Feedback on MRR ESPA

[vivian-ng]

Opening a new issue thread to track feedback for the MRR ESPA.

MRR ESPA is the rename of MRR ESP 3DP. The files for MRR ESP 3DP have been migrated to KiCad for v0.8 and those files will be released on GitHub soon. Meanwhile, in preparation for a proper v1.0 release, I have renamed the board to try and give it some identity. 😄

MRR ESPA v0.9 will be up on GitHub as soon as I can sort it out. File will be on KiCad so that everyone and modify them easily. All symbols and footprints are either from the generic KiCad libraries, or inside a custom library included with the project. You may need to adjust the paths a bit on your system in order to properly find those files. They are in the subdirectory projlib with the footprints in projfp.pretty and symbols in projsym.lib.

The biggest difference between MRR ESPA v0.9 and MRR ESP 3DP v0.8 would be the separation of the VBED rail. It has been isolated from the VIN rail so that they no longer share a common ground line. This means two different PSUs can be used to power VIN and VBED. This is achieved using an optocoupler to drive the bed MOSFET's gate. A voltage divider to half the VBED voltage is the voltage used to drive the MOSFET. I may want to change this to something closer to one-third; but then, it is easily changed by the end user to whatever ratio they want: just solder the corresponding values of resistors instead of the one given in the schematic.

Will upload the files once I can.

[vivian-ng]

Migrating to KiCad ended up in quite a few mistakes. C11 and C12 of v0.8 were connected wrongly, resulting in 5V not going to the 3.3V LDO, and 3.3V not going to the CH340C. These can be fixed by soldering through-hole capacitors in place of these SMD capacitors. Also, the collector and emitter of the S9014 for Q5 needs to be swapped. Again, this can be done by using a through-hole NPN transistor.

The 5V autoselect circuit also didn't seem to want to work. So I just soldered diodes to prevent backflow.

After making these changes (capacitors C11 & C12, Q5, and the diodes to VUSB and 5V1), the board works! However, because IO2 is pulled high, the board cannot be flashed solely with USB. I needed to connect IO2 to GND (via the header pins), reset the board to enter flash mode. Once that is done, the board can be flashed. The next step is to test the motors and MOSFETs. Hopefully, I don't burn anything.

[vivian-ng]

I am having some issues trying to compile Marlin 2.0 with TMC2130 for X and Y axes.

In file included from /tmp/arduino_build_501088/sketch/src/module/../inc/MarlinConfig.h:45:0, from /tmp/arduino_build_501088/sketch/src/module/configuration_store.h:26, from /tmp/arduino_build_501088/sketch/src/module/configuration_store.cpp:47: /tmp/arduino_build_501088/sketch/src/module/configuration_store.cpp: In function 'void say_M906()': /tmp/arduino_build_501088/sketch/src/module/configuration_store.cpp:1926:52: error: 'port' was not declared in this scope void say_M906(PORTARG_SOLO) { SERIAL_ECHOPGM_P(port, " M906"); } ^ /tmp/arduino_build_501088/sketch/src/module/../inc/../core/serial.h:81:69: note: in definition of macro 'SERIAL_PROTOCOLPGM_P' #define SERIAL_PROTOCOLPGM_P(p,x) (serialprintPGM_P(p,PSTR(x))) ^ /tmp/arduino_build_501088/sketch/src/module/configuration_store.cpp:1926:35: note: in expansion of macro 'SERIAL_ECHOPGM_P' void say_M906(PORTARG_SOLO) { SERIAL_ECHOPGM_P(port, " M906"); }

I am not really sure how to fix this on Marlin, don't even know if Marlin 2.0's TMC2130 support can even run on the ESP32, so I will leave the hardware option available for the time being while someone eventually works out a software solution to it. Meanwhile, the TMC2130 support on this board will continue to be listed as "Untested" so that people know it is something to be implemented software-wise.

[vivian-ng]

Update on v0.9.

Soldered everything, and compiled @luc-github Marlin fork which has his ESP3D interface. Prelim tests show that the compile went okay, I could drive the motors to move using the webUI. I attached a heating block with thermistor to simulate the heat bed, and it was able to heat up too, so at least the optocoupler seems to be working. E0 also heated up.

Only thing that seems to be the problem is AUX1. I can't seem to get my MKS TFT32 to work with the board via AUX1. And in the process, I think I fried my RX and TX pins on the ESP32... so while I can see serial messages on the webUI, I can no longer connect to the board with a wired connection (USB or USB<->Serial module) to see serial messages.

Next step is to find time to wire v0.9 to an actual printer to try some test prints, as well as a stress test (dry-print something for 24 hours or more) if the test prints come out okay. Once everything is working, then I will go back to troubleshooting the MKS TFT32 connection (which probably means I need to solder a new board; good thing I ordered 5 pieces).

Meanwhile, v0.9 files will be released soon. It will be accompanied by a sorting of the directory structure, so let me think through the new directory structure first.

[vivian-ng]

Update on v0.9.

The files have been uploaded to the repo under the current directory.

