Will Mowat and Mark Jenkins with two separate comparisons of two very different synthesizers. MJ get first dibs.

At last it's official. After months of speculation and rumours that the Juno 60 was on its way out, Roland confirmed this when they announced its replacement, the Juno 106, at this year's Frankfurt Trade Show.

One of the biggest selling points about the new Juno is that it can do more than the old one, but costs significantly less. With the Yamaha DX9 selling at £899 and the Juno 106 for a hundred pounds less, it's worth talking about the under £1,000 synth buyer. Especially since they're both MIDI and make a perfect combination in tandem.

Stork and butter, or chalk and cheese? Certainly there are plenty of points for comparison between the DX9 and the Juno 106, not the least important being the price.

In each case you've got a very compact, portable, five-octave polyphonic keyboard (although one is a synthesizer and the other may not be) and each machine is blessed with MIDI, which implies similar opportunities for expansion. But there, perhaps, the similarity ends.

As we all know, the FM-based DX keyboards represent something of a revolution in the musical instrument field, whereas the Juno is very familiar from the well-established 6 and 60 models. The styling of the 106 is slightly changed, tending more towards the up-market Jupiter range, but most of the functions are as before. The changes can be summed up very simply — instead of the 64 memories of the Juno 60 you have 128 in two banks; instead of an arpeggiator you have polyphonic glide (portamento); and instead of the DCB connector on the rear panel you have a MIDI bus. Also, and significantly, the recommended price has actually dropped.

Thick And Thin

It may be worth briefly looking at the basic functions of the Juno. There are six completely independent oscillator/filter/amplifier chains, together with suboscillators an octave below the main ones. A chorus unit helps to thicken sounds for string type patches and the LFO can be delayed for automatic vibrato or tremolo. There's only one ADSR envelope generator which has to be routed to both filter and amplifier, but this can be done in different amounts and with the option of 'gate only' (organ action) on the amplifier. There's a high-pass filter to thin out sounds and the usual frequency, resonance, envelope, keyboard and LFO controls on the main low pass filter, which has an appealingly smooth tone. Pulse Width Modulation can also be used to give some movement to the sounds.

As previously mentioned, the 106 can memorise 128 sounds which are assigned by two sets of eight pushbuttons on either side of the central LED display. The LED's show the bank number and patch number in operation, and you can step along a bank using a footswitch or sequencer.

It's possible to set up all the sounds for every song in a complete set in the right order now that there are 128 memories, but some would argue that there aren't 128 different things a Juno can do, which is equally fair.

The patch buttons double as MIDI assign channels for controlling external synths — the only complaint is that these buttons are very low indeed and pretty easy to brush against accidentally or in pairs. Presumably they're a cost-saving exercise and as such can't be criticised too harshly.

Some of the Juno's best functions are hidden on the back panel — the aforementioned patch step, a footswitch control for the Hold function, a simple switch to decide which MIDI parameters (keys, bend and modulation) are transmitted. Another function is slightly hidden away — like the SH101, the new Juno has its modulation introduced by a slight forward pressure on the bender, which gives superb control with only one or two fingers. All the parameters have been quantised for MIDI transmission, so you may be able to hear a slight stepping as the filter is adjusted for instance, but this isn't a big problem.

Overall, then, the 106 looks like a very respectable synth. The sleek grey plastic styling is easy on the eye — it's a lot lighter than the 60 as well — and the number of control functions (particularly MIDI) allow it to become part of a very versatile system. It's when you look at the price that the achievement of the 106 becomes amazing, and visions of two keyboards with detuned sounds linked by MIDI begin to float before the eyes.

Pretty Poly?

How can the Yamaha DX9 compete? The most obvious feature is that it's 16-note polyphonic, which really shows up in long decaying notes where protracted runs have a much thicker effect. The presets vary from the pedestrian to the amazing — percussive metallic sounds such as bells and pianos are best, whereas strings sound thin compared to the lush, chorused synth sweeps we've become used to.

There are many fewer presets than the 106, although of course more can be loaded from cassette, but the argument about the DX range really centres on the difficulty of making up your own sounds. It's not too difficult to put a sound into Edit mode and change the attack of one of the multistage envelopes, the volume of one of the Operators (sine wave oscillator banks) or the algorithm on which a sound's basic structure depends. The problem is that the results of such changes are so complex as to seem almost random, and to set off from scratch and invent a new sound is a task which seems daunting to the most expert (Rod Argent and the BBC Radiophonic Workshop included).

