[sdiy] Keyboards good for scrap / repurposing?

sleepy_dog at gmx.de sleepy_dog at gmx.de
Wed Feb 21 13:10:02 CET 2018


Oh, thanks for the warning, Brian!
I see, sure, the analog stuff is messy. I'll keep in mind not using this 
for a new design (like the keyboard).
I also was mixing up projects while forgetting to mention it: using 
unused ADC pot mux "slots" is for the UI panel of a small desktop-ish 
synth I'm building, not for a keyboard. It's a thing where I, on some 
weekends, add something when I feel like it, hand-wired prototype.
And there are some unused ADC slots and lots of wire mess in place, so 
it is very tempting to just use that also for the missing buttons.
I thought I might get away with it when using a small number of 
resistors (e.g. 4) per line, resulting in wide threshold ranges.

- Steve


rsdio at audiobanshee.com wrote:
> I strongly recommend against wiring multiple buttons to a single ADC input using resistors. That’s really only something to do if you have absolutely no other choice (*). It’s certainly not a good first design option. The problem is that ADC involves a relatively long delay; and it’s analog, so those buttons won’t always be producing the same values every time due to noise, temperature and resistor tolerances. You’ll end up spending too much time testing your software thresholds to make sure that they correctly detect each button alone and in combination. Then, if you have a voltage droop, the thresholds will be all wrong.
>
> In contrast, for the cost of diodes instead of resistors, you can connect those additional buttons to digital inputs that will not suffer from unpredictable analog thresholds. Not only that, but reading the buttons will be instantaneous, rather than having to wait for an A/D conversion. Trust me, you don’t want to introduce analog complexity where it isn’t needed. The code to read a digital input is much simpler and doesn’t require embedded thresholds.
>
> Your 37- or 43-key keyboard will have at least 3 to 5 spare switch positions, and assuming that’s not enough it’s a simple matter to add Rows to allow any number of buttons in addition to the keys. If you start with 7 Rows for a 37-key or 8 Rows for a 43-key, then you’ll have enough spares for 19 or 21 buttons, respectively. Using a ‘138 decoder, you might as well plan on 8 Columns anyway. That’s 11 GPIO pins to read 64 separate digital buttons.
>
> Brian
>
> * Note that the “multiple buttons on a single ADC channel via power-of-two resistor networks” is great for automotive or other situations where you might need to run a single wire a long distance and still carry multiple signals. But in your case you can wire everything within the same PCB or two in the same box.
>
> p.s. If those 37-key or 43-key keyboards are wired for Velocity switches, then you’ll need 10 or 11 Rows, respectively. Still, that’s no more than 14 GPIO pins for 88 separate digital switches. In other words, pick a CPU that isn’t constrained on GPIO. The last design I completed uses an ARM with 90 GPIO pins, and I still haven’t used them all.
>
>
> On Feb 20, 2018, at 6:15 AM, sleepy_dog at gmx.de wrote:
>> ADC mux, yes, the adc has on-board mux also, so I used 4 channels x 8x mux IC -> 32 pots.
>>
>> So far using only 26 of that.
>> So I just remember the MIDI keyboard "trick" mentioned earlier, of scanning buttons with the key scan mux already present. So far only 1 push encoder gets scanned button wise, no mux, so I think for extra buttons planned,  I'll be using that trick to scan the buttons via the pot mux that's conveniently in place :) A few roughly doubling resistance values in series each connected to a button, I guess I should easily get 4 (R) x 4 (lines) = 16 buttons covered with that.
>>
>> - Steve
>




More information about the Synth-diy mailing list