Built a controller keyboard using NG (issue #1: low velocity resolution, issue #2: random pitch wheel values)

[tago]

Hi,

I built two identical controller keyboards using CORE_STM32F4 running MIDIbox NG. They currently feature analog ins, two wheels and octave up/down buttons. It was very hard and time consuming to build these without a proper workshop (used mainly beech wood and aluminium).

Issues:

#1 One big problem i have is that the velocity resolution is very low. It looks like just 5-10 values are generated, nothing in between. It's too coarse for playing properly.

#2 Another problem is that i get random pitch wheel values when in zero position. I wonder if there is kind of a plateau range around the zero position to mitigate that. Or maybe it's because of a faulty pot?

 

Thanks for your help

Edited February 4, 2023 by tago

[Phatline]

look into ng documentation if there can be set a offset for the middle position so it stays on a position... because pots directly to the core is always a bit random... better use for example: http://www.ucapps.de/mbhp_ainser8.html

then you have less random values

also check the quality off PSU...off course a faulty pot can be the reason too

Edited February 5, 2023 by Phatline

[tago]

46 minutes ago, Phatline said:

because pots directly to the core is always a bit random...

Hi Phatline, i'm using an AINSER8 module (see that perf board on the third photo). So pots aren't directly connected to the core. Power is coming via USB.

I'll take a look into the docs for a middle offset function. I think Doepfer is calling it 'plateau' on their keyboard controller boards. I assume there is no way around something like that.

EDIT: do you think it could be because the pot is old/used?

Edited February 5, 2023 by tago

[Phatline]

could be... you can try to filter out some psu-spikes by soldering a 100nf (maybe add also a 1uF or higher for too more stabilize the psu as addition) cap between + and - on the potis legs (most off the time these are the outer 2 legs off the trio)

[Phatline]

ps: cant see much off the vector board.... so its a bit watching in the magic "glas kugel"

[tago]

@Phatline Here the pcb layout.

There is a voltage reference (U3) (center bottom - looks like a transistor) and on the right side an opamp (U2) to scale the voltage range because of the limited rotation angles of the wheel pots. I did that with the help and recommendation of @FantomXR and others. Can't find the original thread at the moment.

See how the Vref (red) goes to the opamp (U2) and then to the jumper J8 where both wheels are connected to.