There are some features intended to help the user, such as Edit/Compare which flicks back to an original sound without losing an edit which you're in the middle of completing. The touch-sensitive buttons on the DX9 are pleasant to use and the LCD display gives plenty of information as to the status of the instrument and its various parameters — the performance controls are very classy too, with a sprung pitch wheel and free-ranging modulation wheel each having a nice thick feel to them. Keyboard Scaling allows a sort of multiple split effect over the length of the keyboard, a facility denied to the Juno, but again an original effect is pretty difficult to calculate.

If you do edit a DX9 sound you're likely to come up with a series of high-pitched oscillator whines which only settle down into musical proportions when certain magic relationships are struck using the Data Entry slider. When they are found, however, you're likely to have something of which the average analogue synth isn't remotely capable. Conversely, there's little point trying to get an analogue filter 'twang' out of the DX — in the time you take to obtain one you could have earned enough to buy a Prophet 5.

Interestingly enough, the strong points of the Juno and of the DX complement each other perfectly. The Juno for lush sounds and twangs, the DX for percussive sounds and hard-edged digital effects. Link them together with MIDI cords and the most unbelievable orchestral effects can be achieved — like tubular bells underpinned by strings, or pianos over brass chords. The unfortunate conclusion from these findings is that in an ideal world it should be possible to own both (for the money it would probably be worth it) and that a decision for one or the other is very difficult to fully justify. Not Stork and butter, then — perhaps oil and water, but in the case of the Juno 106 and Yamaha DX9, oil and water mix very nicely, thank you.

Any other review that compared a digitally-generated synthesizer and an analogue synthesizer would have to arrive at the following conclusion that I touched upon in an earlier article (December 1983), namely that it is not good enough to simply say 'digital is better' or 'analogue is preferable', without severe qualification, because the two systems of generating sound are not mutually exclusive: they can co-exist quite happily so far as the sound is concerned (the comparison of their relative facilities is a different matter — see below).

And yet we have two synthesizers, one from each camp, whose sounds are, in my obsequious view, eminently comparable. And to what is this 'U'-turn in opinion due? Well, it's like this. One one hand, Yamaha have brought out a digital synthesizer within financial reach of the average consumer (the argument being, if you can afford to buy this magazine, you can afford to buy the instruments it features...) and on the other hand, Roland have come up with a machine that sounds remarkably clean and stable for an analogue, a sound which, geologically-speaking, removes it from the average analogue tundra and perches it on a stony outcrop of the digital mountain. Both offer value-for-money in similar proportions, and they are both examples of what can be achieved with a lot of thought, a lot of money, and a desire on the part of the manufacturers to produce the 'next step' forward on the path to perfection, instead of a shuffle sideways.

Real Improvement

This Juno replaces the Junos 6 and 60, and the whole exercise demonstrates the now well-recognised Roland marketing ploy of facelifting and improving a proven seller. This can and does antagonise buyers who thought that the Juno 60 was the bee's knees (and it wasn't around that long, was it?) until this one came along; and because the 106 is so obviously an improvement facility-wise of the basic Juno theme, the values of the 6 and 60 are going to suffer badly on the used market. So often, updates can appear gratuitous to the point of being cynical. But in this case, Roland could be justified to a certain extent by their introduction of MIDI to the Juno — this is the biggest change that has been made to the Juno concept. Certainly, the other 'improvements' that have appeared would not have warranted a re-launch: I mean, it would not be unkind to say that already in the 60 there were perhaps too many memories (you were hard pushed to find 56 different sounds which were not already varieties on a basic theme): and here on the 106 you have 128 memories! Impressive on paper (it has as many memories as the four-times-as-costly Prophet T8!) it probably helps to sell the instrument, but do we need them?

