MIDI-controlled mixing isn't new, but its latest incarnation from Fostex is both highly integrated and very affordable. But is it sufficiently familiar in feel and operation to win over a few more studio traditionalists to MIDI control? Dave Lockwood finds out.

Fostex's new DCM100 and the optional Mixtab controller together add up to a powerful yet easy to use and highly affordable MIDI-controlled mixing system. One DCM100 plus a Mixtab gives you eight stereo channels — however, up to three rack-mount DCM100s can be controlled from a single Mixtab, by way of a 3-position selector switch to determine which unit is to be addressed. With eight stereo signal paths in each DCM100, this adds up to what is surely the most compact means of controlling 48 mix sources. Unlike MIDI-controlled VCA mix automation systems, however, it is not just levels which can be automated; EQ, pan, and auxiliary sends and returns are all addressable in the DCM100.

DCM presumably stands for 'Digitally Controlled Mixer', but don't make the mistake of thinking that Fostex's new launch is a digital mixer; all audio circuitry is analogue, but it is digitally (ie. MIDI) controlled. The little Mixtab controller, although it looks exactly like a conventional audio mixer, in fact simply generates MIDI data — operate one of its controls and continuous controller data issues forth to change the appropriate parameter in the DCM100. The only connections made to the Mixtab (apart from the external 9VDC power adapter) are via MIDI cables.

Though the two parts of the system form such a tidy and convenient whole, the DCM100 rack unit which houses the DCAs (digitally controlled amplifiers) and audio connections, can, in fact, be used as a stand-alone device if you use something other than the Mixtab as a source of controller data; a sequencer perhaps. Many sequencer programs will allow you to generate controller data via on-screen faders, although Cubase's ability to create, and graphically represent, 'virtual devices' certainly gives it an edge here among ST software. However, some users undoubtedly find totally software-based access to the normally intuitive and tactile process of mixing simply too remote and alienating — witness the under-achievement of MOTU's excellently-conceived little MIDI mixer. For those individuals, Fostex's provision of a totally familiar hands-on interface in the form of the Mixtab may well tip the balance more in favour of some of the obvious advantages of MIDI-automated mixing.

CONFIGURATIONS

The system can work in three potential configurations:

• With the DCM100 controlled from a sequencer.
• With the Mixtab operating the DCM100 directly, in real time.
• A combination of the above, with the Mixtab as the control source, but operating via a sequencer which can record all controller data.

It is important to realise that the system has no dynamic memory of its own, (although the DCM100 can store 100 'snapshots' of its entire setup, for instant recall) and therefore only those configurations which include a sequencer will allow recording and replay of mix data, which is surely the whole idea of a system such as this. It is curious, therefore, that the manual makes so little reference to operation via a sequencer, with most of the examples given involving the Mixtab being directly interfaced with the DCM100. Come on, Fostex — a well-written manual should not only clearly instruct in the use of the product, it should also inspire the user with ideas that he might not yet have thought of for himself. Nobody can have a more intimate knowledge of the possibilities of a new product than the manufacturers themselves, but most users who have got as far as buying one of these will be several steps ahead of the manual already. This one also contains potentially confusing errors; you are told, more than once, to press switches "a second time", without having yet been told to press them the first time. In fact there is no 'first' press actually required in either instance. We all know that some people never read manuals at all (until they get into trouble, of course), but there will be others who like to learn all about their new piece of gear 'by the book', and they will find this sort of thing confusing.

THE PROCESSOR UNIT

The DCM100 processor is a rather deep 1U 19" rack device, with integral power supply and fixed mains lead; the review model came without rack ears, though they are presumably an option.

The front-panel, finished in an easy-on-the-eye pale grey (dare I say 'Akai-grey'?), to match the Mixtab, is spartan indeed, offering just nine rotary pots, a headphone socket, and a pair of LED column meters, plus a power switch. The ninth pot is the level control for the headphone outlet, which was disappointing in use, I felt; although it is specified for 8-50 Ohm phones, I thought it should still have made a better job of driving a high impedance set than it did. Many users will undoubtedly try to use this system with high impedance phones, and will be similarly confronted with a lack of level and early onset of distortion.

