This little beastie has succeeded in generating almost as much speculation as R-DAT since it was first shown last year, first because it's a small, mass-produced digital mixing console, second because it has moving faders, and third because it doesn't cost an arm and a leg (and a head and another leg...). I speculated last month that a tarted-up DMP-7 might serve the professional digital mastering buff in place of a more expensive model, and this indeed is the premise under which a certain dealer suggested that they might install digital inputs and outputs.

Some water has gone under the bridge since then, and I have had a chance to review the DMP-7 with the professional recording studio in mind, especially considering the potential role of the console in CD mastering, after which it seems that there may be certain limitations to using the console in applications of this kind. Interestingly, this opinion is confirmed by the decision of the aforementioned dealer not to proceed with the installation of professional digital interfaces.

In the role for which the DMP-7 was designed I have no doubt that it will score considerable successes, and there are definitely several applications within the studio for which it would be well suited. My conclusion subsists mainly in the fact that 'digital' in this case is the means to one particular end, the ease of implementation of effects, assignable controls and remote MIDI automation, rather than the road to exceedingly high audio quality.

SYSTEM ARCHITECTURE

As I pointed out last month, 16-bit conversion is only satisfactory if the analogue signal is carefully placed in relationship to the range of the convertor. This means that it is important for the highest level produced by the sound source to peak near the top of the input range (in order that the maximum number of bits is used), so that quantising noise is kept to an audible minimum. The DMP-7 provides a single assignable input level meter for this purpose (showing from 0dB down to -42dB), together with trimpots on the rear panel next to each input jack, which run from -20 to +4dBm to allow a range of devices with different output levels to be connected.

It is extremely interesting to notice the place in the input stage at which the meter measure the input level (see fig 1 page 45). You will see that the input level meter comes after the EQ! In order to see a true indication of how fully the convertor range is being used, this meter should come directly after the A/D convertor, whereas in the DMP-7's case the meter is measuring the level of the post-equalised signal.

The problem, as I see it, is that it would be feasible for input signals to be converted at relatively low resolution, while appearing high up on the meter if a large positive EQ setting were used.

Conversely, if the input level were to be set with EQ switched out, such that the highest peak just reached the top of the meter (thus using all 16 bits, one assumes), a positive EQ change would take the meter over the top.

The fact that the meter is here makes me somewhat suspicious about the available internal digital headroom, because if there was no problem I feel sure that this input level meter would have come directly after the convertor. My information from Yamaha's engineers is that the internal mix buss resolution is 24-bit.

As it is, you can't have your cake and eat it; either you set your input signal some 12 to 15 dB below the top to allow for the addition of EQ without re-adjusting the gain of the channel (this would mean that you wouldn't be making the best use of the convertor), or you optimise the input signal so it reaches the peak input level (giving the lowest quantising noise), and allow for the possibility of problems when adding EQ.

Further tests performed on the DMP-7 reinforced my worries about level setting and mix buss headroom, because it appeared that if several inputs were used, each with EQ and optimised so that the highest level signal came near to the top of the input meter range, then the point at which clipping was audible tended to come lower down the input meter the more inputs were added.

At no time was the output clipping, because there are meters just below the D/A convertors on the output and these were kept well below the top. This depended partly on the position of the channel fader, which, as you can see on the diagram, is after the input meter, and would affect the range of the digital signal on the mix bus.

AUDIO LIMITATIONS

It is difficult to compare this situation with that found on an analogue console, because there would not be the problem of quantising noise to contend with, and it is not yet clear how Yamaha has structured the relationship between analogue and digital levels in the DMP-7.

From initial impressions, though, it seems more easy to run into the audio limitations of this console than it would be on an analogue machine. Quantising noise or distortion is only present when an audio signal is present, and thus cannot be measured directly as 'hiss' can be measured in an analogue console. It will appear to make the signal more 'dirty', as well as sometimes producing noise modulation on a signal.