The synthesizer of the Juno 106, if I can put it like that, is exactly the same as in the previous Junos — no update has been made to the sound or the parameters — so it will sound no better than its predecessors. And while we're on the subject, let's quell a nasty little myth that is being put about by some people, namely that the Juno has digital oscillators. The Juno has analogue oscillators just like the Prophets or its big brother the Jupiter 8, but whereas the latter's pitch is controlled by a voltage sent out from the key to the oscillators (which is a relatively inaccurate system compared to the more modern DCO circuitry, and depends very much on the quality of the various components as to how much it stays in tune, and in any case has to have a 'realignment' circuit incorporated to occasionally bring all the oscillators into tune with each other), the Juno's oscillators are frequency-locked in tune with each other by a digital process which involves the key providing a code for the Central Processing Unit; the master clock is divided down by a programmable counter, and this counter is controlled by the CPU: the oscillator cannot drift, but in all other respects it behaves like an analogue oscillator, ie chucking out random waveforms which are filtered down by circuitry to provide the Juno's pulse and ramp waves. Because the DCO's are frequency-locked, you will not find them on a two oscillator-per-note machine... yet. And because they are frequency-locked to each other, the machine sounds immediately more clear and pure than a VCO system, without the hint of drifting that makes VCO systems so 'warm' sounding when used in pairs like in the Jupiter 8.

Ergonomics

Here are a few pointers you can be bearing in mind when you look at the Juno 106 in the shop. The styling is quite pretty and 'modern' in muted colours; the layout of the synthesizer controls is fine, following as it does the usual left-to-right flow of analogue synthesis. There are two exceptions to this common-sense approach adopted by Roland. Firstly, the bank and program buttons are too close to one another: a hurried program change in un-ideal conditions (on a darkened stage, for instance) could well mean your finger finding the wrong button — it would have been so simple to slightly separate the little square bits of plastic from one another. The second niggle is more conceptual: I for one do not remotely like the trend, set by the Jupiter 8 and followed by Korg and Prophet, of having a display letting you know what program you have selected. The system looks impressive and high-tech, but it just does not make sense in practice: firstly, on the Juno, there is nothing to tell you that you have pressed the desired program or bank; you must look at the display which, being away from the buttons, necessarily means your eyes switching from your finger to the display. And to the gentlemen in the back row who shouted 'a trifle!'. I would say this: when you use these machines professionally, you want the most ergonomic layout you can get to reduce irritation and confusion. How much neater to have gone for the Oberheim system of a light-emitting diode set in every bank and program button — in one fell swoop it tells you that you have made contact with a button, and that the button is the correct one: no need to go checking elsewhere to make sure.

But I'll say hats off to Roland for carrying on the Juno tradition of banks of programs, with eight programs to a bank, because you can select these eight programs by hitting the desired program selector without having to reselect the bank — one movement less to make, one contented musician more to boot! On the more expensive Jupiter 8 and Prophet T8 you have to make two selections per sound. You can of course override the memories and go into real-time programming by hitting 'manual'; what is nice about it is that, having cooked up a new patch in manual, you can return to a memory to find a suitable home for it and then return to the manual mode, without having lost the patch: almost like having a 129th memory!

The low-pass filter cut-off point is ultra-stable, which is a strong feature of the Junos. You can get this fine filter to a self-oscillate, and by whacking the keyboard tracking slider up to its maximum one-volt-per-octave, key presto (Ed. groan!) you have a sine waveform in perfect tune with the keyboard. And by reducing the keyboard tracking, you can experience the small, intricate world of micrometal composition as practised by small, intricate and ultimately very boring composers. Can't get macrotonal, though. Wouldn't want to, anyway.

Other little things that could be improved on the synthesizer itself (note that the synthesizer is basically a very good piece of engineering, and it is simply a case of fine-tuning its excellence) include a slight increase in noise when the bi-level chorus is selected (the chorus is necessary more often than not to 'thicken' the sound of this inevitably rather 'thin' sounding one-oscillator-per-voice machine), the usual slow-attack scenario found on many analogue synthesizers where designers think that slow attack on the envelope means a great long silence on playing the key, followed by a relatively quick arrival of sound, and lastly two contentious issues: whether it is about time this sort of basic synthesizer should include as a matter of course two full envelope generators (one each for filter and amplifier) and two low frequency sources so that the performance control to the left of the keyboard can inject vibrato at a frequency that is not set by the synthesizer's LFO: at present you cannot have fast vibrato on the performance lever if the LFO is already producing, say, a slow pulse width modulation.

