Finally some open source serious embed platoform is emerging...

[Deleted User]

First the background:
http://cdm.link/2019/03/dadamachines-doppler-fpga-open-music-hardware/

The new doppler board promises to meld the power of FPGA brains with microcontrollers and the accessibility of environments like Arduino. And the founder is so confident that could lead to new stuff, he’s making a “label” to help share your ideas.

>

doppler is a small, 39EUR development board packing both an ARM microcontroller and an FPGA. It could be the basis of music controllers, effects, synths – anything you can make run on those chips.

I remember talk about this possibility emerging time ago as next logical step into Live music production/performance... so let’s talk again?

I can see few amazing apps going standalone hardware... Do you?

brambos

What was wrong with Teensy? That was always already a powerful platform for music projects

[Deleted User]

@brambos said:
What was wrong with Teensy? That was always already a powerful platform for music projects

Teensy is quite similar but it isn’t an FPGA... no? The dev was implementing similar patching environment but IDK how easy/hard is coding for it.
I believe the full opensource side could make jump some devs into it. Also the more modules coming seems something the Teensy dev isn’t aimed for... but I could be wrong you know...

jonmoore

Outside of always being a little wary of Peter Kirn's latest technology to get excited about...

I must admit that I am a little intrigued as FPGA promises significantly better performance than Teensy based projects, with Max like ease of use. If it delivers on that promise, it's a significant development.

mrcanister

i get excited seeing tech like this and then i realize i'm the scarecrow from wizard of oz- going beyond the blink sketch with my arduino makes me feel like my brain is missing lol
i did a module on programming at college and was ok at it, but learning on my own with a book on c++ .... arggh

[Deleted User]

It should be more like max environment and even Beap! modules...

jonmoore

@TheDubbyLabby said:
It should be more like max environment and even Beap! modules...

Going back to @brambos's original point. I find Teensy perfectly adequate for the sort of Arduino music projects I do (mainly custom MIDI control stuff). The biggest benefit of Teensy is that it has a ton of documentation, YT videos and books available to help the less experienced.

brambos

@jonmoore said:

@TheDubbyLabby said:
It should be more like max environment and even Beap! modules...

Going back to @brambos's original point. I find Teensy perfectly adequate for the sort of Arduino music projects I do (mainly custom MIDI control stuff). The biggest benefit of Teensy is that it has a ton of documentation, YT videos and books available to help the less experienced.

I haven't dabbled in FPGA myself. Word on the street is that it requires a totally different way of thinking to implement an algorithm, but if that's abstracted from the user by a clever platform on top of it, it could be interesting. Teensy has impressed me: I could run my Troublemaker engine on T3.6 with very little latency. So that's a known quantity in my books.

jonmoore

@brambos said:

I could run my Troublemaker engine on T3.6 with very little latency. So that's a known quantity in my books.

Now that's some 'open source' goodness I'd love to get my hands on.

ohwell

@brambos said:
What was wrong with Teensy? That was always already a powerful platform for music projects

If I'm getting this right the promise is just lots DSP power at a very similar price point to T3.6, no? (Also, I'm guessing that the DSP per $ ratio of this approach is better (as of now) than just building a Tx.x teensy with a much more powerful ARM chip. (Otherwise the doppler approach really doesn't make sense..))

This part of the blurb caught my eye especially:

"I’ve focused on the FPGA powers here, because those are the new ones, but the micrcontroller side brings compatibility with existing libraries that allow you to combine some very useful features.

So, for instance, there’s USB host capability, which allows connecting all sorts of input devices, USB MIDI gadgets, and gaming controllers.

That frees up the FPGA to do audio only. Flip it around the other way, and you can use the microcontroller for audio, while the FPGA does … something else. The Teensy audio library will work on this chip, too – meaning a bunch of adafruit instructional content will be useful..."

I guess this begs the question: why want more DSP? To your point, I recall Emilie Gillet saying we are very very far from having explored all the things that can be done with the modest processing power of the ARM chips she has been using for a while.

One thing that comes to mind, though: AFAIK getting the most DSP mileage out of these chips requires great programming chops. Perhaps this platform will make it easier to do DSP hungry task with less expertly optimized code? By the same logic better-optimized code should get more dsp mileage than on a teensy?

rs2000

@brambos said:

@jonmoore said:

@TheDubbyLabby said:
It should be more like max environment and even Beap! modules...

Going back to @brambos's original point. I find Teensy perfectly adequate for the sort of Arduino music projects I do (mainly custom MIDI control stuff). The biggest benefit of Teensy is that it has a ton of documentation, YT videos and books available to help the less experienced.

I haven't dabbled in FPGA myself. Word on the street is that it requires a totally different way of thinking to implement an algorithm, but if that's abstracted from the user by a clever platform on top of it, it could be interesting.

That's the point. Doing audio DSP with an FPGA is about as fun as playing the piano with one finger. Abstraction from the user I would translate into a pre-built library of DSP blocks that will most likely face a tough tradeoff between FPGA cell count and actually useful DSP power.

Teensy has impressed me: I could run my Troublemaker engine on T3.6 with very little latency. So that's a known quantity in my books.

