John Bowen Synth Design

Hello all,

John can you please tell us what will the polyphony be in the Solaris? Is this dependent on patch complexity? Is there some way of seeing how much DSP power each patch consumes?

Also what is the multi-timbrality? Do you have some sort of dynamic voice allocation between the patches? Is it possible to set individual patches within a "combi" to last note priority or high note priority? Is it possible to have legato and portamento settings individually?

Also how many analog outputs are there? Does the digital output carry all the signals and in what rate?

Regards,
Yannis

Yannis,
At the moment, we are getting 3 voices per DSP, with the 6th DSP reserved for effects processing. So, this gives 5 x 3 = 15 voices currently. It may be possible to get 1 more voice out of the 6th DSP, but we haven't decided if that's the best way to go yet. The DSP is not dynamically allocated, as it is with the plug-in, so all patches will yield the same polyphony. Hopefully at a future date we can change this.

The other way to increase polyphony would be to drop the system clock rate to 48 kHz. There's been several debates over time at the various trade shows between myself and some other companies' engineers, who claim that most end users cannot hear the difference with a lower rate (48 or 44.1 kHz), and that I'm being 'wasteful' by insisting on a 96 kHz rate. However, I want to make sure the Solaris can produce very high quality output, and so I'm sticking with the higher rate. It would be interesting to have a 'rate switch' in the Solaris, so one could try both (and I thought that we were going to be able to do that). However, I discovered that it's not that easy, as all of the code is being optimised to run at 96 kHz, and to run at 48 kHz would require some tweaks and changes to get things right.

Maybe the answer is to run at 48 kHz, with just the crucial modules oversampled at 96, but that will not be the case for this product.

re: Multi-timbre function - I think I did post elsewhere here that the v1 Solaris OS will not support multi-timbrality. There are 4 pairs of analog outputs, but at the moment only outputs 1&2 are connected. The other outputs will come into play with multi-timbre functionality.

re: digital output - S/PDIF is 2 channels out, and 2 in. The rate can be adjusted as you wish - 44.1, 48, 96.

-john b.

Forget multitimbrality, think Quadraphonic! Ok maybe I'm kidding...

I'd suggest other uses for the 2nd outputs but I've no idea how you would shoehorn them into your current architecture (fx 100% wet outputs--or conversely the typical 'dry' output found on other digital synths, perhaps an m/s encoded version of the signal, or just have them be 4 different destinations available in your mixer, etc).

a 48/96 internal switch would be okay i think, but I personally gonna favor the 96 HiQ output even if it costs some voices (well half of it i think)

15 voices will be more than sufficient for mono timbral work and even with 2 parts is is very good doable, more than 8voice poly a part i never use myself. You could set up like this:

1 x 8 voice part
1x 6 voice part
2 x 1 mono part ( if the coder can get that extra voice from DSP 6)

that will be more than sufficient for me...I really like the non-compromise aproach of the Solaris. Like that 96 khz feature.

But a switch can be a welcome addition i think..especially if one wants to make a whole track with the Solaris only.

I agree with Septic Studio. If sound quality is to be compromised it's better to leave it at 96 KHz despite the extra voices.

Given the fact that there are 4 analog outputs, it makes sense to have a 4 part multi-timbrality, no? Each part gets its own output and everyone is happy.

John, the JP-8000 has a sorta user defined voice allocation. For instance, it has 8 voices, 2 parts multi-timbrality so if you have set up something like a bass and a pad sound you can explicitly allocate 1 voice to the bass and 7 to the pad.

I don't know if you have something like that in your mind for future OS updates, but I think it works ok. I would be very happy to see the voice allocation saved per patch!

Also, if the onboard FX is disabled will the DSP (and hence the extra 3 voices) be available?

Regards,
Yannis

SepticStudio wrote:a 48/96 internal switch would be okay i think, but I personally gonna favor the 96 HiQ output even if it costs some voices.

Yes, but as I tried to explain, this would actually require different code inside, one set for 48 kHz, and another for 96 kHz. He would have to re-write much of the code, which is basically doubling the amount of work needed.
-john b.

ThreeFingersOfLove wrote:I agree with Septic Studio. If sound quality is to be compromised it's better to leave it at 96 KHz despite the extra voices. Given the fact that there are 4 analog outputs, it makes sense to have a 4 part multi-timbrality, no? Each part gets its own output and everyone is happy. John, the JP-8000 has a sorta user defined voice allocation. For instance, it has 8 voices, 2 parts multi-timbrality so if you have set up something like a bass and a pad sound you can explicitly allocate 1 voice to the bass and 7 to the pad. I don't know if you have something like that in your mind for future OS updates, but I think it works ok. I would be very happy to see the voice allocation saved per patch! Also, if the onboard FX is disabled will the DSP (and hence the extra 3 voices) be available?