Theoretically, v0.9 works as long as you are using external MOSFETs for the bed and hotend. During my testing, I discovered that the traces for the bed and hotend MOSFETs were too thin and also the via connecting top and bottom traces were too small. This causes heat buildup, melts the PCB, and gives off smoke. Not a great thing to do. But with external MOSFETs, the current is taken off the board, so the design works as it is.
BTW, if anyone wants to try out v0.9 using external MOSFETs, you can do away with the bed and hotend MOSFETs onboard to save cost. Just use jumper wires to connect IO2 (hotend), IO4 (bed), and GND (can take from AUX1) to their respective external MOSFETs. Of course, you can leave them in and still use them to drive the external MOSFET.

Also tested motor movement using A4988 drivers, which worked fine. I had issues compiling the webui-enabled Marlin fork by @luc-github using TMC2130 drivers, as mentioned above, but when I used the latest bugsfix-2.0.x from the main Marlin repo, I could compile TMC2130 support with no issue using the TMCStepper library. So theoretically, TMC2130 in SPI mode should work too, although I will need to find time to test this (eventually...)

Also theoretically, TMC2130 and TMC2208 drivers should work in standalone mode if the jumpers underneath the stepstick are removed (MS1 and MS2 must both be tied to VIO to enable 1/16 microstepping). I just ordered two each and will test them when they arrive in a few weeks time.

Way ahead:
I am working on the MOSFET trace issue, and will work on the TMC2130/TMC2208 jumpers to allow them to be used for standalone mode (same 3-pin jumpers, with connections to GND and VIO). The next version of this board will not have the current "use jumpers to configure for TMC2130 SPI" feature. SPI pins are already exposed, so future versions of this board will support standalone mode via jumpers, and SPI mode with the "top-facing pins" method used in most other boards.

The next version of this board should be the "production" version. Once it is done and tested to be respectably working fine, I am going to move to add the I2S support. But will need to check with @simon-jouet on what his I2S setup is like.

[emre-koc]

Hi @vivian-ng, any updates on simon-jouet's I2S support. Would like to make a board with ESP32 but I guess simon was not able to look further into this. I would like to contribute if I can.

[simon-jouet]

Hey @daboi @vivian-ng,

Yeah I've been far too busy with work, to prototype the ESP32 i2s board, I've got the basic schematic of it done quite some time ago, I just need time to tweak it and layout the PCB (which is very very time consuming).

I2S does work very well for printing, I've used it with the existing R1 prototype board and some breadboard for some time and didn't run into any issues.

[emre-koc]

Hey @simon-jouet! Wasn't expecting fast reply, thanks! Great approach on esp32 btw. I read almost all of your posts on github about this. Do you have a repo for the latest pcb so that I may have contribute to that?

[simon-jouet]

@daboi it's a bit hit and miss depends if I have to reply or not when I get the email from github :D. Thanks, glad to hear that, I'm still convinced that the ESP32 is the best way to get a very good and cheap controller. The updated PCB isn't on github yet considering it still needs work, but I can try to push it to a new branch in the weekend so you (and anybody else interested) can have a look/contribute

[emre-koc]

I subscribed almost all topics about this so will keep me up-to-date :) Happy to hear that. Which repo will you be using for branch?

[vivian-ng]

@simon-jouet I just managed to finish soldering the latest version of my own board, which hopefully will solve my burning (literally) issues. Trying to find time to test out the board over the next week. Testing with 12V doesn't really help since it is double the current to get the same power compared to using 24V. Also, I managed to figure out the pins for the I2S from the code:
22 DATA_OUT
25 WS_OUT
26 BCK_OUT

Once the current basic board is tested working, I am going to add two 74HC595s to the board (already got the empty space on the board to place those two ICs; other than capacitors on their VCCs, is there anything else that they will need?) to drive the existing 4 stepper drivers. This should free up enough pins for me to experiment with using those pins for UART/SPI_CS with TMC drivers.

[simon-jouet]

I subscribed almost all topics about this so will keep me up-to-date :) Happy to hear that. Which repo will you be using for branch?

my ESPController repo most likely, I can post here once it's done (just need to remember to do it)

@vivian-ng well done hopefully your testing goes well. Yeah the pins for I2S don't really matter too much as they can be remapped so in the current version I used pins that don't have too many functions (but I should double check because it's a bit hard to remember everything). IIRC there is no need for anything else no, what might be good would be to select whether the 74HC595 are powered by 3.3V or 5V so we can get the speed benefits of 5v if needed but 3.3V for instance if it's used as a reference for a signal back in the ESP32 as it's not 5v tolerant.

[emre-koc]

@simon-jouet can we upload gcode file to esp32 from web interface? I saw esp3D repo has that webui but can it work with your implementation? also is it possible to plug an sd card port ? I guess you stream gcode from usb port right? Btw thanks for all info, I am hoping to be a tester of your work =)

[vivian-ng]

@daboi I am using the ESP3D-enabled Marlin fork that @luc-github is working on. It works on my test board (ESPA v0.9). v0.9 can work if external MOSFETs are used to take the current off the board; v1.0 will have larger tracks and hopefully eliminate the need for external MOSFETs.

[emre-koc]

@vivian-ng thanks for the info. That fork has common code with @simon-jouet's one? I am trying to understand which repo contains which part so sorry for asking in detail. Will you branch from @luc-github's work and add I2S option? I can definitely try your V1.0 board design.