The other eight pots are used for gain matching, over the range -30dBV to +0dBV, allowing interfacing with a fairly broad range of nominally line-level sources, including electronic instruments (synths, samplers, drum machines, etc,) and effects devices. Direct input of passive instruments or microphones is not possible — obviously that would have required an extra front-end amplification stage, adding to cost whilst conferring no great advantage to the majority of users. If I have a criticism of the input stage, it is at the other end of the scale; in a situation where it is inconvenient, or even not possible, to lose excess level at source, you tend to find yourself running out of mixing headroom. A slightly wider range, allowing for lower sensitivity, might perhaps have been advantageous to some users. In the context of a typical setup, however, with a -10dBV tape machine and a collection of MIDI gear, no particular interface or operational level problems were actually encountered. I would suggest, however, that individual clip LEDs monitoring each DCM channel would be a major asset — if you begin with the input gain set for optimum sensitivity and then apply a liberal dose of EQ, the stereo bus will soon show clipping. However, it is at the input gain stage that attention is needed, and perhaps individual clip monitoring would help encourage the correct response.

The metering resolution, with just five segments covering the range -10 to +6VU, might seem a little on the coarse side; however, in practice the function of the metering in this context is surely primarily to display peak levels when setting up the input gain controls, and for that it is probably adequate. Each sensitivity pot controls a pair of input signals, for DCM100 channels are all stereo — ie. there are two separate signal paths within each, although they are governed by the same control settings. This is a sensible compromise, allowing the greater flexibility of being able to control up to 16 sources, without incurring the additional expense and greater complexity of implementing a full 16-channel interface. With so many MIDI instruments now incorporating 'stereo' outputs that are actually worth using in stereo, I'm sure this configuration will meet with the general approval of the target marketplace.

Round the back of the DCM100, this naturally translates into two inputs per channel, making for a fairly densely-populated rear panel, as all audio connection, in and out, is via unbalanced quarter-inch jacks. Connection to a channel's left input socket alone facilitates a mono feed to both sides of the stereo mix bus. The system caters for two auxiliary sends per channel, and these too are stereo, hence there are four outputs, accompanied by four further input jacks for the two dedicated stereo effect returns. A multiple DCM setup is almost certain to require aux send parallelling, in order that the same effects will be available to all signals, regardless of which actual unit within the system they are connected to. The DCM100 conveniently provides Aux Bus In sockets, allowing the effects sends to be simply cascaded from one unit to another. Similarly, the stereo bus is provided with Bus-In jacks for mixer parallelling — not necessarily just for operation with further DCMs, however, for the unit will happily work alongside any other -10dBV system you may care to partner it with.

The DCM100 has MIDI In, Out and Thru connections, with the In obviously accepting incoming data and Thru passing it onto other devices within the system. The Out socket is a little less obvious, however. This connection is necessary to allow the DCM100 to communicate its status back to the Mixtab, so that control positions can be compared when a snapshot is recalled. Where multiple DCM100s are being controlled from a single Mixtab, the output data from all the units must be merged (using an external merger) before being fed back to the controller.

Rear panel facilities are completed by an 8-way DIP switch, with the first four switches used to set the unit's MIDI channel. Where multiple DCM units are used they must be set to consecutive MIDI channels. Switch 5 determines whether the DCM100 will automatically execute a dump of all its settings every time it is powered up — when used in a system incorporating a Mixtab, you would probably want this most of the time so that you could match the control positions to the actual initial data values. What those power-up settings will actually be is determined by the setting of switch 6. This gives an option to choose either the default setup, essentially a 'zeroed' desk, ready for action, or alternatively you can effectively determine you own default setup, by having the system recall whatever you have stored as Scene 0.

Switch 7 sets the volume curve, determining the relationship between the control data and perceived loudness. The two choices on offer are Linear and Analog (sic), with Linear producing a direct decibel-MIDI data value correspondence, and Analog replicating the action of a conventional audio fader. The Analog position undoubtedly has the better 'feel' for normal mixing tasks although, oddly enough, the manual states that the default setting is the Linear position (the review model arrived set to Analog, so perhaps this has since been revised).

