The Roland Promars helped to usher in a new era of microprocessor‑controlled analogue synthesizers, but has subsequently been all but forgotten. Gordon Reid remembers the world's only 'Compuphonic' monosynth.
The start of the '80s was a golden era for Roland. The company had finally broken through into the world of large analogue synths but, unlike their American competitors Sequential and Oberheim, their range also boasted a huge array of complementary monosynths, effects units, drum machines and sequencers. Indeed, you need only look back at brochures of the time to see how Roland was beginning to dominate the music‑technology scene. A couple of years before Yamaha launched the DX7, and while Korg was still in the doldrums with the Polysix and Mono/Poly, Roland's catalogue was resplendent with the Jupiter 8, the Jupiter 4, the System 100M, and the MC4 and CSQ sequencers. To these, you could add highly desirable monosynths such as the SH2 and SH09, the now legendary VP330 vocoder and the Juno 6 — a synth whose descendants would change the keyboard world forever. Oh yes, and let's not forget the RS09 and SA09 Saturn, the RS505 Paraphonic Strings, nor the now highly fashionable Space Echo delays. There was even a highly respected range of rackmount synths and effects that included the Dimension 'D' spatial chorus unit, the SVC350 Vocoder, the SBF325 Flanger, and the SPV355 Pitch/Voltage synthesizer.
Yet even among these, there were some turkeys. History sneers at the VK1 and VK09 organs, and the least said about the Piano Plus range, the better. But there was one instrument in the range that is now neither revered nor derided. Indeed, it's almost entirely overlooked. Designed as a partner to 1978's Jupiter 4 'Compuphonic Synthesizer' it remains one of the very few VCO‑based monosynths with patch memories. It's the long gone, and largely forgotten, MRS2 'Promars' Compuphonic Synthesizer.
So what does the impressive‑sounding term 'Compuphonic' actually mean? It's very simple. The controls of a conventional '70s analogue synthesizer (ie. one lacking memories) form part of the synthesizer circuits themselves. In contrast, those of a Compuphonic synth (ie. one that is microprocessor‑controlled) report their position to a computer. This computer then determines the control voltages that control the VCOs, VCFs and VCAs. Like the Prophet 5 and Oberheim OBX that preceded it, the Promars combines an all‑analogue signal path with microprocessor control. You might surmise, therefore, that there is at least one A‑D converter and one D‑A converter in such a synthesizer, and that all the controls will exhibit audible stepping of parameters such as the filter cutoff frequency. And you would be right on all counts.
The microprocessor in the Promars is a particularly primitive device called an Intel 8048. For those of you old enough to remember the first home computers and PCs, you'll be shocked to know that there was a processor even more limited than the Intel 8080 or Zilog Z80 — and this is it. Since the 8048 is an 8‑bit chip, and it has an 8‑bit data buss, it would be reasonable to hope that all the panel controls on the Promars offered 256 discrete steps. No such luck! Roland multiplexed the data from the knobs, sliders and switches so that the fully variable controls utilised six bits each, and the switches utilised the other two. This means that there are merely 64 possible positions for each knob or slider. Mind you, this is superior to the 32 possible values of 5‑bit systems such as Korg's Poly800, so maybe we shouldn't be too critical.
So why did Roland limit the amount of control in this way? That's simple to answer: it speeded up the data handling and reduced the amount of memory needed to store the patches. At a time when a 2Kx1 chip (ie. one that stored 2 kilobits, or 250 8‑bit words of memory) was both state of the art and expensive, this was an important consideration. Given that the programmable section of the Promars has just 16 fully variable controls and 14 types of 'switch' data, this means that the 8048 could store an entire patch in a mere 16 8‑bit words of memory. Since there are eight user memories, that means that the Promars needs only 128 words of RAM to store its complete set of patch memories. In contrast, the operating system of the G3 Macintosh on which I'm writing this currently occupies 41.5Mb of RAM, which is about 325 million times as much memory as the Promars!
The Promars has long been described as "a monophonic Jupiter 4" and in many ways this is a fair assessment. Indeed, much of the service manual and technical notes are common for the two instruments. But if there is one respect in which they differ greatly, it is that whereas the Jupiter 4 has one oscillator per voice, the Promars has two. So let's start here...
