As of 2022, Roland has released an actual Jupiter-4 VST, some of the dialog in this article is thus outdated as it was written before the Roland Cloud Jupiter-4 and Mercury-4.
Jupiter-4 emulation has long been neglected in lieu of the JUNO-60 and Jupiter-8 emulators that are available out there; and it's one reason why many people still seek the "real deal" to get the unique JP-4 sound. So I was excited to see that Roland had officially released a promars plugin. BUT. It's rather surprising they went with the Promars however, instead of the Jupiter-4: a telltale sign that this project wasn't executed like how it should have been. Especially considering the Jupiter-4 is more powerful and can do everything the Promars can do. There's this long held belief (probably in part help propigated by vintagesynth) that the Promars has two oscillators per voice whereas the Jupiter-4 only has one:
That's a painful slice of misinformation since the Jupiter-4 can be put into monophonic mode (multiple oscillators per voice), and its oscillators CAN be independently tuned /de-tuned as well!* You can get FOUR oscillators per a single voice, or two oscillators per two voices: which would equate to having two promars in one.
*The tuning for the oscillators is located at the rear instead of at the front / which of course makes it less functional in that regard, although the JP-4 could be modded to bring the controls to the front like the Promars. And by this time, most Jupiter-4s have been modded to give them more capabilities (such as MIDI and more bank memory).
Despite that, the Promars isn't technically a true two-oscillator monophonic synthesizer like something as a SH-02, because the oscillators are still paired together as they are in the Jupiter-4.
The first test I wanted to see is if the famous square wave was correctly emulated:
And it appears Roland was paying attention since they got the unique shape down accurately (80%): with the exception that the waveform is lacking some of movement of the original JP-4. Just a quick back story, the reason why the square wave is like this is because it's actually being 'created' by multiple saw tooth waveforms. Later on most analog synthesizers would use clinically 'perfect' waveform generation (like the JUNO-60 for instance)-- however if every synth enforces perfect waveforms, you'll start to lose unique timbres found only in 'imperfect' waveforms. As a result of the Jupiter-4's "primitive" square wave, it results in something sonically more interesting than if it actually used a true 'square' wave, ESPECIALLY when set to pulse-width modulation.
Next I wanted to ensure the PWM affected the waveform in the correct way, and unfortunately I was a bit disappointed here as the 'movement' (pulse width) only moves one "saw", whereas on the original hardware both sections of the saw are in motion. And as expected it doesn't sound the same, either:
Here's a recording comparing both--FIRST a real Jupiter-4, THEN the Promars emulator:
Two things that stand out to me immediately, the LFO isn't manipulating the PWM in the same way, and the filter isn't emulated correctly resulting in unwanted unfiltered sounds. This is traditionally why one would consider a real synthesizer more 'warm' and a digital recreation 'cold' because of the difficulties emulating filters.
There are a few complaints I have after using the plugin for a bit. Controlling the virtual knobs on the interface is somewhat eccentric, instead of moving the cursor up (to turn the knob left) or down (to turn the knob right), you actually have to move the cursor in a semi-circle. While this is technically more realistic in moving a virtual knob, the way Roland implemented it is rather imprecise and awkard. Worse, the virtual sliders don't move smoothly and it's hard to bring a slider back to the centre. I'm certain most musicians that use software plugins are used to the up/down control of virtual knobs, so I'm probably not the only one who will find this as an annoyance: at the very least Roland could have added a feature to change between both control methods.
What I don't understand is why Roland deliberately limited the VST plugin to make it more realistic to the original hardware (when obviously a software plugin isn't going to be 100% accurate), but at the same time this makes the plugin worse and doesn't take advantage of what 'could have been'. For instance, why not include what the Promars/JP-4 would have sounded with a sine wave or triangle wave?
Getting more deep into it there become some very blatant disappointments: for starters (and I am certain many were hoping for this) there's no random arpeggiator, meaning it's not a complete Jupiter-4 replacement. There's no option for a polyphonic mode, which I can't understand as it would be very easy for Roland to implement. The filter emulation is as you'd expect, sounds like any other soft synth with the resonance not even getting close to the original IR3109. Reverb and Chorus are plain bread and butter like any other VST on the market.
Is the Promars VST plugin worth it? Priced at $145.00 USD, I think that's an insane cost (even more than the SOUND Canvas VA which has far more content to work with). Roland would need to at least add features that the Jupiter-4 had and a few extras: polyphony, random arpeggiator, better PWM emulation, better presets etc. How does this fare with some of Roland's competitors, like Korg's Legacy Collection? Korg's VSTs have more tuning options, EXCELLENT presets, better price point etc. In the case of the Mono/Poly VST, it has few presets but the quality of each preset is much higher.
What's actually funny is that Roland has the Promars VST tuning limited between: 430 Hz to 450 Hz, whereas Korg (on all of their Legacy plugins) has them being able to reach between 420 Hz and 460 Hz. It's almost like Roland deliberately degraded their tuning options so that Korg would have a 10 Hz greater range. There has been one unique situation where the 420 Hz tuning of the Korg plugins was useful; whereas you can't do that with the Roland Promars VST. Not at all something (most) would consider, but worth mentioning anyways. Why limit the options that SOFTWARE has when so much potential can be put into it?
In conclusion, I think Roland should have approached this as a Jupiter-4 emulator and added more features that weren't possible with the original hardware, so you get the "Promars" functionality with additional benefits. Otherwise, what's the point of going through (some) of the effort to make a partial software emulation of a Jupiter-4? If this was a competent Jupiter-4 emulator I think the $145.00 could be justified; if you have a limited budget, stay away from this plugin and buy something more useful (like the Korg Legacy Collection). The most realistic output I've gotten from the Promars VST was the FM synthesis (when having the LFO affect the oscillator). I will be uploading some audio comparisons of a Jupiter-4 output and the Promars VST at some point.
If I had to rate it, I'd give this plugin a 4/10, but that's primarily due to the fact it has so much potential that Roland flushed down the toilet. A Jupiter-4 VST with all of the voice modes and voice cycling would be one of the best VSTs ever. This one is *so close* to being that, but again Roland didn't add the basic software features that 'could have been'.
Also for reference, here's how Roland's legacy software synth collection stacks up with Korg's:
Sound Canvas VA - $125.00
PROMARS PLUG-OUT Synthesizer - $145.00
SYSTEM-100 PLUG-OUT Software synthesizer - $195.00
SH-101 PLUG-OUT Software Synthesizer - $145.00
Total: $610 / !No bundle option offered by Roland!
MS-20 - $49.99
Polysix - $49.99
Mono/Poly - $49.99
M1 - $49.99
WAVESTATION - $49.99
MDE-X - $19.99
Total: $269.94 ($199.99 if buying all in a bundle)*
*I bought the Korg Legacy Collection when it was on sale for $99.99, even better.