THE CONTROL UNIT

The Mixtab controller measures a diminutive 256mm x 256mm x 52mm; nonetheless, access to controls is not at all cramped — switches are large enough for the less nimble-fingered, and the faders have an acceptable travel on them. Even though the Mixtab is not, in itself, an audio mixer, but merely a source of control information, the control lineup looks familiar enough. Beginning at the bottom of the channel strip, the faders have a 45mm throw, and although the fader tops' are recessed into slots below the control surface, there is still enough grip to facilitate one-finger activation. There are eight channel faders, plus a single stereo master fader. A Mute button provides the standard channel-kill function, but pressing and holding a Mute button for a couple of seconds allows it to function as a solo switch, for it will then unmute the selected channel, whilst muting all the others — a neat implementation of a useful facility.

The channel Pan control functions conventionally on a mono input, determining signal amplitude across the two outputs, but with a stereo signal it acts as a balance control, biasing in favour of one side or the other. Panning completely over to one side on a stereo signal fully attenuates the other side.

AUTOMATED EQ

Moving onto the channel EQ, we find a simple 2-band, fixed-frequency system, with a shelving characteristic. What a shame they didn't manage to stretch to 3-band; how much more would it really have cost to produce? The upper band is fixed at 10kHz, whilst the lower operates at 100Hz. The EQ produces familiar and predictable results, letting you gently fatten up the whole bottom end, if you use it judiciously, though using the EQ for cutting rather than boosting proves, inevitably, slightly limiting. Rolling off above 10kHz can be satisfactory, given the gentle filter slope, but operating just a little higher would let you take out noise or harshness without producing a signal that sounds dull. Similarly, at the other end, when you roll off below 100Hz you will inevitably be left with something that sounds a little thin. Fine, if that is your intention, but the chances are that more often than not you will just want to take a bit of boom, or over-fullness out of a sound, without totally thinning it out.

If this seems to be a little harsh on what is undeniably a very straightforward EQ, it is because I see it as perhaps the most limiting factor in the system. Bearing in mind that the DCM's EQ is 'automated', there are considerable creative and corrective control possibilities on offer — or rather, there would be, given a slightly less basic EQ to work with. A third band, centred on 2.5Khz, perhaps, would make such a difference.

Although there is only a single Aux send pot on the Mixtab, you will recall that the DCM100 actually has two aux busses. Given that it is merely a control data source, the pot can double as the send to both, being assigned via a switch located just above. Both sends operate post-fader, which is the logical choice for effects usage, and each is actually 'stereo' (2-channel, to be strictly accurate).

Given that most effects units will simply sum the two sides together for processing, I am not convinced that this will have any real operational value to most users. It can also be the cause of the odd anomaly — if you're using both signal paths within a channel, but have only connected one side of the aux send (on the grounds that your processor is mono in), operation of the aux pot will be found to be dependent on the Pan setting, sometimes leaving you without sufficient level on the send, unless you swap the single connection to the other side. I suppose there can be no complaint about having the option of stereo sends, it's just that I really can't see this system being used with that many genuinely stereo-in processors. The auxiliary facilities are completed by two dedicated stereo aux returns, equipped with rotary level controls and the same 2-band EQ configuration as the input channels.

SCENE MEMORY

The concept of a snapshot, or scene-based memory system, recalled via MIDI Program Change messages, will be more than familiar to most readers from its implementation in the many popular MIDI-muting systems. The difference with the DCM100 and Mixtab system is that it is used to memorise the setting of every single parameter. It is important to realise, however, that you are not storing Mixtab control settings, but the DCM rack unit's internal settings, which may not necessarily be the same. The system does alert you to this situation, however, and helps you to take the necessary steps to remedy any discrepancies, using the same positional data feedback system employed for matching control settings after memory recall, as detailed below.

Control of the basic memory system could not be easier, with dedicated Store and Recall buttons, and an LED numeric display with Increment/Decrement switches. A digit-shift facility allows you to switch between the tens and units columns, reducing scrolling time drastically when moving extensively around the memory locations. Simply pressing (and holding for a couple of seconds, thus preventing inadvertent activation by the slip of a finger) the Store or Recall buttons will store or recall the contents of the currently displayed memory location — it's as simple as that.