Like all the Promars' synthesizer circuits, the oscillators are true analogue devices. In other words, they are VCOs. Many people are confused as to how the oscillators in instruments like the Promars can be genuine VCOs when they are digitally controlled, but it is perfectly possible. The key point is that while the oscillator parameters are stored and recalled in digital form, the actual pitch is controlled by an analogue voltage. In a digitally controlled oscillator (DCO), by contrast, the pitch itself is controlled digitally.
So, the Promars has two VCOs. But where are the waveform and octave controls for the second? If you look at the 'Dual VCO' section of the control panel, you will find just one octave selector, one waveform selector, and one set of pulse width/PWM controls. Consequently, you could be forgiven for thinking that there is, in fact, just one oscillator with some kind of doubling trick concealed under the Promars' hood. The only nod towards the existence of the second oscillator is low on the far right of the angled panel behind the keys. If you look closely, you'll find two small knobs and a switch here. These purport to control the pitch of VCO2 (or turn it 'off'). So, what's going on?
The answer is weird. The primary oscillator (VCO1) is on the main synthesizer voice board; these boards are the same as were used for each of the Jupiter 4's four voices. The second oscillator, VCO2 (which the Jupiter 4 lacks) is an 'add‑on' that plugs into the main board, and which mixes its signal with that of VCO1 before it passes to the VCF. Having added the second oscillator, Roland then cut corners and, in duplicating the bulk of the Jupiter 4's control panel on the Promars, left no space for independent control of VCO2. Furthermore, the Promars' software has no provision for the extra five parameters that it would need to separate the two oscillators.
Consequently, you must at all times have the same waveform, the same basic pitch, and the same PW parameters (with manual or LFO‑driven PWM) for both VCO1 and VCO2. The only thing that you can control independently on VCO2 is its fine‑tuning. This is unfortunate because it severely curtails the Promars' usefulness. Mind you, there's no reason why you couldn't build a bit of additional hardware to control VCO2 independently. It wouldn't be programmable in the sense of being stored in the patch memories, but it might be an interesting project for somebody who's a bit of a whizz with a soldering iron. Has anybody in SOS‑land ever done this?
These shortcomings might have made the Promars seem more like a single‑oscillator synth than a dual‑oscillator one so, perhaps by way of apology, Roland gave VCO2 a novel pitch‑selection mechanism. Basically, the two knobs mentioned above allow you to set two pitches as offsets from VCO1's pitch. You can, for instance, set one up as a mild detune for rich 'chorused' sounds, and the other as a perfect third or fifth for 'synthy' sounds. The switch then selects between these two settings and 'off'. Since these tuning controls (which actually span somewhat more than one octave) do not form part of the programmable bit of the sound, they offer an interesting performance option. Mine are set up to allow me to switch from thin, precise sounds (with VCO2 turned off) to fat sounds (adding VCO2) to dual‑pitched timbres, all at the flick of that single switch. Furthermore, since both oscillators have square‑wave sub‑oscillators (of course, there is only one switch for this — both sub‑oscillators must be 'on' or 'off') I find that I can generate a wide range of textures with considerable depth when desired.
Following the oscillators, you'll find two filters. The first is a simple high‑pass affair with just a single slider that varies the cutoff frequency between 40Hz and 5kHz. The second is a fully resonant (ie. capable of self‑oscillation) low‑pass filter that offers a full 20Hz to 20kHz range, LFO modulation, ADSR envelope modulation (normal and inverted) and four degrees of keyboard following: 0, 40, 70 and 100 percent. The 100 percent option tracks the keyboard nearly perfectly so that, with Resonance set to maximum, you can 'play' the filter as if it were an extra oscillator. The filter is typically Roland: clean and precise, if somewhat lacking in the depth of character that distinguishes the Moogs and ARPs of the era.
The next set of programmable controls manipulate the VCA and its dedicated ADSR envelope generator. Unusually, the VCA Level control has an associated LED that allows you to see whether you are overdriving the VCA. I'm sure that Roland added this to stop players from pushing the output stage of the amplifier beyond its limits, but I'm equally certain that players today will use the LED specifically to make sure that they are doing so. The result is not at all like overdriving the input of a filter, but nonetheless it lends a certain 'thickness' to the sound.
Roland's documents state that the envelopes offer sub‑millisecond attack time, and they should, therefore, be very snappy. Indeed, the Attack seems very rapid to me, but the Decay most certainly is not. This makes the filter envelope very 'squelchy' in the best Roland tradition and, with maximum resonance, you can easily begin to emulate the sounds of a TB303. (Why you would want to is another matter...)
Modulation was always a strong point on Roland's monosynths, and the Promars is no exception to this. The standard LFO offers four waveforms, and the usual Rate control, as well as a Delay and 'LFO bend'. The first of these allows you to delay the onset of the LFO (for delayed vibrato and other such effects) while the second makes it possible to control the LFO rate using a second slider. This slider has normal and 'wide' ranges, allowing you to push the LFO up to about 80Hz, which is high enough for a reasonable range of amplitude‑modulation (AM) and frequency‑modulation (FM) effects. Unfortunately, the Delay and Bend parameters are not stored as part of the patch. That's a real shame, and I suspect that Roland didn't really understand how useful players might find this.
Other non‑programmable controls include portamento, a volume control, a 'brilliance' knob that is simply another cutoff control for the low‑pass filter, a transpose switch, and a 'Hold' button. If these facilities are fairly basic, the modulation lever (what on a lesser synth would be merely the pitch‑bend control) is not. Together with three associated switches, this allows you to modulate the VCO pitch, the VCF cutoff frequency, and the VCA level using either the lever itself or the LFO, with the lever acting as an 'Amount' control. With two knobs that limit the overall sensitivity of the lever and the applied LFO, you can mix and match these options in any of combination you wish. For example, if you want to create realistic brass sounds you can sweep the pitch up manually, at the same time raising the volume and adding increasing amounts of growl. It's a powerful performance system, and very intuitive once you get to know it.
Once you've programmed the programmable parts of a patch, you can store it in any of the eight user memories. I remember how much I lusted after these in 1979. After all, you could do anything with eight memories, couldn't you? Furthermore, unlike the only other monosynth of the time that sported memories (the Oberheim OB1) the Promars also offered 10 presets: Bass, Strings, Funky Clav, Piano, Voice, Trombone, Sax, Trumpet, plus Synth I and Synth II. The Promars looked like a live performer's dream synth.
None of the presets (except, of course Synth I and Synth II) sound remotely like their namesakes, but many are surprisingly good. For example, the Bass preset is instant nostalgia. With all the performance controls set to their default positions, it's the perfect early '80s electro‑pop bass. If you increase the brilliance and add the second oscillator with just a tad of detune, you get an excellent, and super‑rich, synth bass.
By contrast, the String preset is thin and uninspiring, and it's hard work to get it to sound remotely 'stringy'. Nevertheless, I've found that with the second oscillator added, it forms the basis of an excellent range of sampled pad sounds. Moving on, you'll want to skip over the Funky Clav and Piano presets, which are unequivocally horrible. But then you'll come to the Voice. Oh, the Voice... I love it. With just one oscillator, a soupçon of portamento, and the Brilliance closed down just a little, it's instant Tomita. If you increase the Brilliance it turns into a marvellous Clavioline patch, perfect for those 'Telstar' imitations you've always craved. Then you can add the second oscillator for another range of super, lush sounds. As far as I'm concerned, this preset alone could justify the price of a cheap monosynth.
Of the three brass sounds, the Sax is the turkey. This is not surprising: sax sounds are ghastly on every monosynth, even the wonderful ARP ProSoloist. However, if you stick to just one oscillator, the Trombone and Trumpet patches are surprisingly brassy, especially since the LFO is set up to produce a characteristic growl if you 'play' the modulation lever correctly. Adding VCO2 takes you into synth‑brass territory, and very nice that can be, too.
Finally, we come to the 'synth' sounds. Synth I isn't bad: it's a typical resonant filter sweep which, with detuned VCO2, reminds me of nothing so much as the introduction to Lenny Henry's wonderful Barry White pastiche, Theophilous P. Wildebeest. But, if Synth I is usable in a handful of contexts, Synth II almost certainly is not. It's the obligatory, heavily modulated silly noise and, like spandex flares, is best consigned to history and politely ignored.
