NeoPWM, the best LED controller in FTC
Non-obvious TODO
Connect the NeoPixel, PWM, and CC data lines to pins on the rp2040 that makes sense.Add mounting holes to the PCBMake second PCB for intergrated led panel
Total time: 32 hours
5/17/25, 4 hours
General Idea
In the FIRST Tech Challenge, people often want LEDs on their robot for driver feedback and for looks. There isn't a great way to control neopixel, because IĀ²C is really slow. REV made the REV Blinkin, which allows for control over PWM, but you can't do very much configuartion, and you have to run it on the main robot battery. For high level teams with tons of other devices, this isn't feasible. The NeoPWM will allow for advanced configuration over USB, super fast control using PWM, and power using a 5V USB power bank.
FTC Rule stuff
Because there will be two ground planes (the robot battery and the USB battery), I will need to use a optocoupler to pass the PWM signal to the MCU, which will be powered by the USB battery. Because it contains a programable MCU, in order for it to be legal, it needs to be a Commercial Off the Shelf (COTS) component, so essentially I need to design it, and then sell it and get a EIN and a bunch of annoying legal paperwork. :sob:
USB-C Port
I want to use a USB-C port for firmware updates and such, as well as for power during standard operations. I think the TYPE-C-31-M-12 by Korean Hroparts Elec might be a decent option, it supports USB 2.0 and basic power mode config, so I can pull 15w of power by pulling CC1 and CC2 to ground.
MCU
I think that a RP2040 is probably the right choice, given the amazing documatation and community support. It's also pretty cheap on JLC, being less than a dollar if you order more than 10.
KiCAD VCS
KiCAD 9.0 supports git, so I can just make the NeoPWM the base folder for the KiCAD project. Yay!
Starting on the schematic
I started working on the RP2040 part of the device, and finshing the power input, USB-C port, the crystal oscillator, and the 16Mb flash module (which is a basic part on JLC yay!)
5/19/25, 5 hours
Power (ahhhhhhh)
Leds consume a lot of power. I want to support using a USB-C battery bank, which should allow this to have minimal performance impact. To do this I have 2 options, use USB PD or pull the CC pins to ground with 5.1kĪ© resistors. Using USB PD would allow for a lot of power, up to 5a at 12v = 60w, but it is also very complicated and expensive to implement. The simpler only supplies 3a at 5v = 15w, but that is enough to power 50 leds at full white which should be enough. I will add a software system to manage the power used to ensure it stays below 15w by dimming the LEDs, but this seems like the best idea.
Optocoupler
I have elected use the LTV-357T for passing the PWM signal from the control hub ground level to the rp2040 ground level, mainly due it's availability on JLC.
Bom
I went through and got most of the BOM done, still need to add whatever parts I'm gonna be using for the case though. Also whatever parts I realize I need to add in a week.
3D Models
I've started to add 3D models to some of the footprints that don't have them, this is mianly to assist in making the case down the road (wow im thinking about future self thats crazy)
Done with schematic!
I finished the schematic, assigned all the footprints, and made sure it transferred without issue!!
5/20/25, 4 hours
First pass of PCB
I made my first pass of routing the PCB, although there's still a bunch of tiny things i need to do, as well as fixing a bunch of DRC errors
Optocoupler swap
I swapped the optocoupler for the LTV-217-B-G, mainly cause it's a bsic part of JLC, which saves $3. Also it's a SMD part, which makes case design easier because I don't have stuff sticking out of the bottom of the board.
PCB general layout done
I finished the general layout of the PCB, still need to add mounting holes and a reset button. Still conflicted on whether on not to add a status led. Also KiCAD's 3D view looks soooo cool with raytracing on!
Mounting holes
I added some M2 mounting holes, these will attach the PCB to the 3D printed case!
6/1/25, 4 hours
Minor tweaks
Moved the oscillator around to save some space. Debating whether or not to move the PWM signal input over to save space and make the pcb smaller (and cheaper).
Less minor tweaks
Moved the mounting holes to be in the corners and added a USB-C cable keepout zone. Also added some silkscreen stuff!
6/2/25, 6 hours
I/O layout changes
I decided to move the led output to the left side and the PWM input to the right side to save space. This also allows it to fit into a piece of GoBilda channel, a very common build system in FTC. I also made a first draft of the case, but realized I was gonna have to completely remake it to account the led panel I have yet to make (probably gonna do that today)
More I/O layout tweaks
I moved the USB-C port to the exact center cuz it was odd it being barely on the side. I also moved the PWM and LED connectors out 3mm to get outside of the case and make plugging stuff in easier.
Optional LED panel
Sometime you might want a intergrated panel, like the GoBilda RGB light. However, some teams would only use the LED strip, so having the panel would be useless. I could've accomplished this with a completely different 2 sided PCBA, but that would be very expensive so I decided to have a optional daughterboard which communicates via a 3 pin vertical Dupont connector.
Very rough draft of case
I made a rough draft of the case, and realized a couple things I need to change, including the location of the daughterboard interconnect. This is because the bolt will intersect it's location, which won't work. I also learned to not directly derive stuff from the PCB CAD, because any changes to the PCB breaks everything
6/3/25, 3 hours
Case
Originally I wanted to have m2 screws holding the top and bottom parts of the case together, but after I made a rough draft I printed it out and pressfitting it actually works quite well, so I'm probably gonna do that cuz m2 screws seem easy to mess up. I still have to make the case for the version with the intergrated light tho.
6/4/25, 1 hour
Standoffs
After some thought I realized this is gonnabe on a robot which is gonna experiance a lot of hard shocks, so friction fit is a terrible idea. Instead, I'm gonna use some 6mm m2 standoffs and a decently long screw.
Moving daughterboard interconnect
In order to make space for the standoff, I need to move the interconnect. However, the flash module is in the way so I need to move that, which causes some routing issues because it's very close to the usb data lines. However, I eventually managed to make it work, but I did have to change the clearance between tracks to .15 from .2, which is fine because JLCPCB's tolerance is .1
6/5/25, 3 hours
CAD
I remember why I hate doing CAD, it's so slow and tedious. However, I did eventually get the case completed, for both the versions with and without the intergrated LED panel. I'm planning on getting the clear part on the top 3D printed out of 8001 resin from JLC3DP.
6/6/25, 2 hours
Cost optimizing
Since tariffs happened, saving $5 dollars with JLC assembly actually saves $8 after tariffs. Also, shipping is really expensive (around $38), and this is mainly due to weight. One thing you can do the save weight (and money) is make your pcb as small as humanly possible by putting the footprints right up each other. Additionally, if you can keep it under 100x100mm the PCB is $2. Another cool trick you can do if you have 2 boards is use mousebites and connect them togther, which means you only have to pay the PCBA setup fee once (if you're using PCBA). Using that $9 coupon saved me around $13.
6/19/25, 1 hour?
......
So I got my PCB! Then it wouldn't work with usb. Then I checked my schematic.
so i thought about this for a bit, and eventually decided to make a crossover
usb extension cable. (sorrt for the bad image quality, it was taken on a 5 year old laptop webcam)