How would I find someone to do this work?

[Mbira]

Hi guys,
I've been here at Groupdiy since the beginning.  I have done lots of projects, and learned a ton, but I am by no stretch of the imagination an expert on electronics.  I've also become increasingly busy performing and building instruments so a lot of my electronics DIY has taken a back seat.

I'm writing because I'm hoping to find someone here (since I know most of the older folks here) who would be interested in doing some paid work.  Basically, I'm wanting to find someone that will help me design a "bridge" that will take the audio signals from piezo discs on my marimba and convert them into CV signals.  I use alternate tunings, so I want to be able to individually tune the CV voltage that each individual key produces.  My goal is to be able to plug directly into a modular synth and play my marimba with no latency.  I currently am playing through an Audio to MIDI converter that I built, but the latency is enough to make me not be able to do what I need.  Anyone interested or know someone that would be interested?

 

[JohnRoberts]

I have a friend in OZ who has some expertise in the area... He designed the Peavey Midi-bass  a guitar with midi output, and later did his own guitar to synth device (Chici-pik?).  As i recall it was far from trivial. I can't find any evidence of that on the WWW now but here's his current version bass to midi device.

http://www.industrialradio.com.au/

I haven't spoken with him for a few years and kind of doubt he is looking for outside projects but he would be a good guy to ask questions.

https://www.youtube.com/watch?v=dGOICWocxVE

JR

 

[Mbira]

I "think" that what I'm hoping to achieve is much simpler than a midi guitar.  On my marimba, each key has it's own piezo trigger installed.  So a marimba has around 20 individual triggers.  Those are all going to a DB25 connector.  My thought is that each of those triggers would feed some sort of comparator that would then have a multi-turn trim pot to exactly set the CV output to the specific pitch of that specific note.  All those individual CV outputs would then be combined and sent out of one master CV output into a modular synth....

I'm sure I'm missing some important steps there though....

 

[JohnRoberts]

I have been accused of seeing every design problem in world as something to solve with a microprocessor, and 20 analog inputs is probably pushing it for one chip,  (I just checked and Microchip has some 16b processors with 22 A/D  inputs. )

Using a PWM for output, you could calibrate or dial in each note voltage one at a time, instead of using 20 10-turn pots and save the settings into flash memory.

The output could be combined in the digital domain before outputting a single voltage to your synth...

Do you just play the note for a finite time, or change pitch every time a new note is played (or both)?

I've never tried to multiplex 20 analog inputs, but just for triggers the signal quality doesn't have to be great.

Latency would be the lag in the LPF to convert the PWM to DC voltage, and time to scan and compare 20 inputs sequentially, but those processors run reasonably fast (in audio terms) and comparing the triggers voltage is only a few clock tics, each. .

If the piezo outputs were hot enough you could just make them digital inputs into the micro, and trigger interrupts when those input pins change state making them arbitrarily fast. Still have the output filter lag.

JR

 

[sodderboy]

Two of these might bring joy.  They can probably trigger directly from the piezo.

He used to a "pocket control" that would be the exact ticket.  He has config suggestions for the 192 that is available now that should provide gating etc.
Mike

 

[Mbira]

John-if I'm not having to scan inputs then there will be much less latency, right?  I don't mind building the same simple circuit 20x.  These aren't for production, but for my own use.  The main complexity that I can imagine is that the piezo is going to have some signal coming thru even when not struck, so they will need a threshold.  Also, in the circuit there needs to be some way to have just the peak be the thing that triggers the CV so that one hit won't trigger lots of ghost notes.  There will pretty much only be two notes struck at once at the most (two mallets are used).  So that adds a level of complexity because it'd be cool to have TWO CV outputs that are shooting out so I could have duophonic response. 

sodderboy-that link's not working for me....

 

[JohnRoberts]

John-if I'm not having to scan inputs then there will be much less latency, right?

Yes, an input change interrupt would have zero latency.

I'd have to dig into the specific processor numbers to determine latency. Performing 20 A/D conversions takes time, but you can trade off speed for bit depth and don't need full 10-12 bit resolution for trigger detection.  Further I would set up the A/D processing into two batches of 10 inputs, so while you are scanning the last 10 samples, the next 10 new conversions are being performed.  Once you detect a hit, you set a flag for that and keep sampling for the next hit.

The flag for a hit gets services between A/D batch processing.

This sounds like a lot of time but these processors can clock at tens of MHz, and for relaxed bit depth the A/D conversions can be pretty damn fast.

I don't mind building the same simple circuit 20x.  These aren't for production, but for my own use.  The main complexity that I can imagine is that the piezo is going to have some signal coming thru even when not struck, so they will need a threshold.  Also, in the circuit there needs to be some way to have just the peak be the thing that triggers the CV so that one hit won't trigger lots of ghost notes.  There will pretty much only be two notes struck at once at the most (two mallets are used).  So that adds a level of complexity because it'd be cool to have TWO CV outputs that are shooting out so I could have duophonic response. 

This kind of smart "what if" decision making is far easier in the digital domain than using analog circuitry.

sodderboy-that link's not working for me....

me neither.

JR

 

[sodderboy]

fixed

doepfer.de