 32kHz and 96kHz, while the other three will only recognise sample rates close to the nominal standards of 32, 44.1, 48, 88.2 and 96kHz. Thus the first input can be used for true variable-rate sources (including digital varispeed), while the others are useful for standard rate conversion duties and where sources cannot be synchronised to the local word-clock reference.
All the digital inputs and outputs support fixed 24-bit operation, and the internal signal processing is performed with 56-bit resolution. It is a shame that there is no facility to dither the outputs to 16-bit resolution, as I believe a lot of potential users would like to hook up minidisc and CD-RW recorders to the M1000 output. If you're using this unit with a computer, it is possible to apply the appropriate dithering using your computer audio workstation and USB interface software, although this is a clumsy workaround.
- Roland M1000 MME driver v1.00, WDM driver v1.00
- 850MHz Pentium 3 PC laptop with ASUS motherboard and 512MB RAM, running Windows 98SE.
- Tested with Syntrillium Cool Edit Pro v2.1 and Steinberg Cubase VST v5
At this point it would seem appropriate to talk about the USB port which is tucked away on the rear panel of the M1000. This enables fully bidirectional audio data transfers with a PC or Mac, and thus enables the M1000 to act as a complete input and output interface for computer audio workstation software. However, the supplied drivers must be installed before the highest-quality data transfer facilities become available.
There are two modes of USB operation. The standard mode transfers 16-bit data at 44.1kHz or 48kHz sample rates while an advanced mode enables transfers of 24-bit data at up to 96kHz sample rates. The standard mode essentially employs the standard USB drivers built into Windows ME/98, XP/2000 and Mac OS v9.04 and above. However, to benefit from the high-resolution advanced mode a bespoke ASIO driver needs to be installed. There are dedicated versions of this driver for Mac and Windows platforms, and for Windows XP/2000 users there is a further option to choose between a WDM driver (primarily for Sonar users) and an MME driver (for almost everyone else), but these two alternatives cannot be installed simultaneously.
If the USB port is being used, the fourth digital input is automatically switched away from the rear-panel S/PDIF input socket (and SRC) and instead handles the output signal from the computer. Unfortunately, though, there is no front-panel indication that the fourth input has been reallocated in this way (other than the Lock LED being illuminated). That said, I would imagine it would be a set-and-forget function for most users.
To prevent a howlround between the USB output and input within the M1000, there are two separate mix busses. The first carries the analogue and first three digital inputs and is routed only to the USB output. The second carries all five inputs (the fourth digital input being the USB return, of course), and is routed to all the main analogue and digital outputs, (but not the USB port, obviously). Thus there is no way that the USB return can get back to the USB send within the M1000 itself.
An interesting additional feature is that, when the USB port is in use, the analogue input can be deselected from this second mix buss. (If the USB port is not in use the analogue output is passed through the second mix buss automatically.) This option might initially seem odd, but the idea is to prevent a possible double output of the analogue recording source — once direct through the M1000, and a second time via the USB send/return or some other external loop. To select or deselect the analogue pass-through when the USB is active, a rather obscure switching process is used. Pressing and holding the Clock Source button for more than a second engages the analogue pass-through, while breaking the direct through-path involves pressing and holding the Sampling Frequency button. The clock rate LEDs flash to confirm the appropriate action has been taken, but sadly there is no permanent indicator to remind the user of the current signal path status. However, the selected mode is remembered after powering down so that the next time a USB port is used the M1000 reverts to its previous operating condition.
The M1000 is a 1U rackmount unit (although the rack ears can be easily removed, if necessary) and measures just 167mm deep with a weight of 1.9kg. It has a fixed-voltage internal mains supply with an IEC mains cable inlet. Included with the unit are a USB cable (A-type to B-type sockets), a CD-ROM containing the various computer drivers, the owner's manual, and some stick-on rubber feet if you don't wish to mount the mixer in the rack.
The unit is constructed well, with a steel chassis containing a large and surprisingly heavily populated circuit board. The predominantly surfacemount components include a Motorola DSP, AKM A-D and D-A converters, and Analog Devices sample rate converters, with a considerable number of other 'big chips'. A smaller PCB is mounted on the front panel to carry the indicator LEDs and audio meters, but all the controls and connectors mount directly to the main PCB.
