Let's build a midi controller

[andrew]

taking some inspiration from the 16n here - I wanna build an open source box of knobs

my naked boards mc-24 just broke after a shaky relationship with the build quality, my launch control is ugly and large, 16n is beautiful but I slept on a pcb and I don’t need all the features. but I figured these things aren’t exactly rocket science, and I have access to some quality fab equipment before I finish undergrad this year

let’s build a thing

what I’m looking for:

• both a 4x4 and 4x8 array of knobs. matches the size of 64 and 128 grids. a low form factor would be great but it’s not a priority

• just a USB out is totally good. I just want to send high-resolution midi, maybe osc

• HIGH QUALITY potentiometers. the good stuff. and they don’t wobble (this is the problem I had with my mc-24). the mannequins modules have the best knobs I’ve ever tried out - it would be great to match that feel and durability

• no PCB. that’s the downside of 16n for me - I’m really looking for a build at home type of thing. I feel like there’s potential for this to be a beginner level project. I’ll open source it of course. I want to lower the entry level as much as I can

• I’m able to laser cut and even cnc a case out of wood, but I image I could even work out a design for hand tools

questions:

• will I run into noise issues ?

• what microcontroller should I use ? teensy seems popular but looks like I’m gonna run out of analog ins for the 4x8 size. I’m sure there’s 10000 other options tho

• potentiometers and caps ? anything from some gear we know and love that isn’t too hard to acquire in low quantities ?

• wobbliness ? seems like mounting all the pots on like a proto board inside or something would do the trick, but I feel like I run into this problem a lot on other devices. I want to shoot for something that’s rock solid if I can.

[andrew]

oh RIGHT this thread

Four-N 128 sized knob controller

[image] grouped into powers of four: 1, 4, 16. to address different levels of granularity. no visual feedback, just high quality pots. usb-midi for plug-and-play compliance, CC 1-21 i basically created a 128 grid sized shield for the teensy 3.1. or, it could just be considered a very expensive wire harness. the board uses the same high quality Alpha pots we import directly for our eurorack modules. the teensy code has one stage of noise filtering and is incredibly responsive. this is quite p…

probably covered for potentiometer options there. being relatively new to hardware stuff I’m still curious what the importance of a PCB is in simple builds like this

[wheelersounds]

Others will surely chime in, but I think a PCB will be a good idea if not totally necessary-- it’s likely that we’ll need to multiplex in order to get all those knobs into a micro, so there will need to be something that manages that. Although it’s possible to get a multiplex on a breakout and solder from the micro to the breakout boards, so maybe with a 16chan mux you can do it without a PCB. I forget how many analog ins are on a LC/3.2 vs 3.6 or even 4.
Another advantage to using a PCB (or proto) is that if you use board-mounted knobs, they’re wayyy less wobbly than if they’re mounted only to the panel. Because you have the rigidity of both the board and the panel keeping the knobs in place. PCB will be a bit cleaner but not much more than a protoboard.
Actually, if you’re open to using a protoboard you might not need to use a Mux breakout, and you can just get a through-hole version. This might add a bit to the overall size.
Regarding midi, Teensy has a great advantage in that setting it up as a USB midi device is very simple. The code will be super lean.

[andrew]

yea main main motivation for steering clear of a PCB was to keep the cost low while also not outsourcing to china , and also make it easier to build just one. and lower consequences for soldering I guess ? kinda want to make like more of a “recipe” than a “product” if that makes any sense.

3.6

https://www.pjrc.com/teensy/techspecs.html

meant to link up top but yea 3.6 just has 25 ins. If I could figure out something that’s just knobs > proto board > microcontroller that would be exactly what I’m looking for. just a straightforward soldering job so it would be easy for beginners ((me)). I’m pretty comfortable programming whatever controller.

[okyeron]

wobbliness ? seems like mounting all the pots on like a proto board inside

Fwiw you’ve just described mounting them to a pcb.

Work the panel height so it’s flush with the top of the pots and you can use the pot panel mounting nuts to hold them in place. This is what you see on nearly every modular synth device with pots.

Alpha brand pots (you can buy from any synth diy place like Thonk, etc) are the industry standard and are usually quite sturdy.

For a “local” pcb solution -OSHpark.