Full Of Meat

The polyphonic portamento is not programmable but I find it a lot more relevant than the previous Junos arpeggiators. There is also a key transpose control which gives you an extra octave up or down on the keyboard on top of the usual 16, 8 and 4 feet and also allows you to transpose to any key you like. If you're finding that playing in the key of B is too confusing — just press the button and C sharp and you can play your B scales in C — you know, the one with all the white notes! Before we get onto MIDI, you may like to hear about the two Roland 'Poly mode' routines. The Juno allows you to select either 'Poly1' or 'Poly2' (but not Unison: mere words cannot describe the disappointment one feels in not being able to stack those meaty bass lines. This probably has something to do with the DCO's which are frequency locked and which would therefore sound like one oscillator even when stacked.) Each time you hit the same key with one finger the same voice is triggered — the CPU does not trigger each of the six voices in turn like the Oberheim system. The difference between the two polyphonic modes then appears when you play subsequent notes: in 'Poly1', each successive different key is accorded a fresh voice, which allows each voice envelope to go through its full cycle; this is especially apparent with a long release, unless of course you play a riff of more than six notes in which case you start 'robbing' the first voices again (in a block chord, the seventh key to be played will not sound). In 'Poly2', each successive different key will be accorded the same original synthesizer voice, with fresh voices being triggered only when more than one key at a time is played. The net effect is this: you use 'Poly1' for one-key-at-a-time solo work, since this mode will allow in effect only the last note played to have a release.

And now for the Musical Instrument Digital Interface. As the whole world knows by now, it allows synthesizers to speak Japanese and English with equal ease, and I'm thinking of having one fitted to my head. With sixteen MIDI channels, you must check that the instrument that you are interfacing with displays the same channel as the Juno before they will communicate. Roland have included a handy selector switch to allow three levels of information to flow through the MIDI bus, ranging from information about the keys being played, and the optional 'hold' foot-switch being activated, through to the information about the parameters in a patch. Of course, in these relatively early days of MIDI, there still exists the situation where the only time you will get a full 100% flow of information is between two identical instruments. It is a fact that MIDI has become very popular with those who have access to the instruments blessed with it, and one of the most popular link-ups is between the Prophet T8 and the Yamaha DX7, giving analogue and digital sounds at the same time. But don't forget: MIDI will not improve your synthesizer's ability to make sounds when used on its own, and for this reason, and the fact the Juno 106's synthesizer is the same as the Juno 60's you shouldn't immediately trade in your 60 for a 106, or get upset about the appearance of the 106 so soon after buying a 60. The new machine cannot sound any better than the old one!

Best Of FM

The Yamaha DX range of synthesizers is everything that people say it is... and more. It is infuriating, breath-taking, complicated, dazzling. The only thing that would make anyone want to buy a DX is the sound of the animal, not the controls, for hand in hand with digital generation comes the assignment principle of controls, where you have to be very clear-minded about what you are doing before you take the plunge and actually lay a finger on the control panel.

The DX9 is the least complex of the DX range; it basically has less of everything compared with the slightly more expensive DX7. The irony is that whereas the Roland Juno 106 has 128 memories for a synthesiser that is by comparison very simple and whose ability to create a large variety of sounds is, again by comparison, strictly limited, the DX9 has just 20 memories on board — not nearly enough for the imagination of the average synthesist. I belong to the generation that learned synthesizer programming on analogue synthesizers, and as a consequence the arrival of Frequency Modulation synthesis at the beginning created a few headaches. How did one know, on the DX, what each algorithm meant in terms of sound? How could one predict what the relative tuning and output level of each operator would do to the sound? Well, in the fullness of time, these questions have been answered to the extent that, whereas many sounds are still being discovered by accident, one can begin to aim for a particular sound now and be reasonably sure of reaching it. In the field of inventing new sounds, the DX9 is superb; it's in the field of mimicry that a good working knowledge of the fundaments of sound will help, for Yamaha give no assistance in their manual for those who want to know which of the eight algorithms one should start with for a violin, for instance. The ethics and so on of mimicry versus invention in synthesis will rage on for a long time yet, but there is no getting away from the fact that the DX9 is very good at the mimicry of certain instruments, given careful programming, and as with all synthesis, careful equalisation on the recording desk.

