Standalone Poly Plaits


#1

I was thinking about what I want in a synth and came to the conclusion that I would like the Plaits sound engine in an elektron box. Like the DigiTone but capable of 8 plaits worth of mayhem. Naturally this is just a fantasy, but can I make this a reality?

What platform would be necessary if I wanted to prototype this concept? I’m not looking to make a modular system into a groovebox but make a groovebox with Mutable’s gorgeous sounds inside.


#2

As far as products go, check out the archive for information about the Ambika. It is a 6-voice hybrid poly (somewhat similar in architecture to something like the Novation Peak), based on the Shruthi. While discontinued, you see them pop up used every once in a while and people in the community have taken the open source schematics and produced PCBs for building yourself.

Plaits has not been open-sourced yet. If/when it is, the DSP will probably be ported to a bunch of platforms (for example, there are externals that run Braids in Puredata, VCV rack, Max for live patches and a bunch of other things), so if you have a capable enough computer, you could for example, create a virtual VCV rack with 8 Plaits in it (you could do this with Braids now–it is the “Audible Instruments Macro-oscillator” in VCV).

The interface of Plaits is pretty tuned to being used in a modular-environment (used in accordance with other modulation sources and audio modifying modules), and while I think it’d be cool to explore some of the modes in a Poly or in a multi-part, parameter-locked-in-sequence context (in fact, I have 2 Plaits in my rack and that is a ton of fun!), I think it would get very complicated quick (in terms of both interface and things you’d need to include to get the most out of the various modes, as well as a project).


#3

Ambika looks really interesting, regardless of my current obsessions, I will have to keep an eye out for that.

Precisely my inspiration. I have been using the digitakt and yarns to control the modular, but occasionally I wonder how much could be done with the mutable firmware if it were in a stand alone synth. I know that I like the sounds of Braids/Plaits as much as anything I I have found elsewhere and the tools found within the elektron requencers should be sufficient to modulating, parameter locks and LFOs cover all your bases.

But Elektron isn’t exactly encouraging people to load alternate firmware on their devices, so it’d be a ground-up project if I go forward with it.


#4

Plaits’ models go from 50% CPU to 90% CPU on a Cortex M4 clocked at 72 MHz. You can make things more efficient by increasing the buffer size to amortize the cost of all the stuff that is done once per buffer (exponential conversion of pitch data, preparation of lookup tables). And in a standalone device, you won’t have to do any CV processing. All in all, it should be possible to run 7 or 8 instances of the Plaits engine on a STM32H7.


#5

Said it before, say it again; I really hope @pichenettes will do a polyphonic digital desktop synth some day, maybe as a collaboration with a bigger brand.


#6

Korg or DSI or nobody else.


#7

bring it on (-:


#8

Someone please email Tatsuya and Dave.


#9

#10

I know. But he’s still involved in an advisory role.


#11

Ahh, this is exactly the type of information I was looking for. Many many thanks.

Also, DSI HQ is located a few minutes away from me. I have always been interested in visiting there as they give informal tours to synth enthusiasts. If I go, I will be sure to mention your somewhat satirical interest in a collaboration :wink:


#12

If you’re interested in Ambika, it’s alive and well in the open-source DIY community. Tube Ohm offers a Tubika DIY kit or just pcb’s with 4 filtercards available. It’s on my list, but I’m still working on an Anookum (Anushri) and a Deckard’s Dream.

I’m very thankful to MI for open-sourcing these older synths, as I only just discovered them…long after they were popular! Now I just need to get my hands on a Hyve Touch Synth PCB…!!


#13

I guess you could mount 4 Plaits, a Yarns and a 4:1 mixer in an 84 HP skiff (with room for some FX as well!), to get a 4-voice Plaits. Yarns has a ‘dispatcher’ mode which would assign the polyphony across the 4 Plaits.


#14

I was thinking about getting a discovery board for the STM32H7 just to play around with some DSP heavy algorithms, but they cost something like $600 each for some strange reason!


#15

The thought has definitely crossed my mind that I could sell off some of my unused gear and get a 4 x plaits system. I just wanted to parameter lock more settings than just pitch though. Modular is also possibly the most frustrating poly synth you ask for as well. Whenever I set up a poly patch it invariably starts to become a tuning nightmare and eventually devolves into a multitimbral patch anyways.


#16

I see! Maybe I have found an ambition that can blow a larger hole in my savings than modular!


#17

nice to see you’re so selective! And what about an open-source design by and for the community?


#18

I realized that the STM32H7 is pin compatible with some less powerful chips (e.g., STM32F7 STM32M4), so it might be possible to use an older design for experimentation. E.g., you could potentially swap out the STM chip in Braids/Rings/etc and work on some new algorithms to take advantage of the extra computing power…


#19

This would still be a souped up eurorack module, no?


#20

The only issue I see here is the smallest (pin-count) packages on any STM32H7 released thus far is QFP-100, and all the M4-based modules are QFP-64 based. So I guess it wouldn’t be a drop in replacement, but certainly possible to tweak a design without too much effort.

Though I remember @pichenettes mentioning somewhere that the Clouds-successor’s processor isn’t yet mass produced, so I’m curious to see if it uses: a) a yet-to-be-announced 64 pin variant (possible, but maybe not likely for eurorack volumes?) b) QFP-100 or c) something BGA. As much as DIYers and "let’s shrink this UI as much as possible"ers of the world may hate to hear it, c. almost seems most likely to me (potentially more pins used with differential inputs, maintains smaller physical package size, maybe more feasible now manufacturing larger batches)!

Another consideration is Olivier’s stmlib is based on the STM32 standard peripheral library, which doesn’t exist for M7 or H7, so some work would have to be done updating things to instead use STM32cube. Something to contend with, but certainly quite doable especially since Olivier has presumably already contended with this and hopefully he’ll continue to share this work with the world!