I'm of the opinion that faders are better than twist pots, and wanted something compact that I could use to control a bunch of parameters in an Ableton drum rack. I also wanted to solve the "jumping" problem that you have when using a controller with multiple banks - if you change a bank, you don't know what the real position of the knob should be, and it's easy to accidentally jump to a wildly different value than you want just by slightly moving the potentiometer. Here are the features that the LumaFader has, some of which help solve this problem.
Demo
Features
- 4 x long-throw faders for precise control
- 4 x mechanical keyswitches for bank management
- 16x RGB LEDs per fader to show the last fader position for that bank
- "Pickup" mode - only send a message when the last value is passed
- USB C for easy connection to your computer / phone / whatever
- Mini TRS Out to directly connect to a synth
- 17 banks = 68 total assignable sliders
Usage
I designed this to use with a drum kit in Ableton. For example, you can map the global sliders to overall kit FX (reverb, volume, filter, etc.). Then, map each bank to an individual sample's parameters. So holding one button will access 4 parameters of the drum sample, holding another button will access 4 parameters of the snare sample, etc.
- Move slider(s) to send CC messages
- Hold button(s) to access other banks of CCs
- Double click a bank button to lock the bank (otherwise the global bank will be used if no buttons held)
- CC messages are only sent if the slider moves past the last value sent (as displayed on the LEDs next to each slider)
- You can also enable jump mode - send CC messages immediately if the slider is moved
Design
I had already successfully designed a custom PCB based off of the RP2040 - see the Loopster 2.0 - and each time I design a PCB, I try to do something I haven't done before in order to improve. In this case the new elements were USB C - I've only ever used micro USB in the past - and the use of a large number of RGB LEDs (NeoPixels). I've used these pixels in the past, but this project uses 69 of them - a good factor of 4 more than the Loopster 2.0 has.
First, I had to add the RP2040 and the associated circuitry. Following the RP2040 hardware design guide I added the appropriate power regulator, memory chip, crystal, and decoupling caps. The only changes I made from the official design:
- Swapped the ceramic cap on the power supply with an electrolytic - something that a Redditor suggested. This seems to have eliminated the high pitched buzz you can sometimes hear.
- Swapped micro USB for USB-C. This ended up being simpler than I expected - just needed a few extra resistors
The rest of the circuitry is pretty straightforward - connect the slide pots to the analog pins of the RP2040, connect each button to a digital pin, and then connect the neopixels to one digital pin. These tiny addressable RGB LEDs are great in that they only require ONE digital pin for all of them. Finally, add the TRS MIDI Out circuit, which requires just one more digital pin.
Here is the full schematic
Layout
The goal was to make this device compact and portable, so efficient use of space was a priority. I used the middle of the board to house the main circuitry, the left for the buttons, and the right for the 4 large sliders. I decided to break out some additional GPIOs from the RP2040 since I didn't use them all. I don't have any plans for these, but perhaps in the future they can be used for something?
Enclosure
I usually go with a 3D printed enclosure made from PETG, but this time wanted to try the ol' "PCB Front Panel" trick. I drew up a simple front panel in KiCad and included holes for all of the LEDs, Sliders, and Buttons. I thought they came out great, especially considering how cheap they are to get made. I used some brass standoffs to connect the main board and the front panel.
The tiny LEDs shining through the panel holes were a bit harsh, so I designed little diffuser inserts and printed them out of transparent PETG, which did the trick. Note that the bottom layers are printed in black PETG in an effort to prevent too much bleeding between the pixels.
For the hell of it, I also created a nice Walnut case for one of them. I designed a router template in Fusion 360 and hand routed it out of a solid block of walnut. It was then hand sanded to a high grit and finished with a hardwax oil (Rubio Monocoat)
Software
The software is written in CircuitPython. I used it because of my familiarity from other projects, and because it's easy for a beginner to modify (for example if someone wanted to utilize the extra GPIO pins).
In an effort to try yet another new thing, I heavily used AI to help write the code. It's definitely a great tool if you use it appropriately, but I made the mistake of having it write large chunks of code that pretty much worked... until they didn't. Then I had to read through it all carefully, and as you may know, troubleshooting someone else's code is.. not very fun. In the future, I'll only have AI write small functions, deal with formatting and documentation, and other very specific tasks.
As always, the hardest part for me was just keeping track of the state of everything in a logical way, and dealing with detecting events (did the slider change enough? Was that a double click? What mode are we in - should we even care in this mode? etc.). This project has an extra layer of "keeping track of stuff" because of the "pickup mode" behavior. We have to store the previously sent CC value sent for each CC number in order to know how to set the lights up when changing the bank, and where the "passthrough" point should be for each bank / slider combo.
Tools used
- Fusion 360 - Light diffusers, case modeling
- KiCad - Schematics, board layout, front panel
- JLCPCB - Manufactured and assembled PCBs (minus the sliders, buttons, and aux jack)
- VSCode - IDE for development
- Thonny - Loading and testing the code
Overall I'm pleased with the final product - It's the most professional looking device I've created (using a PCB for the front panel rather than a 3D printed enclosure helped a lot on this front).
Takeaways
- Don't rely on AI to write more than like ~20 lines of code at a time. It's tempting because it really does work well for the most part, but troubleshooting later is quite a headache.
- USB C isn't as hard to deal with as I thought - use this for everything in the future
- Using PCBs as a front panel is surprisingly cheap and professional looking
- Using tons of tiny NeoPixels is not a problem if you are mindful of brightness. ~70 pixels lit up at once at around 50% brightness @ 3.3V draws less than 150mA, including powering the RP2040.
- Always overclock the RP2040. you can safely double the speed and it only takes like 1 line of code
