Someone asked me what MIDI stood for the other week. I gave the usual reply. "Nope," he retorted, "it means 'Many Idiots Desire It'." A bit harsh, I thought, but I can see how he feels. This 'universal' connection system for synths has been pounced on as the greatest single advance for the keyboard player since God invented the thumb.

Certainly it's an excellent and creative tool, but we've almost reached the point where the synthesiser itself doesn't matter. It's merely a convenient case to put around two five pin DIN sockets. You got MIDI, you can fix anything, a bit like Evostick.

Here's a new definition: "Makes Italians Do It".

Both the Opera 6 polysynth and the matching Expander module come from the land of the bruised tomato. The Expander is the latest release from the Siel company. The Opera was launched in Britain last year. The former is essentially the latter without keyboard or controls. Within the free standing 23in wide box are 12 DCOs (six voice polyphonic), six 24dB/octave low pass filters, three LFOs, six ADSR envelope generators, and 95 factory programs.

It links to the Opera via two MIDI cables, but of course could be connected to any suitably MIDI-ed synth. All you see on the front panel are ten small, round pushbuttons that call up the memories, a two figure LED display, master controls for volume, tuning and detune for the second DCO bank, plus a section for interface mode which we'll arrive at in a min.

The purpose of the Expander is to enhance the sound of the Opera 6 – either adding two more oscillators per note on the same sound, or contributing an entirely different voice for layered effects. When both boxes arrived, my immediate thought was, does the Opera 6 need expanding? Better look at the keyboard on its own to begin with.

It boasts a five octave, C to C, touch sensitive keyboard putting it one step ahead of the present round of under-a-grand, analogue polys at least in terms of expression. The dynamic effects can be assigned to the ADSR (the harder you hit, the louder the notes), and the attack time (strike fast for percussive effects, press gently for slow fade ins). The degree of dynamic sensitivity is pre-set and can't be adjusted.

The Opera is a simple synth – 'uncluttered' would be the generous term. You've a choice of ramp or square (pulse width) waveforms, or both and they operate at footages 16, 8 or 4. The oscillator banks can be detuned for chorusy effects, but can't be separated by much more than a tone which wrecks the idea of intervals. Shortsighted, that.

There's a single ADSR envelope generator that caters for the VCF and VCA (two would be better), the filter has an unusually high range (very bright at the top) but complains when you attempt to use it as a third oscillator. You might send it into oscillation by turning up the cutoff and resonance but it tracks uncertainly across the keyboard, occasionally slipping out of tune.

There is a strong white noise source, and the usual master tune and volume controls, plus two, sloppy modulation wheels for pitch and vibrato.

The programs are selected by a bank of the same, small push buttons that appear on the Expander. They're numbered 0 to 9, but once you've dialled in a new choice – say 27 – nothing happens until you press a red button marked enter, then the Siel accepts its instructions.

Most of the remaining switches have several functions. For example, for waveform selection, push the switch once and you get ramp, a second time and it moves to square, on the third tap they're both on, and for the fourth they both go off. Saves on switches, but you have to make repeated attacks to travel from the current set up to the one you desire.

The final mystery is the interface mode – two lights, one button, again with four positions. In practical use it allows you to call up one program on the Opera and make the Expander jump to the same number, or select one program on the Opera and a different option on the Expander – a versatile, convenient technique.

The factory settings are hardly dazzling examples of how the Opera performs. The Italians have opted for thin, reedy sounds, and gone overboard with the modulating LFO. The benefits of pulse width mod have largely been ignored and if you write this into the patches, the Siel immediately sounds fatter.

Found a neat trick while I was playing around with this. Both the DCOs have pulse width controls to alter the mark/space ratio of the square wave. At the extremes of their travel, the waveform becomes so thin it disappears. If you set one DCO close to this limit and the other slightly further back then modulate both, they slip in and out of hearing at different places. With a touch of detuning and a couple of low notes, the results 'grunge' around in the lower registers – very Aboriginal. That's my thought for the month!

The Opera 6 is a basic machine. I was disappointed that I couldn't draw more from it. A few extra tricks such as reversible envelope generators, a sub oscillator, fast ways of shifting the keyboard up or down an octave and a chorus unit are generally standard on cheaper Japanese models. So are arpeggiators and even small polyphonic sequencers, albeit programmable in step time. The Opera featured none of these.

If you're after soft, full string or brass sounds and bright pianos, okay, but the Italian Stallion goes little further than that... no synced up growling sounds, for example, and it won't convert to a mono synth for bass lines. Less than full marks, I think.

And so to the Expander where, obviously, the same comments about sound apply since the circuitry is identical. One variation, however. The programmers have been more 'Japanese' in their approach, serving up fatter, brassier and lower register portions.

They've also chosen the patches to complement those already in the Opera. For instance, the low strings at 32 on the Opera will be augmented by a suitable brass on 32 in the Expander. Or a punchy Clav will have strings fading in slowly underneath.

The touch sensitivity also comes in handy. It can be programmed individually so the Expander might be giving you a straightforward organ voice, unaffected by the touch sensitivity, but if you hit the keys harder, the Opera will then jump in with, say, a saxophone blart.

To my mind, editing the Expander is cumbersome and wasteful. The MIDI link allows you to swap programs, but you can't influence the Expander via the Opera's front panel controls. You'd have to dump the Expander's programs into the Opera, carry out the work there, then shunt them back into the Expander again. Where does that leave you if you've got a non-Siel MIDI-ed synth, but you'd still like to buy and apply the Expander? I suspect you'd have problems making fine adjustments.

Even so, the link up does function impressively. The Siel duo can hit the extremes – soft and furry underneath, bright and tinkly on top – or deliver one thick wedge of sound. It's here that the Mediterranean preoccupation with heavily modulated patches and long decays, bears fruit. One set of oscillators produce the essential, solid sound, the other set tinkle away in the distance like echoes – ineffective on their own but seductive when teamed.

Siel are promising a change in the EPROM to allow a keyboard split, and there's a computer interface on the way which should (for under a ton) run a stack of Expanders from a cheap micro such as a Commodore 64. They shouldn't have any problem getting the electronics right since American experts Sequential Circuits acted as consultants in the original MIDI planning. They probably suggested the in, out and through sockets on the back panel.

But, to return to the intro and make some sense of it all, I can't help feeling that powerful though the Expander system may be, it's partly making up for weaknesses within the Opera itself. MIDI isn't everything. You do have to start with a good machine and for the price, the Opera is already outdistanced by many of its competitors.

synth £1299
expander £699