I've recently tried a few different cool solutions (mostly found in Github projects) for audio synthesis and I was really surprised what one can get out of an Atmel MEGA168 (Arduino Nano) without even coding in native assembler, just plain C
I'd say the best of them was an octave-divider based polyphonic string synth featuring the typical phaser-like effect, all on a $2 board

[Deleted User]

Meanwhile taking a deeper look into the CDM article and documentation we can find the Teensy Audio Library (full compatible with this project) also based in block patching ala pd/max

https://www.pjrc.com/teensy/gui/index.html

https://github.com/adafruit/Audio/
https://www.pjrc.com/teensy/td_libs_Audio.html

About audiomemory usage:
https://www.pjrc.com/teensy/td_libs_AudioConnection.html

About Interrupts:
https://www.pjrc.com/teensy/td_libs_AudioProcessorUsage.html

How to create new modules:
https://www.pjrc.com/teensy/td_libs_AudioNewObjects.html

Roadmap:
https://www.pjrc.com/teensy/td_libs_AudioRoadmap.html

Let's take in consideration this is the teensy/adafruit Library which is awesome but isn't the one dadamachines is going to develop dedicated to the board (but compatible as I said) so we can expect something similar but on stereoids and hopefully Teensy developer will help in improve all since he is an amazing human kind.

So to put light on this Doppler works with Arduino IDE ergo this and lots of coding out there.
FPGAs have the ability to become hardware from coding them and it could be possible to develop something in this way and later export it as standalone hardware with its own memory (inside the FPGA or attached as module) like GEM works on Max(msp) environment.

This is still young but I see a huge potential and not so complex coding metaphor to get an standalone box someday. It's like the missing piece on the puzzle between pd/Max environments and iOS developing (the other similar piece to this is MobMuPlat) so it's a matter of time that some kind of wrapper will emerge (it could be possible to port LibPd into this hardware and just make it run pd patches but I can imagine the dadamachines proposal gone even further since Zmors modulars is similar approaching)

From CDM article too:
https://ds0.me/oshug.pdf

FPGA structure:

• Modern FPGAs aren’t just LUTs and DFFs!
• Block RAM: dual port SRAM, usually 4-36 kbit
• DSP: multiplier and adders
• High speed IO blocks for DDR memory, HDMI, PCI Express, etc
• Even CPUs! (Zynq ARM cores, Virtex-II Pro PPC)

From https://blog.hackster.io/dadamachines-doppler-development-board-packs-both-a-sam-d51-and-ice40-fpga-f09a62aadd29

On the hardware end, the Doppler features an Arm Cortex-M4F (SAM D51), an FPGA 5000 LUT (1Mb RAM, 6 DSP cores, OSC, PLL), and is both Arduino IDE compatible and breadboard friendly. The board sports a micro USB port, 3.5V to 5.5V VCC, 4 X 4 LED matrix, 2X DAC pins, 10X ADC pins, and all GPIO have 3.3V logic level. It also offers a pair of user buttons, an AREF solder jumper, SWD program pin headers, and uses a UF2 bootloader with firmware uploaded via USB stick mode.

I hope this is helpful or almost helps to continue a healthy discussion about this possibility...

Telefunky

The 'problem' with FPGA is the so called 'IP' (pre-programmed function blocks), which are those 6 DSP cores in the example above.
For Altera FPGAs there was a vast availability of such IPs, but for a premium charge.
No idea what happened to all this after Intel aquired Altera a while ago and no idea about the details of that open source hardware/software combination.

To make a long story short: in any case none of the systems (CPU native, DSP or FPGA) will save you from knowledge about algorithms - which in the end define the audio quality that will leave the DA converter.
Results will vary widely, no system is superior by default - it simply depends on the task.

The famous Bricasti hardware reverb uses 6 relatively cheap Sharc Blackfin DSPs.
A specific feature of this chip is it's capability to handle program loops without checking a counter for boundary, which may be an advantage in time critical applications.

On the other end of the spectrum there's Fractal Audio's AxeFx guitar processor with a super high powered TigerSharc DSP, a very expensive chip.
The sole reason it's used in that guitar fx unit is that the developer knows this chip's programming inside out from a former military job.

Learning DSP code is a though thing, but the FPGA approach may be even tougher as it's more abstract from the usual coding paradigm.
But I definitely wouldn't want to discourage anyone from accepting the challenge.
I've recently posted a guitar example with just the reverb of a late 80s Boss ME-5 pedal.
This reverb/delay unit uses the same technology as FPGAs, but the chip was a factory programmed gate array. Just like those 'Lexichips' in Lexicon reverbs.
I really like the ME-5's reverb's color, which is very guitar focussed, but it shows that you get in fact nice results from the technology.

[Deleted User]

Please read the pdf I’ve linked
https://ds0.me/oshug.pdf

It talks about this. I can’t understand it so deeply but I think the development is focused towards that problem you point in mind...

[Deleted User]

https://cdm.link/2019/11/op-ed-korg-has-transformed-synthesizers-by-letting-them-run-plug-ins-says-sinevibes/

Telefunky

Smart move by Korg, jumping the train early as usual...
This way ambitous users can focus on audio programming, while the hardware and interface (affordable) already exist and add to Korg's revenues.

waka_x

what about
http://www.axoloti.com/
?!?!?!

[Deleted User]

@waka_x said:
what about
http://www.axoloti.com/
?!?!?!

Yes that too. Also ELK...
https://elk.audio/dev-kit/

[Deleted User]

[Deleted User]