We did discuss some fixed allocation of voices for parts, as you suggest. And yes, 4 part Multitimbre has been the plan all along....
As to disabling the onboard FX to free up the DSP - I have no idea if this could work. I'll have to ask.
-john b.

+1 on sound quality. There are a lot of awful sounding vst's and other stuff. Good sound is critical.

+1 on predictable DSP allocation.

I think the decision logic there is really good. Just wanted to offer a positive agreement on that.

Totally agree with this design.
Anyone here ever heard that Viscount VA version of the Oberheim?
Being an old Oberheim user I was interested but noticed the 12 voices of polyphony.
Immediately that translated into 12 weak voices.
When I played the first chord I confirmed my suspisions.

I have never played a VA synth I liked yet except Solaris.
I am happy that JB chose to use 6 X ADP21369 chips and 16 voices.
If I wanted a synth with 500 presets and 96 voices I would have made another choice, and then probably had to layer patches to get a giant sound, which in my experienc3e sounds like Gooo and lacking any definition.

This synth is a surgeons tool, that's why I am buying one.

So the final spec is 6 AD21369 DSP (so half the power of the XITE-1) ?

So to say it is equivalent to a full 50 old Scope DSP system.

CheerZ

John,

can you please tell us if any of the filters in the Solaris self-oscillates? If so how does it handle the resonance buildup close to 12 KHz? Do the filters use oversampling internally or is there something like a soft compressor to keep the buildup well-behaved?

Also can you please tell us how does the mixer behave when all the signals from the oscillators are at full level? Does it scale back the signal so there is no overload or does the user must decrease the respective levels manually so as not to have any overload?

Thanks,
Yannis

96khz internal processing was the right decision IMO. With the solaris, sound quality should be #1. And there IS a difference in SQ. I testet with difference between 44.1 and 96khz and synth plug-in sound with my pulsar2 cards, and it was quite noticable.

ThreeFingersOfLove wrote:John,

can you please tell us if any of the filters in the Solaris self-oscillates? If so how does it handle the resonance buildup close to 12 KHz? Do the filters use oversampling internally or is there something like a soft compressor to keep the buildup well-behaved?

Also can you please tell us how does the mixer behave when all the signals from the oscillators are at full level? Does it scale back the signal so there is no overload or does the user must decrease the respective levels manually so as not to have any overload?

Thanks,
Yannis

Hi Yannis,
All filters in the Solaris do self-oscillate. As far as any internal control for build-up - well I don't know, since I didn't code them, but I do not find any problem of a "resonance buildup' around 12 KHz (or anywhere else). For the mixers, as you say - the user has to decrease the levels to avoid overload, especially if you are using all sine waves for each input, but this is also dependent on where the Master Volume is set.
On the other hand - if you want the overload, it is there...
-john b.

I have a question about the 24 bit 96 kHz internal processing of the Solaris. I am completely new to digital synthesis & am still learning. I've looked for similar specs on other digital synths like the Access Virus & Roland machines, but cannot find any similar specs reported. Can anyone tell me how the 24/96 processing rate of the Solaris compares to other digital synths, e.g. Virus (just as one example)? Digital specs do not seem to be quoted in a common language.

If this has been covered elsewhere, I apologize for missing it in my search. If you'd be good enough to just point me in the right direction, I would be grateful...

3rdConstruction wrote:I have a question about the 24 bit 96 kHz internal processing of the Solaris. I am completely new to digital synthesis & am still learning. I've looked for similar specs on other digital synths like the Access Virus & Roland machines, but cannot find any similar specs reported. Can anyone tell me how the 24/96 processing rate of the Solaris compares to other digital synths, e.g. Virus (just as one example)? Digital specs do not seem to be quoted in a common language.

If this has been covered elsewhere, I apologize for missing it in my search. If you'd be good enough to just point me in the right direction, I would be grateful...

I would say the Solaris is in an entirely different league then any Access or Roland synths or romplers. The processing power blows them all away. No comparison. As for sound quality for a digital synth, it will be unequaled. You can't find similar specs because they don't exist in hardware. You would have to find it in a Vsti.

viewtopic.php?t=459

viewtopic.php?t=413

viewtopic.php?t=417

good luck finding any digital synth or va with specs this powerful:

Solaris
* 4 Oscs, 2 Rotors, 4 Mixers, 4 Filters, 4 Amp/Pan sections, 8 Envelopes, 5 LFOs, 4 External Inputs, Flanger/Chorus, Phaser, Delay, 3 band EQ, Overdrive
* 2 Vector Mixers
* 2 separate AM (amplitude modulation) sections, with Ring, AM, Rectify, & Clip algorithms
* Joystick and multi-touch Ribbon controllers
* Arpeggiator and control Step Sequencer provided, with MIDI output
* Performance buttons include 2 assignable switches, Arpeggiator Start, Sequencer Start, Hold, (Tap) Tempo, Unison
* Polyphony count - expected to be 24 voices with all 4 oscs, 4 filters, 4 mixers, envelopes, LFOs, etc. running
* 96 kHz internal processing rate
* Insert FX pre-/post each filter section, with BitCrush, Decimate, and Distortion
* 4 pair of outputs; Main pair for v1.0, additional outputs reserved for future use (when Multi-Timbre Mode is implemented)
* separate Headphone out

Detail per section:
Oscillators - each osc type selects from standard waveshapes (MultiMode Osc), wavetable (PPG) type, sample (WAV) playback, CEM (Prophet 5) type, or Prophet VS type. The MM (MultiMode) type provides the following waveshapes:
Sine, triangle, ramp, saw, pulse, noise, S&H, morphing sine-to-saw, morphing sine-to-square, and a stacked "supersaw" with varible detune (based on the Shape parameter).

Hard Sync is only available for MM saw, ramp, pulse, and the CEM osc models.
New types will be added as they are developed via an upgrade to the OS.

Individual "analog-style" glide is available for each oscillator.

There are 4 mod paths. Each one is freely assignable to select exponential frequency (normal pitch mod), Linear FM, or Shape as their destination. Mod Sources include any oscillator, any filter, the 4 external inputs, any of the lfos, envelopes, controllers, etc.. A 'sidechain modulation' function is provided for each path, using Controller (non-audio rate) signals. Controller signals are all lfos and envelopes, velocity, note, aftertouch, mod wheel, ribbon, joystick, select MIDI controllers, assignable CC knobs, etc.

Filters - 4 filters, each with selectable inputs.

Filter types include:
1) all pole possibilities for the MultiMode (MM1) filter, including 24 dB Lowpass, Highpass, and Bandpass, 12 dB Lowpass, Highpass, and Bandpass, and 6 dB Lowpass, Highpass, and Bandpass, along with some other combination modes, for a total of 23 variations.
2) 24 dB Lowpass modeled on the Prophet 5 Rev1 filter (SSM2040)
3) 24 dB Lowpass modeled on the Rev 3 Prophet 5 (CEM3320)
4) 12 dB Lowpass modeled on the Oberheim SVF
5) Comb/Tube filter (the "tube" is a comb with negative feedback)

New filter types will be added as they are developed via an upgrade to the OS.

For filter modulation, it's the same structure as the Oscillators - there are 4 mod paths. Each one is freely assignable to select Cutoff, Resonance, or Damping (if Comb/Tube is selected) as their destination. Mod Sources include any oscillator, any filter, the 4 external inputs, any of the lfos, envelopes, controllers, etc.. A 'sidechain modulation' function is provided for each path, using Controller (non-audio rate) signals. Controller signals are all lfos and envelopes, velocity, note, aftertouch, mod wheel, ribbon, joystick, select MIDI controllers, assignable CC knobs, etc.

ADSRs - there are 6 standard DADSRs. Each overall amount can be modulated by Velocity, and each segment can be individually modulated from Velocity, Note, Mod Wheel, and assignable Midi Controllers (CC1-CC5). Also, each segment can have a variable slope, from linear to exponential.

Looping Envelopes - there are also 2 looping envelopes, each with 8 Time&Level segments. There is overall modulation possible of Time and Level.

LFOs - there are 5 identical LFOs, with the fifth being permanently connected to the frequency of all oscs (therefore, it is called the Vibrato LFO). The LFOs have the standard waveshape types, and range from 0-524 Hz. There are parameters for Delay Start, Fade In, Fade Out, Rate, Waveshape, Retrigger, Phase, Level, MIDI Clocking, and Offset (offset provides a positive unipolar signal for the lfo outputs). There are 3 mod paths, similar to the Oscillator modulation structure. The destinations here are selectable for Rate or Level.

VCAs - There are several models implemented for the final output stage circuit. VCA Types include: Linear, Log, and Sigma (Minimoog style). There is 1 mod path for the VCA, and 1 for the Pan position.

Vector Synthesis - There are 2 Vector Mixer sections. The Joystick (non-spring loaded) in the leftmost section is normally connected to both Vector Mixers, but can be disabled.

AM Sections - 2 Amplitude Modulation sections, each of which have Carrier, Modulator, Algorithm and Shaper parameters. Ring Mod is one of the algorithms provided.

Effects - Initially available will be delay, flanger/chorus, reverb, EQ, Overdrive. As with the other sections, additional FX types such as a vocoder or resonant filter bank will be added as they are developed via an upgrade to the OS.