OPERATING MODES

The Mixtab can be selected to operate in Direct, Preview, or Enable mode. Direct sends data from the Mixtab controls whenever they are moved, giving instantaneous control just like a conventional mixer. Once you have moved every control at least once, they are then all reflecting their true 'data position', and everything will be entirely predictable, with no sudden changes. However, any discrepancies between the initial memorised settings in the DCM100 and the actual settings on the Mixtab will, of course, produce discontinuities at first — for example, if the DCM has the master fader value all the way down, and on the Mixtab it is actually all the way up, the moment you move it in Direct mode, the DCM will recognize its new position, and jump instantly to full level. This is a problem that confronts the designers of all automated systems, other than those which use self-positioning motorised controls, or 'virtual' controls. Every control on the Mixtab has its own tri-colour status LED, which is very cleverly implemented as a data feedback device to address precisely this situation, but in Direct mode, the LEDs merely flash yellow to confirm activation of a control.

Switching to Preview mode and recalling a scene displays the Mixtab in all its multicoloured glory. The control status LEDs indicate where discrepancies exist between the memorised data and the actual control settings by flashing in either red or green; red indicates that a control is set too high, and green that it is too low. The speed of flash also indicates the distance from the target value; as you approach the correct position, the flash rate increases, the LED extinguishing altogether when you arrive at the target position. The whole mixer can be reset very quickly, as the process soon becomes intuitive.

I find the colour association (red for high, green for low) very logical, but would personally prefer the faster flash rate to indicate the greatest distance from the required setting, slowing down as you get nearer.

In Preview mode the Mixtab controls no longer transmit data — you need to switch back to Direct operation, after control matching, to resume mixing. Enable mode, however, is the logical combination of both these two — controls are in Preview status, until they are matched to their actual data value, whereupon they switch automatically to Direct control, confirmed by their illuminating in yellow, rather than extinguishing, at the null point. Enable mode allows you to operate much more instinctively, without having to constantly think about mode selection.

DATA THINNING

One of the biggest headaches with MIDI continuous controller data is reaching a compromise on exactly how much data you actually need. The more finely incremented it is, the finer and more accurate the control resolution will be, but you also run the risk of producing data at such density that the system will not be able to keep up with it, particularly in a multi-controller situation, such as presented by the Mixtab — data lag will occur, and events may well be shifted in time by other events occurring simultaneously. Conversely, send too little data, by increasing the increment size, and the result may be audible 'steps' in what was supposed to be a smooth fader movement. Fostex have chosen to offer a user determined data thinning rate, rather ambiguously tagged as a Smooth control, with Fast and Slow settings. The idea is that if you know you are going to be moving a control very quickly, you should set smoothing to Fast, so that the data will be thinned to a manageable level. Less rapid movements, on the other hand, can be performed at maximum resolution, in the Slow position.

My experiments with the Smooth control produced some curious results. Moving a fader as fast as possible over its full travel, with smoothing set to Fast (maximum thinning), I found it was possible to get from 0 to 127 with just two intermediate steps. The same rapid movement, with Smooth set to Slow (which I presumed would create a system-choking flood of data), produced exactly the same result: just two intermediate data increments. Reasoning that there might be some overriding 'protective thinning' designed in, to prevent precisely the kind of lunacy I had just indulged in, I repeated the comparison using a precisely-timed slow fade: same result. Both Fast and Slow positions produced the same volume of data; 0 to 127, in about fifty steps this time. The only conclusion I can draw is that, unless this is a rogue unit, the controller value is in fact incremented at the same rate, regardless of the Smooth setting. I was unable to clarify this with Fostex before going to press with this review.

The obvious next step was to see whether this had any detrimental effect on the performance of the system in a 'worst-case' mix of the utmost complexity. For my tests, I configured the system firstly with the Mixtab/DCM100 having a MIDI port all to itself, and then subsequently with it having to share with Note data. I must admit I was quite surprised to find that, with all channels as active as would be conceivable in any 'real' situation (and fed with sustained note signals so that anomalies would be readily audible), nothing at all in the way of undesirable side-effects occurred, until I exceeded an apparent 'critical rate' of movement. The system coped perfectly well, in either configuration, with normal mixing movements.

Entering some timing-critical mutes, with all the other channels' information turned off, and then turning the general flood of data back on again resulted in some time shifting (when controller events were on the same buss as note data). I must stress, however, that this is the result of the way in which the sequencer is organized to assign priority; although different sequencers will have their own data priority algorithms, they will all, quite correctly, place a higher value on notes than controllers. If at all possible, it seems one should try to configure this system with a port of its own, for optimum timing accuracy. Mute activation does generate a significant degree of switching noise if you interrupt a signal with it, however, as it is not 'ramped' in any way.

