Next-gen and mini Axoloti hardware discussion


#430

Sure,
A ready to code board is pure freedom for Makers like most of us.
We can do whatever we want with it. Self Taylor made :smiley:

A ready to use module will be usable by musicians without electronics skills.

Note that the later can be based on the former. These options are not contradictory, they are complementary.


#431

Personally I'd like a less "hacky" code environment. It's super confusing to try to edit code in each of the little tabs in the properties window, especially when you realize everything's being pasted into one file and you have to watch out for namespaces and stuff. And there's no real benefit to doing fixed point math everywhere, we're not on a Cortex M3. Make it user friendly for both the GUI people and the code people.


#432

Don't want to stir up any heated debate again but I do have some insights to share if anyone are serious about getting into audio professional/commercial business.

Any core boards should be as close to your own business as possible (design files, bill of materials etc). You should need to be able to switch to any PCB manufacturer at any moment for any reason.
If you have bought loads of periheral parts to have in stock (i e header cable with connectors/jacks as suggested to be of use in the connectivity of these boards in this thread), you'd easily be in deep trouble if you are only in second position in sourcing the core boards that you are planning to connect with your peripheral parts. What do you do if you have customers waiting to buying your products but your original core board supplier has withdrawn their ability to deliver for any reason and you don't have any design files etc?

Secondly, who would be responsible to handle any warranty of your products if there are a core board that you have pretty much no control over (any design alteration may affect your product capabilities drastically)?

Any type of connector/jack with header cable is quite an manual labour in production. Which unfortunately usually means expensive to buy unless you buy a huge amount from a supplier. There are quite some options out there regarding connectors/jacks to choose from. And what lengths should the header cables be provided in?

Big professional brands may/often use such solution between core, ui and i/o boards. Some of the reasons are ease of manufacturing, further development and/or alteration in any of those boards area. As well as ease of fault test/repair.

I haven't seen it myself regarding the connector/jacks themselves having a header cable at their end. But they probably exist somewhere in some product out there. Point is that the costs probably speaks against it unless it is a niche product. Which probably is why you/me/we don't see it more often in commercial products.

In production/assembly you often see either an i/o board solution (header - header) with specific configuration of connectors/jacks OR a core board with specific configuration of connectors/jacks.


#433

Hey everyone. I really appreciate your enthusiasm.

A few of you have contacted me directly with various concerns.

I want to clarify something: everything I've shown in this thread is my work. Johannes is not involved. The device itself will not be called Axoloti. It will ultimately be a separate open source project, derived from the original open source project.

We're approaching a point where it will make sense to move discussion about my project elsewhere, out of respect for the original project and to avoid confusion, etc. This transition will occur when pre-orders go live. Stay tuned for details. I'm not going to post anymore here until then.

I agree with @timeorspace. Let's keep it chill.

Jump on Discord here if you'd like to chat in the meantime: https://discord.gg/xeztjzh


#434

ohh as a solder freak this turns me on <3


#435

Hi! I need clarification. Is it planned that your akso board will have fpaa onboard? I would love to make an axo synth with analog filters but I'm not sure what is possible. If I wanted a full 8 voice poly would it be feasible with the fpaa to have 8 or 16 filters? I suppose this depends on the design of the filters but how exactly is the signal passed out to and back from the fpaa?


#436

at the moment there is only a axoloti-extended board planned as far as i know.


#437

This means what exactly? Like a shield type addition? I don't mind if it is not on the board, but I'm wondering if it would be possible to program the fpaa from the axo software. Also I'd like to know more about how sending signals back and forth to the fpaa would work... Like if we have an 8 voice synth presumably that would mean we need 8 channels out to the fpaas and 8 back...

I am very interested in an architecture that will allow for analog filters so I'd like to know what is realistic and what to expect.


#438

at the moment there is an akso board developed, which has a stronger processor but is very similar and compatible to the old board, so no analog addons. these may folllow in the form of shields, but i guess @urklang is better suited to give you an in depth answer


#439

Urklang has stated that he would like to do that in the future, to be able to use for example Axoloti objects, Csound classes, fpaa and maybe even VCV rack modules in the same environment. But I don't think you should expect this in the near future.


#440

@Ke10g The new base board is DSP-based just like the original core board. FPAA tech will come on an expansion board; it is not part of the base design. The exact details of signal flow between the two domains are to be determined. I definitely want that to be as flexible as possible as you mention. My estimate is that the expansion will have four FPAA devices. That should give you a rough estimate of processing headroom if you take a look at the Zrna documentation. A very rough rule of thumb is that you can do 8 filter poles per FPAA device, so that would be a maximum of 32 for the expansion or roughly a 4-pole filter per voice for 8 voices. The plan is to make it possible to program the FPAA expansion (and other expansions) directly from a future patcher.


#441

Amazing! Count me in for both your new akso board and this fpaa expansion board.


#442

:open_mouth:
Yes! Me too!


#443

@Ke10g- my opinion, FPAA isnโ€™t a great idea if it requires a D-A and an A-D, only to be mixed in Axoloti and sent back out through another D-A.

The FPAA operates on analog, so makes most sense after the final D-A. Or before input A-D, in the case of an Axoloti FX processor.


#444

@tele_player yes this is what I wondering. Can you not have the conversions happening in the fpaa allowing you to do the transfers between the axo and the fpaa digitally?


#445

Even if you could, itโ€™s still too many conversions. And you canโ€™t.


#446

Thanks for the clarification... If the analog pins were running at an appropriate frequency, in could work though no?


#447

There's some subtlety here. Tele is spot on in the sense that we need to be intelligent about avoiding excessive conversion, but I'm more optimistic about what we can achieve. The upstream and downstream cases are the easiest as mentioned. To what extent we allow more complex cases, i.e. feedback loops rather than a linear chain, is to be determined.

One subtlety is that parameters in the analog domain are under direct digital control. This means that, for example, a digital signal in the Axo domain could modulate a parameter in the FPAA domain purely through digital means without any conversion in the traditional sense. The signal can be "connected" to the analog domain without actually running out on a real wire as a voltage. So there could be a feedback loops without any extra conversion: FPAA into the normal Axo ADC, Axo DAC to audio output or FPAA, Axo Signal to FPAA digitally.

Technically you actually can do AD conversion in the FPAA itself and send out a digital signal back to the host processor. This is a feature of the particular FPAA we're using.

We also have the normal integrated processor ADC channels and DAC that can be used for exchanging signals with the FPAA domain independent of the normal Axo codec. In this case there would be an actual analog signal on a wire in between the two devices.

Just sketching here; the final solution will be some mixture of these approaches.


#449

This is really interesting. Thanks for the introduction to the topic.

Another thing I wonder, since I've only started thinking of fpaa is: what human ears like about analog filters is their various characteristics presumably not only due to their architecture or design but due to the components and materials used. The fpaa changes the architecture on the fly, but the material characteristics stay the same. What do people say about the sound of analog filters instantiated on fpaa vs analog filters built with discrete components? What I mean is, is it worth it? Do people even hear a difference between a dsp filter that is purely digital and an fpaa filter?


#450

Excellent questions!