SMD Stereo Ambika-Braids XT

I’m about 80% into an Ambika build and I’ve been having some ideas about how it would be fun to re-design it a bit using surface-mount components, and along the way changing out all of the things that had to be sacrificed to make it DIY friendly.

My thoughts were:

  • Power would be 9-12VDC (much easier to obtain), negative rail would be generated with a switching reg.

  • Voices would be somewhere between Ambika and Braids (I think some of the braids modes wouldn’t make much sense in a poly, maybe I am wrong though). They would run on 2x STM32F4 (3 voices each). From what Olivier has mentioned on a few occasions this should be feasible. I2S and quad DAC for the oscillator outputs, 16bit PWM for the voltage control signals.

  • There would still be 6 individual outputs, but the main output mix would be stereo. Each voice would have a pan control. Pan would be a mod destination, but there would probably also be a dedicated “spread” control of some kind.

  • Filter could either be fixed (gasp!), or it could use plug-in filter boards. Similar to Ambika in that sense, but the “voice cards” would actually just be “filter cards”.

  • UI would essentially be the same, but probably running on some lower end STM32 (more room for growth/hax). In addition to the main LCD+knobs UI of the Ambika, the main board would have IDC headers on the left and right that could jump over to the XT-style interface boards. Obviously the case would have to be different depending on non-XT vs. XT but the main board would remain the same either way.

Now, the question is: am I crazy?

I think Braids has its root in an (abandoned) SMD voice card for the Ambika?

> am I crazy?

I’ve worked on such a project in 2012 and had almost completed a working prototype by the end of that year. The code I had written for it ultimately became Braids and Yarns. You’ll see images of the voicecard PCB on Mutable Instruments’ facebook page.

Originally, each voicecard was a STM32F103 + AD5644R ; then I combined two voices per card, with a STM32F405 MCU, CS4344 for audio, DAC128S085 for the CVs (PWM is a bad idea, the carrier will be horribly low if you want your 16 bits of resolution).

I evaluated the cost of launching it, and did not continue because I did not want my first manufactured project to be that complicated - and that would have been an investment in the 400k€ range.

In retrospect, I’m really happy I did not continue with it. Either it would have sucked, or it would have killed Mutable Instruments.

I have completely lost faith in the “digital oscillator + analog filter” formula. There are so many interesting digital techniques (including more than two thirds of those implemented in Braids) that do not really benefit from going through an analog filter. The result is already balanced enough, maybe some cutoff would help, but certainly not resonance. And don’t get me started on the ultra-conservative approach to routing and modulations used in Ambika…

This kind of variety you can find in all these digital oscillator models… It’s really suited for a Eurorack module. In a Euro system you can decide to throw the oscillator signal straight to an output, or through a filter, or through a VCA, or mix it with another layer…

As for the “XT” interface - I’m kind of unhappy with what I did with the Shruthi: all the parameters are exposed… except the most important… the modulation matrix!

I’ve experimented with tons of things since 2012 - some of these experiments having given birth to popular Eurorack modules. If I ever returned to a desktop synth, it will be something a bit more original…

> Now, the question is: am I crazy?

No. The question is: Are you strong enough to do it just for fun == for no profit. I would totally love it and salute you for doing it. But if I’ve learned one thing over my years of DIY: It’s going to be a shitload of work and you must be really serious about it to keep your motivation high enough to eventually complete it.

And believe me, I know, what I’m talking about. I’ve been thinking about an Ambika XT, to start with. Then a monosynth (where I had some very serious ambitions+circuits+prototypes until I realized that I actually need a sequencer). The (not up to date and incomplete) threads are here on the forums if you want to have a laugh at me ;-). It only gets serious when your first euphoria has passed. Will you go on?

Anyways, I’d love to see a SMD ambika rework! Do it!

Something like the modor nf-1 proves that doing a digital synth is fine. Would be nice to see another full synth, Euro isn’t for me in terms of synthesis despite having lots of cool modules.

Good point about the PWM - I hadn’t checked how many bits would be lost with an acceptably high carrier. More DACs would certainly increase cost and complexity…

… the cost (and lack of free time) would probably be the one thing stopping me from actually attempting something like this. Especially the XT with all the extra pots, knobs, and buttons. There is an assembly place local to where I live, but having them do the SMD work on some protos (which there would probably be a few of - only ever got a board 100% right in the first go once) would probably cost a good amount, as I really don’t want to mess around with it myself (done enough SMD stuff at work when it’s absolutely necessary, and it drives me bonkers).

I totally understand the lost faith in the “digital oscillator + analog filter” design formula - but that is kinda how it goes for standalone synths right now (although you will probably change that in the future I bet). I guess I just feel a little left out because I can’t really get into modular, but love the tech+music crossover. It’s a love hate thing - I love the idea of modular, but can’t justify the cost (to myself or my lovely wife). I follow the modular scene even though I have nary a module… even getting excited whenever a new Mutable module is announced. Making something like this would feel like being involved in some way, something that I can’t do with modular.

TheSlowGrowth - I know that feeling exactly, I always get ideas to start this or that project, thinking small at first. But then I think, well I might as well add this… or it won’t be worthwhile really, unless it does x thing. Then the scope is just too big to think about so I don’t start anything. Or I think, someone has already done something along these lines… so why bother. Rinse and repeat. But hey, your sequencer project looks legit! I’ll review your schematics again as you get further along :slight_smile:

Upbeat, thanks, you’ve been a great help! I’m honestly trying to finish the project this time. :wink: