Newbie - hardware question - where to begin ?

[johndough]

Hi
I want to "play" with some DSP programming.
and start with a "real world" powerful chip
(I already have Arduino board),
which later can be used commercially, and not a chip which can be used only for "proof of concept".

A "must" is that it has a C compiler (unlike Spin FV-1).

I read posts here that mention Analog Devices (SHARC/BlackFin),
( http://www.diystompboxes.com/smfforum/index.php?topic=112463.0 )
And I also saw that Strymon use SHARC (at least on some pedals - http://www.strymon.net/dsp-anyway/)

Any other options you can suggest, or ADI is the way to go ?
Of course I prefer something that have a ready development board
(I'm not good with soldering...), I dont know if they all come with that.

Thanks for any help,
Regards.

Si.

[Digital Larry]

I'll pop off with a couple of ideas.  Keep in mind that I'm pretty enamored of the FV-1 in spite of its limitations.  Moving right along,

You might consider developing C DSP code right on your x86 processor (Windows/Mac/Linux).  Challenge with this is ability to cost reduce it adequately while maintaining adequate performance.

Take a look at Faust.  https://en.wikipedia.org/wiki/FAUST_(programming_language)

Mark Seel has posted some designs he's done using Xmos chips on this Forum.  http://www.xmos.com/  Last I recall, Mark was an applications engineer with Xmos so that gives him a certain advantage.

The Hoxton Owl uses an ARM Cortex and can apparently import patches from Pure Data.

http://hoxtonowl.com/

None of these are general purpose DSPs, but maybe it will expand your vision a little.  Let us know what you get into!

[slacker]

I think the problem with the Blackfin or SHARC is that the development software will cost you about $1,000 or are there other alternatives?
http://www.analog.com/en/design-center/processors-and-dsp/evaluation-and-development-software/adswt-cces.html#dsp-overview

An audio codec, maybe some SRAM and a 16 or 32 bit micro controller makes a nice cheap decent quality DSP rig to play with. You'll probably have to do some soldering though. I have something along those lines that I'll be sharing soon.

[johndough]

Thanks a lot for your help !

The Hoxton Owl Pedal looks perfect except for the price (~ $360).

I was thinking about doing some simple stuff, in pedals format, just for fun/hobby,
and it's nice enough, I can build some more for my friends.

So the "pedals" format takes x86 out of the equation as far as I can tell.

As for the FV-1, I prefer something with c/c++, I will feel much more comfortable with that.
I also read in other forum (that was before I found about this forum, so please excuse me    )
"the EQD Rainbow Machine appears to be using the default pitch-shifter program on the FV-1 in combination with various feedback mechanisms to do its wacky sounds"
here:
http://www.thegearpage.net/board/index.php?threads/educate-me-us-on-reverb-dsps-belton-others.1111366/#post-13494913

And from this I understand, that the capabilities of the FV-1 are limited and require extra hardware to achieve some effects.
Am I wrong/correct ?
I want the option to do everything with code.

Any opinion much appreciated.
Thanks.

[Digital Larry]

An Intel Edison is an x86 experimenter board in about 3 square inches.  Doesn't run Windows to my knowledge but I've goofed around with Linux on it.  As far as suitability for DSP I have no idea.

Here's a blog of a guy who got it working with USB audio dongle.  Others report using it successfully with its built-in bluetooth.

http://alextgalileo.altervista.org/blog/lets-make-noise-play-audio-edison/

Spin FV-1 has an optimized architecture that can do things like LFOs, pitch shifting, all pass filters for reverb structures very efficiently.  I've written a CAD application for it that lets you build effects from small functional blocks, to avoid writing code completely.  At that point you are limited by the blocks i've provided, but assuming you can deal with Java and the Eclipse IDE, you could define your own.  To my knowledge, nobody has actually done this yet.  In everyone's defense, I haven't exactly documented it in a straightforward way.

[slacker]

"the EQD Rainbow Machine appears to be using the default pitch-shifter program on the FV-1 in combination with various feedback mechanisms to do its wacky sounds"

The FV-1 comes with 7 built in programs that you can use without doing any programming, companies do use it like that, EHX for example have a couple of pedals that use the built in reverbs and I think the tremolo.
One of the built in effects has a pitch shifter on one channel and a delay on the other do it's possible EQD are using that and doing some mixing or feedback externally. The FV-1 only supports 3 pots for adjusting settings and the Rainbow machine has 6 pots so they must be doing something outside of the FV-1.

Here's some examples of what the FV-1 can do

Ignore the LCD that's handled by a separate micro controller. All the effects are done purely in code on the FV-1, there's a rainbow machine tyoe thing at about 3 minutes.

Here's a couple more.

[Digital Larry]

The OWL uses an ARM Cortex M4 chip, same as is used on some experimenter's boards that are pretty inexpensive, e.g.

http://www.mouser.com/ProductDetail/STMicroelectronics/NUCLEO-F401RE/?qs=fK8dlpkaUMvGeToFJ6rzdA%3D%3D&gclid=CK_qtpnU3MoCFVh7vQodxEIFWw

You just have the challenge then of attaching a decent amount of RAM for delays and audio I/O.

[johndough]

Thanks for your help.
I will do some reading on the chips you've mentioned,
and see how it goes.

Thanks again !