alnwlsn's Voron 2.4 3D printer notes and mishaps

2024-04-12 | [3d-printing]

Getting CAN bus toolhead to work at 1000000

I replaced my toolhead PCB with a CAN bus version some time ago (this one, the EBB SB2240_2209 CAN), but it always had some problem with it because despite having a built in accelerometer, it will fail while doing TEST_RESONANCES with the error "Timer too close". My best guess is that this is what happens when too much data is trying to be sent down the can bus, so I need to up the baud rate so it doesn't pile up.

• manufacturer seems to recommend 1000000 baud, and I think I was at 500000 before. 3 places need to be updated:
  • the toolhead PCB itself, with new klipper firmware
  • the base BTT Octopus (I was using it as the CAN interface to the Pi, took a long time to figure out this needed to be updated too), with new klipper firmware
    • if you don't update this, then when you scan for CAN devices, you'll only be able to see ones at the lower baud rate. I kept flashing the toolhead PCB and was really confused why I could only see it when I set the firmware to 500000.
  • network interface for CAN configuration in raspberry pi
  • also the configuration within the printer - printer.cfg, etc. but you probably knew that already.

firmware update (EBB):

1. Refer to klipper's official installation to download klipper source code to host
2. Building the micro-controller with the configuration shown below.
   • [*] Enable extra low-level configuration options
   • Micro-controller Architecture = STMicroelectronics STM32
   • Processor model = STM32G0B1
   • IF USE CanBoot
     • Bootloader offset = 8KiB bootloader
   • ELSE
     • Bootloader offset = No bootloader
   • Clock Reference = 8 MHz crystal
   • IF USE USB
     • Communication interface = USB (on PA11/PA12)
   • ElSE IF USE CAN bus
     • Communication interface = CAN bus (on PB0/PB1)
   • (1000000) CAN bus speed

BTT octopus firmware update

1. similar process, but harder to search for for some reason. Do make menuconfig again
   • [*] Enable extra low-level configuration options
   • Processor model (STM32F446)
   • Bootloader offset (No bootloader)
   • Clock Reference (12 MHz crystal)
   • Communication interface (USB to CAN bus bridge (USB on PA11/PA12))
   • CAN bus interface (CAN bus (on PD0/PD1))
   • (1000000) CAN bus speed

• compiled formware goes to ~/klipper/out/klipper.bin
• I didn't use a bootloader. Most tutorials will have an extra step to install the CanBoot bootloader, which will allow you to update the firmware in the future over the can bus (ie, without taking anything apart), but when that happens I'll probably have forgotten what to do and will have gone back to uploading the firmware in DFU mode over USB instead
• put device in DFU mode
  • BTT octopus has a jumper you set, then press the reset button
  • toolhead has a button you hold down, then connect the USB cable (after unplugging the CAN/power plug)
• make flash FLASH_DEVICE=0483:df11 will upload the firmware.bin onto the device from the raspberry pi itself. The FLASH_DEVICE is the USB Id from lsusb
• you can also just copy the .bin elsewhere and install dfu-util, then run the command make flash runs, which is sudo dfu-util -d ,0483:df11 -R -a 0 -s 0x8000000:leave -D out/klipper.bin

contents of /etc/network/interfaces.d/can0:

allow-hotplug can0
iface can0 can static
    bitrate 1000000
    up ifconfig $IFACE txqueuelen 1024

• txqueuelen needs to be much higher than the 128 I saw everywhere, because if you set it to that you get "No buffer space available" when trying ~/klippy-env/bin/python ~/klipper/scripts/canbus_query.py can0 scan command.

other pitfalls

• the BTT octopus board does have the 120 ohm CAN terminating resistor (well, 2x 59 ohm).
  • when trying to measure the resistor to see if it was there, I forgot that the CAN lines put out voltage, which messed with my voltmeter and made it say 6.6 Meg, suggesting no resistors. Took a while to figure that out, but measuring with power turned off yields the correct 120 ohm measurement.

For what it's worth, after the 1000000 upgrade, the accelerometer does work properly.

Moving to USB (January 2025)

Late last year, I tried to do some upgrades on the printer. I was having some issues getting the first layer correct, so I upgraded to a Voron TAP kit. In doing so, I got another SD card, to start over with a fresh Klipper installation on the Raspberry Pi, because I could not figure out how to remove the auto z calibration plugin.

The new Klipper install wanted updated firmware on the MCUs, so I did that, and then I started to experience some problems related to the CAN bus. I could not trace it effectively, but I think the problem has to do with the BTT octopus working as a CAN bridge. I am more inclined to believe it is software/firmware related than hardware.

• You can use another CAN bridge. Here's what Klipper has to say about them
  • You can use some other USB-CAN device. I think one is sold with the EBB board, but I did not get it.
    • also seems redundant to go USB -> CAN when I know the EBB can use USB.
  • I have a bunch of MCP2521 based cheap CAN modules. Klipper does not recommend them as it is more demanding than the hardware can take.

So maybe it is better to use USB instead.

First, I did the previous steps with make menuconfig and make flash but changed the settings to use USB instead of the CAN interfaces (not serial, I think this puts the interface on the mcu's UART instead of a USB Serial port).

Then, in the printer.cfg file, swap over the interface for the MCUs to use a serial port instead of CAN:

[mcu]
##  Obtain definition by "ls -l /dev/serial/by-id/" then unplug to verify
##--------------------------------------------------------------------
serial: /dev/serial/by-id/usb-Klipper_stm32f446xx_320034000550314D35323820-if00
# canbus_uuid: 32813f9e77c0
#restart_method: command
##--------------------------------------------------------------------

[mcu can0]
serial: /dev/serial/by-id/usb-Klipper_stm32g0b1xx_1F0029001650425539393020-if00
# canbus_uuid: e291d48ebcaa

Since I didn't feel like running a USB cable through a wiring harness to the toolhead, I thought I would try to run the USB data lines through the two wires that used to carry the CAN bus.

1. Go look at the schematics and board views for the EBB 2209
2. desolder L5 with a hot air gun. Lift off the component and store it for safekeeping. I put mine in the old Accelerometer Compartment. Anyways, this will isolate the CAN bus connector. Be sure to also remove the 120R terminator jumper.
3. The exposed bottom 2 pads from L5 connect directly to the CAN connector. Use 2 pieces of jumper wire to connect them to R44 and R40, which go to the USB connector. Connect on the side of the resistors closest to the edge of the board.
4. On the other end of the cable, disconnect the 2 wires from the CAN bus and solder them to the green and white wires from an old USB cable. You can trace thw wires if you want to, but it's easier to guess and try both ways to see which work.

So now, I have a tether with GND, 24V, and USB D+ and D-. Seems to be good so far. After 1 day and printing a fe hours, it seems to work OK.

Nozzle Camera

I got myself one of those 3DO Nozzle Camera V2s. Most of the solutions of this seem to print a new cover for the stealthburner and put the camera right in the middle (exactly where you normally look to inspect the nozzle). They also normally put the USB board behind the part cooling fan, but I have the breakout from the EBB kit there.

A better location seems to be as hung on the ADXL accelerometer mounting point. I used this one for the Minition Camera to get the location close for my camera.

3DO Camera V2 Stealthburner ADXL Mount

get the step file | get the FreeCAD source

I ran the cable behind the hotend cage. It's not a perfect fit, but the exposed parts of the cable do not seem to be in a position to be crushed by anything.