Wish List for future additions and upgrades to board

#1

SO here's what I'd like to see added to future revisions of the design

  1. An FPGA.
    Take much of the processing load off the CPU and increase overall processing power enormously.

  2. More RAM, 4 gig, Maximum a 32 bit CPU can address, why not? Load more and bigger samples and play back faster

  3. Digital i/o, spdif and ADAT

  4. clock in and out for syncing to other devices

  5. More in/out channels

  6. more flexible analog audio, balanced lines in and out, level control, good preamps for recording, hi impedance input for guitar etc.

I figure much of this functionality, apart from the RAM and FPGA could be put on addon boards that could connect to headers on the existing pinouts.

#2

1
Dunno what that is, but if it gives more processing power I am up for it.

2
Not sure if it is possibe, cause I am not a programmer, but again, sounds like a great idea, if it doesnt mean that the price will rise too much.

3
There is an object in development, that lets you stream audio digitally from one Axoloti to another. I look forward to try this out.

4
You can sync over midi using regular midi clock. BUT you cannot pass clock through Axoloti. If you want to distribute Axolotis clock to more than one device, I'd suggest getting one of these midihubs. It is cheap and works really well, it even got its own internal clock. 1 in 6 out:
https://www.tindie.com/products/hotchk155/midihub-6-way-midi-thru-and-metronome-module/

5
Yeah, why not. Using streaming digital audio, as described above, will give you some more I/O. QThen you can use the digital stream between Axoloti and the in&out for other synth or effects.

6
I think you can add these thing on your own. Axoloti is designed so it is very easy to add a preamp or whatever yourself, even for a noob like me :wink: I have been looking at thw one in the link below. It has got an analog drive stage. THis could be used to beef up Axolotis sound. Unfortunaly itr is not in prodution, if you want it you have to build it yourself from his github descriptions.

#3

Thanks, I had thought most of those things could be added to the existing board using the pins that are broken out and will be working towards building add on boards to do most of those things.

FPGA stands for Field Programmable Gate Array, programmable logic that can be reconfigured to do whatever processing you need, I had thought it's primary use would be to do hardware FFT taking a huge load of the CPU and increasing processing power massively, it should be possible to do FFT's in only a few clock cycles using an FPGA as opposed the to hundreds or even thousands needed to do it in the cpu.

Could also be programmed with any other cpu heavy task being done already or for new more complex processing algorithms.

MOre RAM shouldn't be hard at all , the CPU has the capability to address up to 4 gig, that may require using a package that has more pins than the current one though to get a 32bit address and data bus.

STM32 comes in a few different packages with widely varying package size and number of pins out.

#4

FPGAs are hard to program, require a proprietary compiler/toolchain, involve long compilation times. The latest Xilinx Vivado Design Suite is a 11GB download. It is tempting to think of an fpga as a "modular digital audio fabric", where modules are mapped to physical areas on the fpga fabric. However, that does not come close to the clock frequencies that fpga's are capable of, while running out of fabric area. So to exploit the power, it needs to share functional units across audio objects, and time-multiplex them. And then it starts to look similar to implementing a DSP architecture in a FPGA, with the benefit that custom instructions can be implemented. And then write a compiler for that... but life's too short. Unless it is only used as a fixed function co-processor for, say massive FIR filtering or FFT's. But even then, there will be limits in communication bandwidth.

More RAM, I believe it is possible to replace the on-board 8MB sdram with a 32MB or maybe a 64MB sdram chip, but this really requires specific tools and skills ("BGA rework") and also firmware modifications, I really can't support such surgery.

#5

Strum, I am very interested in seeing what you're doing with the FPGA add-on board. I've been developing a board to connect to the Axoloti, and the serial communication is rock solid and never steps on the audio processing or vice versa.

Additionally, the mix of graphical and code programming has been a real timesaver, both in building my infrastructure quickly and that it's already been tested to be robust. So, I recommend letting Axoloti do what it does best (Building synth models, MIDI, Audio I/O, Processing potentiometers (works great with 16 ch mux on the I2C line) and focus on what the FPGA can do, but an Axoloti cannot. But whatever your idea is, I hope to see it come to fruition.

What sorts of features / applications are you thinking of? Looking forward to hearing more :smile:.

#6

Sorry I think you misunderstood me there. I was suggesting that a FPGA be added to future revisions of the board not that I was going to build an addon board with one on it, I wouldn't have thought the bandwidth of the pinouts available on the existing board would be enough to make effective use of an FPGA off the main board, I could be wrong about that though and it may be a viable option to add processing power without a new board.

#7

I know that time has passed, and you probably already know
but now there are new FPGAs that have been hacked and freeware is being created with them .. using free software ...
As an example the ICE40 series of Lattice.
There is a FPGAwars community using the board
FPGA development board IceZUM ALHAMBRA 2.
Multiplatform: Linux / Mac / Windows
For more information https://github.com/fpgawars/icezum/wiki