The only option would be to appear to set all the inputs well below the top, to allow for greater flexibility in mixing without digital clipping. This is perhaps intended by Yamaha in a subtle manner, as the input level meter is green up to about half-way (-24dB), orange from there to about -6dB and red above this, an encouragement to keep the signal in the green, thus converting it at a higher quantising noise.

The DMP-7 is really an assignable console: there is only one set of controls (except for the faders). On each channel is a SELECT button which assigns the controls to that particular channel, after which parameters can be edited in DX-7 fashion on an LCD display using either the DATA ENTRY SLIDER or UP/DOWN buttons. Parameters which fall into this category are such things as EQ settings in three parametric bands (which have a very wide range of frequencies for each band), and panning (in 16 steps left to right).

The EQ allows for variable 'Q' from 0.1 to five and for the high and low bands to be either peaking or shelving, as well as up to 15dB of boost and cut. As the cursor is moved between the parameters on the display, the data entry slider takes control of the selection, allowing the user to hear the effect of changes.

I found that small dynamic changes of the EQ were relatively smooth, but that larger changes tended to result in a jump from one level to another. This was partly hindered by the fact that the data entry slider has a very small amount of travel, so that tiny movements cause large changes in settings, although these can be trimmed with the incrementing controls. It is possible to turn off the EQ for each channel by toggling the relevant buttons, although this has to be done three times, once for each band.

FADER FLIP allows all the channel faders to become send level controls for either SEND 1, 2 or 3. Their motorised nature is useful in this respect, as they can jump to indicate the relevant setting as each different send is selected. Two of these sends are reserved entirely for internal effects (generated by three individual digital effects sections within the mixer), while the third can be sent out to an external device via a D/A convertor.

Send levels can be monitored on a single assignable meter, and the send currently being displayed/adjusted is indicated with large lights at the top of the console indicating SEND 1, SEND 2, SEND 3 or Channel.

On the return side, again one and two are reserved for internal use, while three is switchable between an external analogue stereo input and the internal effects buss. Amazingly, EQ is provided for the two analogue return channels as well!

The internal effects supplied are, as might be expected, very comprehensive, offering similar but sometimes enhanced facilities to the familiar SPX-90. Considering that to buy the effects alone would probably cost between £1500 and £2000, the DMP-7 shows itself as a bargain.

MIDI CONTROL

It is in this area that the console will appeal to the modern MIDI-fied studio, because, as you must have heard by now, virtually every parameter on the console is available to be controlled via MIDI and every parameter is available to the MIDI output. This can be implemented in several ways.

The MIDI spec, allows for several 'continuous controllers' (of which the modulation wheel and foot pedals are examples), up to a maximum of 128, although some of these are standardised functions and some use two controller numbers per physical controller for sending high resolution information.

Any of these controller numbers can be assigned to any control on the console (mod. wheel to SEND 1, channel 1, for example) so that a musician could vary certain points about the mix for the instrument he was playing. It's also possible to assign particular note numbers to parameters, in which case it is the velocity value of the note which changes the setting of the parameters.

I tried this out by assigning a note to a channel's EQ setting, which seemed like a good idea because I wanted the sound to become brighter (more high EQ) the harder I hit the note. As it happened I'd chosen the only parameter which doesn't react very fast to control changes via MIDI, and the effect didn't take place, and that it has been decided that EQ is the lowest priority of all. This was certainly true, because assigning the mod wheel to the same high EQ setting still caused about four seconds delay after having moved the wheel before the effect took place. In all fairness, though, if this console were to be used with synthesisers it would normally be possible to introduce velocity-controlled filtering on the synth itself.

Other controls, such as PAN, reacted much faster to MIDI control, with almost instant effect, which makes me wonder why the EQ is so much slower than the rest. The motorised faders are reasonably zippy, being quite fast enough for most applications in which this console would be used, although they became a little less sluggish when asked to respond to note velocities (which is possibly an unrealistic thing to expect anyway).

Dynamic changes of any setting are available as continuous controller data or note data at the MIDI OUT. In other words, any movement of a fader, EQ setting, send level, etc, would result in a new value for that controller being sent over MIDI. Exactly how many times per second these are updated is not entirely clear, but there seems to be no obvious problem with 'staircasing' effects on most music.

It is important in cases such as this to take care that the MIDI buss is not overloaded with data, as it would be when a large number of controllers were moved together, and which could be achieved by multi-layer sequencing even if not physically on the console (due to the assignable nature of the controls).

For this reason, it would be advisable to synchronise either a separate sequencer (which could be fairly basic) purely to store console automation data and to spit it back out over a dedicated MIDI cable, or to use a sequencer with a number of separate MIDI outputs, connecting one exclusively to the console. I suspect that overloading would only become an obvious problem in large only become an obvious problem in large systems using all 16 channels, with large amounts of performance data flying around.

(Click image for higher resolution version)

STATIC MEMORY

If dynamic change is not a must, it is far more sensible to store changes in the console's own internal memories, which are in the form of snapshot stores for every setting of every channel. These can be recalled with virtually instantaneous change in the audible result, due to the fact that the processing is all internal, as opposed to the delay involved in sorting out reams of MIDI data from outside. Subsequently, any of these snapshots may be recalled either from the front panel, or using MIDI program change commands (a single message which can be given a high priority).

With the addition of a RAM cartridge, up to 97 memories are available to the user (30 without the RAM cartridge), and the changeover between each snapshot is made satisfactorily free of noise.

If required, a cross-fade can be instituted in-between each change of snapshot lasting up to 10 seconds, but I found this only to operate on the fader settings, and not on any other parameters. Thus if EQ were to have changed between two snapshots there would be a subtle cross-fade of fader levels coupled with a rather unsubtle jump in the EQ setting at the halfway point of the cross-fade.

Partnering the DMP-7 is a useful microphone amplifier called the MLA-7 which sports phantom powering (a feature often omitted on budget desks), and eight XLR inputs with a gain control for each input. I wasn't able to test this at the time of the review, but the specification for equivalent input noise is very good which bodes well.

In any assessment of the DMP-7 one cannot but remember its price!

Certainly no one could accuse Yamaha of not providing good value for money in this product, and it is only excited onlookers like myself who have held out wild hopes that this might be the answer to a lot of people's prayers, speculating about the console's viability in CD mastering studios and so on.

To be fair to Yamaha, they have never made such claims, and thus should not be criticised for not having quite lived up to the industry's expectations!

The main problem has been that people have seen the magic word 'digital' on the DMP-7 and equated it with immediate perfection of every kind; this is hardly likely at a price of a few thousand pounds.

SIGNAL QUALITY

There are a number of areas in which I believe that limitations would be encountered when using this console in a demanding professional environment.

These lie mainly in the points which I have made earlier about difficulties in setting up audio levels in the system, and the clear possibilities for either clipping or excessive noise to be incurred.

In terms of audio signal quality and audio flexibility (being able to change signals, EQ, and levels with reasonable freedom from worry about reaching the limitations of the equipment), I believe that one could still be better off with a good analogue console. The compressor on the output of the DMP-7, for example, is very limited (no pun intended) and distorts when driven hard.

There's no control over its characteristics, except a RATIO control. Nevertheless, there are most definitely exciting applications for the DMP-7, especially as a submixer for a number of MIDI controlled devices (a drum-sampler with multiple outputs, for example).

In such an application as this each channel could be optimised with the EQ and the level settings desired, and using the myriad effects which are available internally, providing a stereo sub-mix of percussion with effects changes, programmed mutes, etc., being controlled from a sequencer.

BARGAIN

It still remains a bargain despite the problems revealed, and it is clear that the use of digital rather than analogue technology in this machine was more for reasons of operational flexibility than one of audio quality.

It is easier to implement all the assignable functions, MIDI control and digital effects in the digital domain than in the analogue domain).

Moreover, it shows what is possible with digital mixing technology, and what Yamaha have clearly done in developing customised VLSI's for digital audio processing.

If this were applied to a full-blown professional device then a few other companies would have to take some serious deep breaths.