The Promars was never a 'hot' item on anybody's shopping list, and it has been almost entirely overlooked by the "it's analogue therefore it must be wonderful" brigade. Indeed, it's hard to believe that the Promars ever looked modern or advanced. Nevertheless, it still boasts a number of excellent features. With its presets and memories, strange oscillator and sub‑oscillator facilities, typically 'Roland' envelopes and filters, plus its interesting performance capabilities, it deserves more attention than it currently gets. Considering that the far more limited Roland SH101 unaccountably commands prices above £200, the Promars — occasionally seen for about £150 — is undoubtedly a bargain. Well, it was, until SOS published this retro. Now watch the prices spiral upwards!
The Promars' relative, the Jupiter 4, was Roland's first true polyphonic synthesizer. However, although it was a contemporary of the SCI Prophet 5, the Oberheim OBX, the Yamaha CS80, and the Korg PS‑series, it was never going to compete directly against any of them. For one thing, it offered just one oscillator per voice, whereas each of the others offered two or, in the case of the PS3300, three. Its secondary facilities — chorus/ensemble, sample and hold, three Unison options, and an excellent arpeggiator — were first‑class, but they could not disguise the limitations of its basic voice architecture. Furthermore, with a 4‑octave keyboard that offered neither velocity‑ nor aftertouch‑sensitivity, and only 4‑note polyphony,it was no match for the wonderful poly‑aftertouch keyboard of the CS80, or the 49‑note polyphony of the Korgs.
On the other hand, the Jupiter 4 was affordable and undercut each of its competitors by 40 percent or more. Maybe as a result of this, the list of its owners became a Who's Who of the era, including Kitaro, Gary Numan, Tangerine Dream, Stevie Wonder, Tomita and, perhaps most famously, Nick Rhodes of Duran Duran.
In retrospect, its home‑organ styling and limited total of eight programmable memories might have guaranteed that the Jupiter 4 would never become a classic, but interest in it has never been higher than it is today. This is bizarre: as the forerunner of the mighty Jupiter 8, the 4 may hold some interest for collectors, but the Juno 60 is, for all practical purposes, a far better polysynth. It looks nicer, it plays better, it has more memories, and it's far more reliable. Maybe some players just want the Jupiter 4 for that sound — four VCOs arpeggiated in unison through a classic Roland ensemble effect. I guess the Jupe has something going for it after all!
It's fair to say that the Promars was never one of the darlings of the electro‑pop era, and only a tiny handful of 'name' bands and keyboard players used one. Perhaps the most famous of these were Depeche Mode (who used one briefly on their 'See You' tour in 1982), Klaus Schulze and Peter Vetesse. More recently (in 1996) Martyn Ware used one with Heaven 17.
One of the Promars' least appreciated capabilities at the time was its auto‑recharging backup battery. This means that, in theory, you never need to replace the battery that maintains the memory and, therefore, you never risk losing your patches while doing so. In 1982, Roland claimed that the backup battery would hold its memories for six months without power, and that a few hours powered‑up would be enough to recharge the battery fully. Well, they were wrong. The longest period my Promars has remained starved of power is... more than two years. Nevertheless, all my patches have remained secure. What's more, at 18 years old or thereabouts, the original battery is still going strong. Bravo!
The Promars is well endowed with CV and Gate inputs and outputs, and it even offers a modulation CV input for vibrato, filter sweeps and so on. Consequently, you may feel that it's superfluous to add MIDI to it. However, Kenton Electronics provides a MIDI interface that has a number of advantages over mere voltage control. Most important is that you can assign aftertouch, velocity, or any two MIDI controllers to affect modulation, pitch‑bend and filter cutoff. Then, once you have configured the synth and the interface, you can store the assignments (together with a pitch transposition if desired) in non‑volatile memory. A velocity‑ and pressure‑sensitive Promars also responding to a ribbon controller and joystick...? Now, that's something I would like to try!
The Promars shares its unusual oscillator structure with just one other synthesizer. This is another Roland, but it's one that lacks a keyboard, and offers limited envelopes and no modulation in its patch structure: it's the SPV355 rackmount synth.
Designed as a guitar or wind synth, the SPV offered pitch‑to‑voltage conversion (hence the name) and an envelope follower that, in principle, allowed you to play it from other sound sources. However, in common with other instruments of the era (such as the Korg MS20) it was never very successful in this role, so it's fortunate that the SPV also offered CV and Gate inputs and outputs. This makes it an ideal expander for the modern analogue studio, and for this reason you will probably find that the SPV355 now commands a higher price than the Promars itself. Isn't life strange?