ESP-07S on SKR 3 ES ?

[raoullector]

Hi,

I’m using an SKR 3 ES with TMC 5160 drivers and GrblHal, and everything is working perfectly.
Thanks to @terjeio and @dresco for their great work!

But now, I’d like to connect a WiFi module ESP-07S, and I can’t find a reliable guide.
Has anyone successfully set up this configuration?

[terjeio]

Not that I am aware of - if the ESP-07S uses the same command set as ESP-AT then the ESP-AT plugin can be used?

[raoullector]

Good idea, I will have a look at ESP-AT.

[dresco]

I'll add the additional serial port definition (for the ESP header) to the SKR3 board map. I have the BTT ESP32 module from the Scylla board, so can do an initial test with that..

For ESP8266 support, guess it would first be a question of finding suitable ESP8266 AT firmware. I have a few old ESP8266 modules here, nothing with the same footprint/pinout, but may do for a PoC test against the plugin..

[dresco]

fyi, have just added the 2nd serial port definition to the SKR3 board map, and successfully tested a telnet connection with my existing BTT ESP32 module.

Worth noting that I previously updated the binary for the BTT ESP32 module (using the Espessif at.py tool), to move the AT function to UART0 (TXD0/GPIO1 & RXD0/GPIO3). I originally did this for the BTT Scylla board, I should probably do a write up on this, and put a pre-built binary somewhere..

[raoullector]

Thank you both for your answers.

@dresco :
On my side, I'm now testing in Station mode, but so far, it isn't working any better. I'm still stuck on the same error when the "AT+CIPMODE=1" (passthrough mode) command is sent.

Something doesn't add up because :

• "AT+CIPMODE=1" (passthrough mode) requires "AT+CIPMUX=0" (single connection),
  but
• "AT+CIPSERVER=1,23" (start Telnet server) requires "AT+CIPMUX=1" (multiple connections).

Which version of ESP-AT are you using? Have you modified anything in your ESP-AT firmware?

@terjeio :

It likely is since the module might be held in reset otherwise.

I'm not sure that's the case, because on my side, the ESP seems to be working fine: the Wi-Fi starts up, the ESP-AT dialogue runs smoothly (up until the infamous "AT+CIPMODE=1").

[raoullector]

@terjeio, I've tried the Scylla mods but it doesnt't change anything in the ESP-AT start up process on my SKR3 : still unable to retrieve the Reset and Boot ports.
For this board, neither COPROC_RESET_PIN nor COPROC_BOOT0_PIN are defined, so COPROC_PASSTHRU equals 0.
The same for AUXOUTPUT8_PORT and AUXOUTPUT9_PORT.

[terjeio]

I am using this version:

at param mode: 1
AT cmd port:uart0 tx:17 rx:16 cts:-1 rts:-1 baudrate:115200
module_name: WROOM-32
max tx power=78, ret=0
v3.4.0.0

For this board, neither COPROC_RESET_PIN nor COPROC_BOOT0_PIN are defined, so COPROC_PASSTHRU equals 0.

Then the reset and boot ports can not enumerate. They are only required for passthru programming and will be used, if available, to reset the ESP during the plugin startup sequence.

[dresco]

Apologies, am a bit tied up with other things this week, so haven't had much time for this..

@terjeio Can confirm the Scylla passthru is working again for me with the above changes. I'll add the relevant coproc pins to the SKR3 map with the next update.

One thing I've just noticed, when I put it into passthru mode, I can talk to the ESP chip with esptool.py, but I can't (for instance) query the firmware version with AT+GMR from a terminal, I guess is already switched into bootloader mode?

@raoullector I'm using ESP32-WROOM-32-AT-V3.4.0.0 firmware, with the BTT ESP32 module that I got with a Scylla board.

I used the ESP at.py tool to customise the factory WROOM binary to use UART0 for AT commands, just due to the way the Scylla board is wired. Have carried that UART0 pinout over to the SKR3 map, and had previously tested telnet access with the same module on my SKR3 here.

I still haven't tested with an ESP8266, will do that when I get the opportunity.

[terjeio]

I guess is already switched into bootloader mode?

It is, I'll change this line to set IO0 to 1, but I think I'll leave the chip in reset mode to block incoming connections/traffic until DTR is asserted.
FYI if ESP-AT is reconfigured to use UART1 passthru mode cannot be used to program the chip.

[raoullector]

It's OK, everything works fine with an ESP32, but the esp-at plugin won't work with an ESP8266.

To summarize, the root of the problem is that until esp-at firmware version 2.3, running a Telnet server (AT+CIPSERVER) required multiple connections mode (AT+CIPMUX=1), which was then incompatible with software passthrough (AT+CIPMODE=1). Unfortunately, for the ESP8266, the last supported firmware version is precisely V2.3.

This is not the case for the ESP32 (I flashed esp-at V3.2) : the Wi-Fi connection with the SKR3 works in both Access Point and Station modes (although a bit slow in STA mode, but that's probably due to my network).

Thanks again for your work and support!

[dresco]

Hi @terjeio

Just checking, should the latest core changes support passthru for the BTT Scylla board? I've just updated, but the co-processor port is no longer enumerating for me.. (I do have the AUXOUTPUTn_PORT and AUX_CONTROLS changes already in my local branch).

[dresco]

Just another data point from some further testing;

If I comment out the new AUXOUTPUTn_PORT definitions, the ESP AT plugin works correctly (just no passthru support, and the higher aux outputs are not present in $pins - all as expected).

If I re-add the new AUXOUTPUTn_PORT definitions, the ESP AT plugin reports as disabled, and the higher aux outputs are listed as aux outputs & not CoProc pins.

[PIN:PE9,Aux out 8,P5]
[PIN:PE10,Aux out 9,P6]

Hope that gives you some insight, am finding it a bit hard to troubleshoot...

[dresco]

I've pushed my latest updates to another branch here.

I now see that the ESP AT plugin is failing because the reset pin is being pulled low (being treated as an aux out pin & not a CoCproc pin). However, am struggling to follow the ioports logic to understand why it's not being claimed properly..

[terjeio]

However, am struggling to follow the ioports logic to understand why it's not being claimed properly..

I am struggling as well, so I am refactoring it a bit more now. It is somewhat complicated to have ioports available for both the core and for claiming by plugins... I expect to have completed testing later today.

[terjeio]

I have just commited changes to the core that should fix this.

[dresco]

I have just commited changes to the core that should fix this.

Cool thanks, will test tomorrow..