[andrew]

will the snap-in variety work or do I need a particular lead/mounting configuration to fit ontp a proto board ?

pulled this from four-N thread (something about 6mm vs 9mm yea?)

https://ie.rs-online.com/web/p/potentiometers/7293423/

[okyeron]

Some examples at thonk:
https://www.thonk.co.uk/shop/alpha-9mm-pots/
https://www.thonk.co.uk/shop/alpha-9mm-pots-dshaft/

The snap-in type should work with protoboard depending on the size of the mounting lugs. Those alphas have smaller lugs (I think that’s what those snap-in parts are called)

Also - US source: Modular Addict:
https://modularaddict.com/9mm-potentiometers

[andrew]

cool I’mma stick with what yr linking

[frankchannel]

I feel like “outsourcing to china” is looked at as a negative thing, and I’m just curious – if that is indeed the sentiment – why that is? I also feel like this is completely unavoidable (e.g., consider how many parts on a teensy or any other microcontroller are made in china, for instance)

[andrew]

making the assumption here that the lower cost = lower wages (not necessarily specific to china). definitely a larger conversation than DIY specific stuff - I’m sure it’s difficult to avoid completely (though I believe monome works hard to that end).

for this project specifically I’m really just looking to avoid manufacturing anything to lower the cost for single quantities - it should wind up being a simple circuit anyway so I’m not losing much. also, I do not know how to CAD :^) .

[andrew]

oh && does resistance matter ?

loose part list so far:

Teensy LC - DEV-13305 - SparkFun Electronics

SparkFun Analog/Digital MUX Breakout - CD74HC4067 - BOB-09056 - SparkFun...

https://www.thonk.co.uk/shop/alpha-9mm-pots-dshaft/
https://www.thonk.co.uk/shop/intellijel-grey-knobs-sifam-selco/

[joshhh]

Work the panel height so it’s flush with the top of the pots and you can use the pot panel mounting nuts to hold them in place. This is what you see on nearly every modular synth device with pots.

This is definitely a common solution, but there are notable exceptions, such as Make Noise being only PCB mounted with the shaft not anchored to anything…which is often a contributor to “wobbly” pot feel.

(not picking on MN or meaning to split hairs for @okyeron, hopefully just adding some perspective for @andrew/others)

I think PCBs are so common for DIY projects since otherwise you’re typically doing a lot of “panel wiring”, often literally mounting the pots first and then hand wiring the whole thing. Trying to connect 32 pots on a protoboard actually sounds even harder to me, since you only have copper on one side vs a two layer PCB, which helps a lot in routing around the three closely placed pins on a typical pot. Wires, while messy, can at least move in three dimensions

IMO, if you want a one-off and the goal is primarily to use the controller (as opposed to primarily a way to learn EAGLE or how to use multiplexers or whatever) I’d personally be inclined to:

• find the microcontroller dev board with the most analog inputs possible (Teensy 3.6 with 25 is the max I know of in a common dev board)
• lay out a controller that uses that many knobs
• get a bunch of “solder lug” panel mount Alpha pots from Tayda - these should work
• wire all those pots as voltage dividers, each one going to an analog input on the dev board
• find/write/adapt code to translate those 0-3.3V ADC readings to MIDI or whatever!

[andrew]

• lay out a controller that uses that many knobs

any particular disadvantage to using a through-hole multiplexer and less inputs instead ? I’m kinda set on 32 as opposed to 24 to keep up with the multiples of 4 logic that a lot of my grid stuff follows. I think the price would come out similar.

[joshhh]

No real downside for this application based on my understanding! It’s just a bit more complicated…or, rather, complicated in a slightly different way.

[ParanormalPatroler]

I will be following this thread with a l o t of interest, so thank you for making this. I need the smallest midi controller possible for a live case and there’s nothing out there that fits my needs or size. I wish I could make my own but that is beyond my skillset right now.

[thopa]

Its daunting to have to learn Eagle or Kicad at the beginning , but the results, and the future ideas you might have and will be able to transfer to a pcb are countless. Its not all debugging free but you will learn a lot from simple designing and building.
It might not make sense as other people have said, if this is going to be just a one off, or you dont want to spend money on soldering | electrical equipment.
There is a great joy to holding a device you have designed, build, tested and are going to use!

