This is a big one for me. Besides the fact that it is easy to switch for testing things etc, I don’t have a computer anywhere near my CNC machine, and removing the Jackpot from the CNC is a huge difference in effort compared to removing the ESP32 from the Jackpot.
I often have 2 or 3 different ESP32s that I’ve used to switch firmware versions for test, etc,. or just to flash new firmware when I didn’t have it connected to my home wifi.
If I were buying a new Jackpot today, I’d rather pay an extra $5 to keep the flexibility.
I keep thinking, the datagram format of the TMC2209 is simple enough, one could make a UART “router” that accepts input addresses beyond 0-3 and routes to multiple UART outputs. Like this:
Then addresses 0-3 go to the upper set of devices and addresses 4-7 go to the lower set of devices and the router would overwrite the address portion of the datagram to be 0-3.
Then the client (ESP32) can just treat it as if the addresses go beyond 0-3 and the software is oblivious to the workaround. The bad part is you would have to incorporate and somehow flash the router chip.
This would be a fun little mini-project but unfortunately I think it’s not really practical.
I know, this isn’t germane to the voltage regulator… but, when the number of drivers in the chain go above 3, the serial interface becomes unreliable. Proven in the earlier threads when developing the jackpot.
In effect, the scheme on the jackpot already does this for drivers 3-6.
You could abstract this back to every single driver in the chain is behind the same mux arrangement as used on Jackpot, then one ESP-32 UART could support an arbitrary number of TMC2209s via thier UART interfaces, up to the point where the mux is too slow for responsiveness. (The UART isn’t generally used in FluidNC except when setting up or when responding to errors).
Yes, and having the mux handled in FluidNC makes an alternative pretty much pointless. Even if a hardware mod could make the code “cleaner” for the client, we could never remove the mux code.
It might be a different story for Marlin on Jackpot. I have not checked if Marlin enforces the limited addresses on TMC2209 or if it just passes it along. In the latter case, Marlin would work without extra unofficial hacks.
Yes we could. The only reason we can not use it now is that devkit has the dang pins scrambled for some reason. I have a stack of them here right now.
I am confident I could solder it, but could I remove it??
I totally understand this, I have swapped out a CNC version and a Laser version myself. But there are a lot of new boards coming out, lots of stuff happening with product pricing by the time it gets to my door.
The Jackpot as is will not be gone, I just think there is options for the less power user. For example, I stopped swapping my esps when I realized I can just change the config name in the settings and reboot to swap firmware. I also don’t think there will be any major functional differences between a integrated board and the current board. Maybe one or two less outputs.
Integrated drivers and esp32 saves me a lot of headache. I have to keep everything in stock, and assemble. This also means I can much more easily ship international direct with flashing and testing included, currently that is cost prohibitive.
I am realizing integrated drivers means the heat sinks would end up on the bottom of the board though, or maybe they go on the actual chips since the board will have a much larger ground plane / heat sink. We could also use the tmc2226 for a little larger contact area. Integrated drivers also means we can control the sense resistors, I think that means we can control the heat a bit better since we run above the midway point of the current. so maybe a 100ohm instead of 110?
Sorry, I thought you’d have to choose between integrating an ESP32 module that either includes an internal antenna but not the antenna connector or includes an antenna header but no internal antenna. I now see that the ESP32-WROVER-IE module picture has both. I had looked multiple times on my phone, but couldn’t see antenna traces until I looked on my desktop now.
Make sense, I do hope you end up with multiple JackPot/Controller board options that still includes at least one fully flexible offering, as well as a cost/assembly optimized fully embedded variant right sized for LR4, MPCNC v.next and MPEDM.
Vaguely recall JackPot’s swappable MCU connector setup potentially enabled a RP2040 with DevKitC pinout being worked on so that Klipper could be used?
Espressif seem to be churning out MCUs at faster pace than many companies can absorb reving/iterating their consuming controllers/products… Guessing pace will increase as AI’fication of all the things seems to be the current trend.
At all, shoot I thought that was just a pin move or something. DANG (its okay for that one, you had to move a tiny resistor anyway).
I will carry both as long as they sell.
We need the MUX code, that is above my paygrade, and honestly I just wanted to say it could do it. There is absolutely no reason to spend time or money on that as there are thousands of printer boards.
Well I am only keeping an eye on stepper drivers and what fluidnc and grblhal support. That makes it easier.
Just took a look, for the most part it does work we just lose pins 16-17, for us pin 17 is the most important (I2s), pin 16 is one of the high power outputs. When I flashed and tested it I was at a loss for a while until I saw the difference.
I am taking notes, I love all the exploration but let’s keep this thing moving. We can keep exploring all the other non-power supply topics as well as we keep poking at this.