"Alive, Alive 0"

Live, I fear the DX9 displays its Achilles Heel, with the assignment principle cutting out any rapid parameter changing on stage. When the PPG Wave first came out, everyone said how dreadfully complicated it was to program or edit in a live, adrenalin-loaded situation; it was however thought that familiarity would improve its user friendliness on stage. But this has not really happened, and the Wave continues to be used as a preset instrument on stage. This is what is happening to the DX range, with the DX9's further problem of having a small memory. But when you sit down on your own, without the pressure of time on you, and concentrate on the DX9, you don't notice the hours slipping by. Although it is not really fair on the DX9, you find yourself, when making a comparison between it and an analogue synthesizer, having to relate to it in analogue terms. You can play up to sixteen notes at one time, or put in more human terms, release times will be allowed to run their course more readily than on a six-voice synthesizer. It has a five octave keyboard like the Juno, and modulation and pitch-bending wheels. The fun starts when you begin programming and it is then that you have to cease making comparison with analogue, it is enough to know that the DX9 gives you the possibility of a vast range of permutations of sound. You will find envelope generators; not one as on the Juno, but one for each FM operator, and depending on whether the operator is acting as a modulator or a carrier, the envelope will determine the behaviour of the loudness or the harmonic content of that operator. In fact, even with this relatively simple DX9 (simple compared to the long-awaited DX1) the number of permutations of various parameters is awesome, even terrifying.

So many places to go wrong, but equally so many chances to make a happy mistake! None of the performance controls of the DX9 are programmable, and in this sense it is similar to the Juno. By entering what Yamaha call Function mode and what everyone else calls the 'Brown mode' because of the colour of the selector (well, me, anyway!) you access the two wheels, the cassette interface, the effect of the Yamaha Breath Control has on various parts of the sound, portamento and so on. The Purple mode is the one used for assembling a patch and comparing your edit with the original sound (the Purple mode is the complicated one!), the Green mode selects your twenty memories, and the Pinky/Red mode (highly technical stuff, this!) allows the storage of a new patch into one of the twenty memory slots, as well as the time-saving facility of copying the values of one operator's envelope characteristics onto another.

And the sound? This is going to be the main reason for opting for a DX rather than analogue. If you can get past the irksome (to some) assignment foothills of the panel, you arrive at a typical digital landscape: outlandish sounds, some of them; extremely interesting on the percussive front, since the attack times are much quicker than on an analogue; the ungracious would say that on the whole, the sound is rather thin — complex, yes, but if you want a 'background wash' of sound, it would be fair to describe sustained sounds which attempt to resemble strings as 'sterile'. And this is what leads to a most interesting conclusion, for which you will have to wait, because the MIDI interface needs a mention! As with the Juno, you can use the MIDI to the full only when interfacing with the rest of the DX range — that is when you can transmit and receive the details of a patch. Otherwise, there are two basic ways of using it: real time control of another MIDI-blessed synthesizer (like the Juno) or interfacing with a MIDI sequencer or a remote keyboard.

There would be various ways of concluding this feature. One way would be to say that if you want ease-of-use go for the Roland, but if sound is more important, it's the Yamaha. Another way would be to ask is mimicry important to you, in which case go for the Yamaha, or if all you want is background wash then the Roland is the one. But all these are half-truths. Yes, the DX9 is better at percussive, surprising sounds. Yes, the Juno is, at first, easier to use. But the surprise is this; if the Juno did not have the chorus function, its sound on sustained notes would be as flat, featureless and barren as the DX9's. Roland's use of the DCO and the one-oscillator-per-voice system has made the analogue synthesizer sound, under certain conditions, identical to the digital. If you wanted to be cynical about it, you could put the DX9 through a Roland Boss chorus pedal and have the best of both worlds! I will not push my conclusion any further than to encourage you to try out the competition; the DX7 costs a little more than the DX9 but has lots more control and sound possibilities. The Roland JX3-P is a weird kettle of fish, but is priced near the Juno and does lots more, partly because of its two oscillators per voice. In fact, when you look at it, the competition is not Roland v. Yamaha, but JX3-P v. Juno 106 and DX9 v. DX7. It's civil war!

Roland 106 £800
Yamaha DX9 £899