There is one aspect of the construction which concerns me greatly, and that is the absence of any insulation around the mains inlet wiring on the back of the IEC connector (see picture). While there is no immediate danger to life, as the mains terminals are well away from both the earthed case metalwork and each other, I consider this lack of insulation a poor and potentially dangerous practice. The vast majority of equipment manufacturers would have used cable sleeves, or a rubber boot over the socket itself, to prevent any possible risk of electric shock, and I feel Roland's lack of attention to this important detail is disappointing.
I initially used the M1000 as a stand-alone monitoring controller for a variety of sources, with its monitoring output feeding some active speakers and this worked very well indeed — it's nice to have the luxury of input SRCs to take care of the unsynchronised digital sources. I also used the unit to extend the monitoring section of an analogue console and to provide headphone monitoring while doing some on-site SADiE editing. It also served usefully as a 'tape dubbing' facility feeding one analogue and one digital master recorder.
It initially seemed odd to me that the analogue input level is controlled in the digital domain. If you allow the A-D to become overloaded, turning down the level control only reduces the volume of the aliasing distortion, not the distortion itself! There is a peak LED associated with the analogue input, by the way, so, although there is no input meter as such, overloading the A-D is made obvious. However, with a maximum input capability of +22dBu (at the +4dBu nominal input level setting), this is unlikely to be a problem.
In fact, many users might be concerned that they cannot drive the analogue input enough to 'use all the bits'. The provision of a -10dBu input level switch helps in this regard, since you can reduce the amount of headroom by 14dB, but with a 64-times oversampling 24-bit converter, leaving 12dB or so of analogue headroom does not affect audio quality. Roland quote an overall signal-noise ratio in excess of 102dB (IHF-A weighted).
When mixing analogue and (already mastered) digital signals together, the latter will have almost zero headroom and the former will have a lot. Consequently, I often found it necessary to turn down the level of digital sources by 12-15dB in order to balance the signals properly against an analogue one. This lost level could be partially reclaimed at the output stage using the +6dB of gain available there, but I would ideally have liked 12dB of gain in hand.
After loading the appropriate standard and advanced drivers into two of my PCs (one running Windows 98SE and the other Windows XP Home), I was able to hook the M1000 up and use it as an interface, mainly with Cool Edit Pro v2.0. I had no trouble loading or configuring the drivers, and everything seemed to work well. If loaded, the advanced drivers are used by default and the analogue pass-through is switched on — which suited my way of working anyway. I did a little multitracking in Cool Edit Pro and found no latency problems — I was able to play along with myself quite happily — although there are provisions on the driver setup to adjust buffer sizes and processing offsets to help fine-tune the system if required.
Overall, I found this to be a very handy little box, often providing a useful means of solving an interface or monitoring problem. There are aspects of its design that could be improved and expanded to make it even more versatile, but at its current UK asking price it already represents something of a bargain. It would have been nice, though, if there had been balanced analogue inputs with a mono-normalling facility for single inputs. It would also have been good if there had been some input configuration facilities to handle M&S sources, to swap digital channels over, and so forth — real problem-solving tools.
In terms of pure audio quality, I didn't find its performance to be quite up with the best in a purely digital domain — the penalty of using SRCs rather than synchronous digital interfaces — but the upside is its flexibility when dealing with unsynchronised sources, of course. As a stereo USB interface to a computer, it appears to work well, and with ASIO and Direct Sound compatibility, plus 96kHz/24-bit audio paths, it should suit just about any software platform.
It's always easy to point out the limitations of a product, but I'm sure most people reading this will quickly see a way in which the M1000 could extend the functionality or versatility of their own equipment installation, and I would certainly recommend it for personal investigation. This is a unique product, and one which has already found a permanent home in my studio.
With any digital device, the issue of clocking soon rears its ugly head. The M1000 is unusually competent in this area, with options to derive the unit's clock source from its internal crystal (96, 48 or 44.1kHz), the first digital input, an external word-clock input, or the USB data stream — the last with 96kHz send and receive modes to improve clocking stability with high-resolution signals. A word-clock input is accessed through a BNC on the rear panel, complete with a switchable 75Ω termination. There is also a word-clock output BNC socket which can be switched to a 'thru' mode which directly outputs any signal presented on the word-clock input. With the 'thru' mode switched off, the clock output signal is derived from the internal clock, whatever that may be slaved to. Thus, when linking clocks through the M1000, you effectively have the option of a straight through path, or a regeneration path — and the flexibility of being able to choose is often extremely useful.