Multi effects processor with Micro controller

[Pedro Cardoso]

Dear all,
I am starting a project in effects development using micro controllers.
I would like to enter the DSP world and try to develop several modulation effects. I will do the distortions in analog fashion :-).

So I am looking for a micro controller capable of doing DSP suitable for audio applications. I started looking at the STM32 boards, which seem powerful and are quite cheap (here in the UK from 10 pounds upwards with a Cortex M0).
The STM32F411E (14 pounds), for example, uses an ARM Cortex M4F and has an integrated audio DAC with class D driver. The board seems pretty good to begin with but it doesn't have audio input, just some mems mics (maybe it will be possible to connect the guitar to that input, taking the mics off).
Another option would be the TI MSP432 Launchpad Evaluation Board (14 pounds), but I don't know if it will be suitable for my application.

So I need your opinion on the best development budget board with a DSP capable micro Controller.

Thanks in advance,

Pedro

[ElectricDruid]

I wouldn't even describe what you're talking about as a "microcontroller"! The STM F4's are 32-bit processors and quite capable of serious DSP - this is not a little micro controller, it's a heavy-duty computer.

You could have a look at the Axoloti platform. This provides STM-based hardware and a software framework to use it. It already includes audio inputs and outputs and MIDI and so on, so you're pretty much there, hardware-wise. If you wanted to get "bare metal" with it, you could.

HTH,
Tom

[izikg]

hi
i have similar idea runing in my head .
like you i was thinking that the overdrive/distortion should be analog while the delay and other time effect should be digital.

the st32f looks fairly good with some open issue .
regarding the a/d d/a,  will a 12 bit be sufficient,  specialy since on a delay line where we sum the analog undelayed signal to a 12bit delayed one,  will that be high quality
another issue is the memory size
another issue is digital noise

one other design thought i have is to control the analog parts digitaly so there will be many  configuration options with minimum analog pot. maybe also to add a bluetooth and control it all from a smartphone

i am looking forward to see where will this all go to

[PRR]

> 12bit .... will that be high quality....

Dammit! This is not the 20th century any more!

What does another 4 bits cost now? $10? It is more than worth it to go 16-bit.

Back three dozen years ago now, some LPs were cut with a 12-bit 2-second delay in the mastering path. This really simplifies (cheapens) the mastering process. Customers and their dealers returned many of these cuts. They had never heard a digital artifact, did not know what was wrong, but the sound was clearly inferior, "dirty like bad vinyl".

The dynamic range of club music exceeds the range of a common LP.

ONLY in pedal-effects world do we put up with abominable delays. I have a $10 cell-phone with almost enough ADC and more than enough RAM and DAC to do 96dB dynamic range at great length.

[Pedro Cardoso]

You could have a look at the Axoloti platform.

Tom, I have taken a look at the board and it seems quite good to me and capable of doing what I want. However I think that, for guitar, 24-bit converters might be an overkill and maybe there is a less expensive board with relatively less resolution converters. Moreover i don't know if I need the MIDI ports ....

Main Specs:

• 168MHz STM32F427 microcontroller
• 24bit/96kHz capable stereo audio ADC/DAC (editor and firmware currently only supports 24bit/48kHz)
• 8MB SDRam

Do you thing that if I remove the mems mic from the board, I can plug-in there the guitar ? I don't really know the voltage levels to expect from that mic ...

Thanks,
Pedro
[/list]

[Mark Hammer]

I don't know nearly enough about a/d conversion, but I'll mention something that effects pioneer Roger Mayer mentioned in the interview with him in Dave Hunter's effects-pedal book.  Hunter promotes use of BBDs over digital for delay, in the interview, largely on the basis of how they behave in the decay.  He emphasizes that while full digital resolution can be applied to the peaks of signals, as the signal decays during final repeats, one ends up applying fewer and fewer "bits" of that resolution to it, leading to a poorer-quality sound.

That's what HE says, and should be viewed through the lens of whatever technology was available to him and the industry at the time the interview was conducted.  But the gist of his argument makes sense: base your resolution needs on the softest parts of the input signal, not the loudest ones.  If a few dollars more gets you 24-bit, go with that.

[Pedro Cardoso]

Yes, I think your are right, by the end it is worth spending some dollars/pounds/euros :-) .
My point was based on the circuit used by boss for their famous DD-2 and DD-3. As these pedals use 12 bit DAC/ADC I thought it should be more than enough, if there was a significant price difference. By the end, Slash uses them ;-)

[ElectricDruid]

24-bit convertors aren't expensive these days really. What's much harder is finding the kind of variable-rate 12-bit ADC/DAC chips that things like the Boss pedals you mention used. But at this point, that's just a question fo software, right? After all, you don't *have* to use all 24-bits if you don't want to. You can always throw away some of the incoming data by bitcrushing it, and then only use some of the output resolution you have available. And you can resample to simulate the variable rate if required.

So yeah, 12-bit might be enough, and might even give you "a certain sound", but you can for sure get the same sound with 24-bit convertors and 32-bit processing and some good programming. It might be overkill, but it's probably cheaper nowadays, and it's certainly available , which might not be the case for the more "original" solution.

Tom

[blackieNYC]

There is a DSP forum right here within DIY Stompboxes, btw.

[PRR]

> the Boss pedals you mention...

1983!! In 1983 a MegaByte of RAM cost $2,396 (*)

2012 through 2017 we have been running $0.007/MB or less. 300,000 times cheaper than in days of DD-2.

(This is very much along Moore's Law. I work it out as should-be 500,000X cheaper now, but the price flutters and I did not plot a running average.)

The ONLY reason for 8-bit 12-bit audio is to save RAM (and ADC/DAC) cost. It's like thatching your roof with straw to save money. Low-price shingles have made thatch (and its problems) a thing of the past.

[Pedro Cardoso]

Thank you all for your comments.
I guess all of you are right, there's no sense to go with a 12bit ADC when higher bit rate are available a very low cost. :-)

I will try to "savage" a STM32F407 by removing the mems micro and getting a way of of "converting" analog to PDM, so it can be understood by the uC.

If I fail to do that, I will buy the board that Tom suggested.

Thanks,
Pedro