Apart from time-shifting, I was looking for audible discontinuities — a 'stepped' effect, or audible 'clicking', in what should have been a smooth fade. Very rapid movement of faders or the bass EQ controls does seem to produce this effect, regardless of the general level of data density. If you restrict yourself to the sort of control movement rate involved in the majority of normal mixes, however, no problems are evident at all. I suspect that there will be two distinct attitudes to this situation — some potential purchasers will test the system by zapping faders up and down as fast as possible, looking for noise, and will dismiss it once they hear what they are expecting; others will take a more reasoned view, deciding that the creative possibilities of normal usage far outweigh the occasional restriction.

Obviously the data-density situation would be exacerbated by the use of three DCM100s in combination, but I only had one unit on test, and cannot, therefore, verify performance in a larger configuration.

CONCLUSION

This is an exciting product, not because it is new — there have been entirely MIDI-controlled mixers, and dedicated MIDI-controller sources before — but because it works well at such an affordable price. It is precisely because it works well, just as it is, that I find myself wishing that it was just a little more sophisticated — offering slightly more sophisticated EQ, perhaps, and maybe one more auxiliary send. The essentials of decent mixer performance are all there, and it is conspicuously clean and quiet, subjectively introducing no undue noise or degradation into the signal path, once gain levels have been optimised.

The absence of insert points on channels, and particularly on the stereo buss, would seem to militate against the idea of the DCM100 as the sole mixer in a system of any sophistication. With only line-level signals catered for, however, it would be possible to patch any individual processing required directly between the source and the DCM input, though patching, say, a compressor on the output side, after the master faders, would be less satisfactory. With any significant degree of gain reduction applied, the compressor would tend to fight against the start of a fade out, and would be left ultimately with no input, but feeding its full output level into the following stage — not exactly optimal in terms of noise.

The DCM100 system, either complete, or the rack processor alone, will surely be attractive, not just to the small system user who might want to automate a few audio signals alongside his already automated MIDI sources, but as an add-on to larger systems. If you already have an automated desk, running alongside a MIDI sequencer locked to tape, but find you never have quite enough automated channels to take care of all the effects returns, this is certainly an option to consider. If you could afford the luxury of having a DCM 'floating' rather than dedicated to a specific task, you could patch one in for its automated EQ alone (this would be a lot more tempting if they had managed to make it 3-band, of course). Had Fostex also managed to provide a direct output from each channel, I think they would have been certain to sell bundles more units, for it then would have been the perfect device to add a limited degree of automation to an otherwise unautomated mixer, via insert points. DCM channels could then have been patched individually across a few key signals where automation would be most valuable, like a conventional 'add-on' automation system — but better, in that it could offer more than just level control.

The DCM100/Mixtab combination achieves what it sets out to achieve. It's an excellent facility for the money, but it raises a few questions at the same time, not least of which is whether MIDI is really the right control source for any form of mixing automation. The obvious next step is to a larger, more sophisticated mixer, with a similar level of control, but will that actually work? Will not the MIDI bandwidth ultimately be taxed to the point where even a multi-port system can longer cope?

The fact is, if you want a no-compromise automated mixer, you must be prepared to pay a no-compromise price for it. However, if you are prepared to work within the few practical limitations imposed by the DCM100/Mixtab system, it offers a very neat and cost-effective solution indeed.

Further information

DCM100 £449; Mixtab £299. Prices include VAT.

Fostex UK, (Contact Details).

Delving inside reveals a large main PCB, covering most of the case area, with three sub-boards, and a surprising amount of point-to-point wiring. A flexible metallised screening pad (removed in the photo here) lies above the main board where the audio wires cross over it. The audio connectors have a sub-board all of their own, securely mounted to the chassis — good to see this, since it will withstand frequent reconnection without the fatiguing of joints frequently found in inadequately supported boards with connectors on them. All ICs are soldered, except the main operating system ROM (V1.02 in this case), which is socketed, and mounted, slightly inconveniently for access, half underneath the MIDI-DCA interface board. I am not sure if the review model is from a production batch or not, but presuming it isn't, I would be surprised not to see production models a little tidier inside than this.