I

[wheelersounds]

oh && does resistance matter ?

Are you talkin’ pot resistance?
If so, my understanding is that resistance and slope are important – I’ve always seen 10k linear pots (B10K) because 10k is enough to bring the pins all the way low, and linear taper makes the most sense for knobs that are controlling parameters other than freq and vol.

[infovore]

Some brief notes having made a lot of different kinds of midi controllers in the past years (and having several rotary prototypes rendered up in various places):

• “no PCB” is a recipe for a) wobbliness and b) build issues down the line.
  • The former: well, you have an anchor between the thread on a pot and the front panel, but even screwed tight, there’s still play. If the pins of a pot are all connected to something solid, and that solid thing is also anchored to a panel, you remove wobbliness. Wobbliness is largely down to mechanics, not pot brands. Have a look at some of the Mannequins hardware to see how robustly those panels are mounted to the circuit boards.
  • “Build issues”: those 9mm alpha pots are designed to go into a PCB, that’s why they have tiny pins. You can get 12mm pots with “solder lugs” - big pins with holes in, which are designed for hand-wiriing. They’re common in guitar pedals. But every pot is three chances to make a mistake with the wiring looms, which for a 32-pot controller, is a recipe for pain or grumpiness from end users. And the moment you’re in to stripboard… I dunno, I’d just bang out a circuit board. Vertical torsion is basically the enemy, and a top/bottom panel that don’t move, combined with lots of anchors everywhere, is the way to be safe.
• PCBs do not have to be made overseas. There are lots of in-territory manufacturers in lots of countries. Whether or not people want to pay what local labour costs is another matter.
• Not making a single board, however, might be a good idea: clusters of 4x4 you can chain are likely to be more cost-effective. (I’ve definitely got a design where I did this. It’s quite elegant.)
• Don’t be distracted by numbers of analog inputs. Use a multiplexer. They are very straightforward to implement, and their design (as in, the physical layout of pins) encourages multiples. if you have, say, two 16-pin multiplexers, they can share “control” pins and even share an analog in: you activate the first one, read 16 pins of it, then activate the second, and do it again.
• In terms of microcontroller - although given your examples you mean dev board - you do not need anything fancy; you just need something that can act as a USB HID device with minimal fuss. Teensy LC is $10 and will be fine for this; similarly, any of the 8-bit “32u4” based Arduino-alikess - eg the $5 “pro micro” you can find on ebay - will be fine. Stay away from Atmega328 and similar, you want the USB with minimal fuss.
• I would recommend not relying on just specifying a recipe of ‘dev boards’ one can chain together - once you’re into ‘get this mux board’ you are in trouble when Sparkfun stop stocking it, or there’s a shortage - especially the boards that are just commodity components like those muxes.
• “It’s a simple circuit” - is it? Have you drawn out the schematic? There’s not many components, sure, one or two mux chips and a cap for each power supply, an MCU dev board, N control pots, possibly some resistors… but the wiring loom for this is going to be revolting. I appreciate I have a particular perspective, but having done my fair share of spaghetti wiring looms… it just feels like a risk.
• If you do want to learn schematic capture and layout, this is a great project to do it on, incidentally. There are easy parts available in tools for things like the Teensys and Arduino devboards, parts for the pots, and really the problem is learning the tools, none of which are user-friendly in any traditional way (though their madness has many upsides). One minor problem is that the multiplexer I’d use - the 74HC4067, which has sixteen ins - is only available as a surface-mount package, the 24-pin DIP version is obsolete and hard to track down. If you want to use through-hole multiplexers, you’re going to be chaining 8-to-1 chips.

[andrew]

clusters of 4x4 you can chain are likely to be more cost-effective.

this is smart ! I’m making three 4x4 sections actually

tbh I may start thinking about this as more of a prototype and then a finished thing

[Oootini]

I’ve built a no pcb midi controller into a Hammond box. It’s not given me any issues. Pots are bolted to the Hammond enclosure. Very simple, but works totally fine. Some basic notes and the teensy code for a 12 knob, 3 button box.

I used perfboard to space pots evenly, then had a second smaller board for teensy, which I wired to the main board. Easy Peasy.

aidanreilly/hammond_midi_controller