With a lot of help of @David_Gonzalez, Arduino SimpleFOClibrary now supports the ESP32 boards.
If you are interested in using the library with the ESP32 visit the supported hardware docs to see how to install the support package. It is very simple.
I am looking forward to hearing your impressions, ESP boards are really awesome and they are a great match for this library.
Let me know if you have any troubles running the code!
Antun
I used the new 1.4.1 version of SimpleFOC with my esp32 last night, it worked great!
Iāve tested it with i2c AS5600 magnetic encoder and 2804 gimbal motor using voltage, velocity and position control. Some minor speed wobbles on velocity control (even at higher speeds) but that could well be my PI tuning.
It looks well implemented. Thanks @David_Gonzalez and @Antun_Skuric. This is helping me learn loads! It would be interesting if anyone could get a single esp32 to control 4 motors. Looks like the code is setup ready to support it.
Hi, Iām trying this library on different microcontrollers, and when using esp32 Iāve found some issue when resetting and entering in a reset loop. Sometimes it lasts until I change the initial position of the motor, sometimes it is randomā¦
The setup is ESP32 DevKit, DRV8302, 5330 Brushless motor, AMT102-V.
It seems to me that NOT using the index pin makes things better.
My main question is: Are there suggested pins to use/avoid?
Should I turn off wifi and bluetooth?
Hey @dab,
If I understood correctly your ESP32 is restarting randomly? If so, my first guess is that your power supply is not being able to deliver enough power (this is sometimes typical when using wifi and bluetooth). How are you powering the DevKit?
There a lot quirks with some of the pins. This is a great guide to know which pins to use and which pins to avoid: https://randomnerdtutorials.com/esp32-pinout-reference-gpios/
Iām powering the esp32 via my Dell USB 3.1, which on the papers should give at least 1 Amp. but i didnāt measure that. But the strange thing is that when using INDEX reading, it seems the reset loop depends on the motor positionā¦
Iāve tried different pins, but if there is a āset of pinsā which work for sure, it would be nice to put it here, avoiding everyone to read all the esp32 pins behaviour.
Thanks for helping, Davide.
It should be enough powering via USB, but Iāve seen different behavior powering from an external source. When youāre not using the Index reading does it work as expected? What GPIO are you using for the index?
Iād tried to establish if this is a hardware (e.g power) issue or software issue. Not sure if you are using Arduino ide or platformio (i use the latter) but for software issues you typically see a guru meditation error message in the logs + register dump then a reboot.
Can you tell us which pins work and which do not?
Iāve found some spare time to do some testing.
Summary: ESP32 + DRV8302 + AMT102-V + DX2205-2300kv BLMotor.
esp32 is the common devkit, also known as wroom-32.
AMT102 is set at 512 ppr, and powered from VIN pin (5vdc)
Since this motor becomes neatly hot, Iām powering from a lab unit, set at 12V - max 5A.
// DRV8302 pins connections
#define INH_A 25
#define INH_B 26
#define INH_C 27
#define EN_GATE 14
#define M_PWM 18
#define M_OC 19
#define OC_ADJ 21
// AMT102
#define EN_A_pin 23
#define EN_B_pin 22
#define EN_X_pin 5
// motor instance
BLDCMotor motor = BLDCMotor(INH_A, INH_B, INH_C, 7, EN_GATE);
// encoder instance
Encoder encoder = Encoder(EN_A_pin, EN_B_pin, 512, EN_X_pin);
// Settings:
motor.voltage_power_supply = 12;
motor.voltage_sensor_align = 0.5;
motor.velocity_index_search = 3;
motor.PID_velocity.P = 0.01;
motor.PID_velocity.I = 1.0;
motor.voltage_limit = 2.0;
motor.PID_velocity.output_ramp = 50;
motor.P_angle.P = 1.0;
motor.velocity_limit = 50;
RESULT:
With these pin and settings there is a good calibration and in 10 tries no reset loop (guru meditation).
but sometimes, in position mode, after being still in a position for 3-5 seconds, it suddenly try to go somewhere at exagerate speed, asking for exagerate Current (my lab unit clamp it at 5Aā¦) and I can only reset the board to stop itā¦
Iāve done a Video which reproduce the behaviour.
After calibration i set new position to 62, then to 0, then to 62 and 0 again, then there is an error on the serial, but it doesnāt seem a guru meditation, and the motor flyies away.
Iām looking to how post the videoā¦
EDIT:
hereās the link:
Copy and paste from the terminal (Iām programming via Arduinoā¦)
MOT: Monitor enabled!
MOT: Init pins.
MOT: PWM config.
MOT: Enable.
MOT: Align sensor.
MOT: Absolute zero align.
MOT: Searching...
MOT: Success!
MOT: Motor ready.
Motor ready.
Set the target angle using serial terminal:
Target : 62.00
Target : 0.00
Target : 62.00
Target : 0.00
ets Jun 8 2016 00:22:57
rst:0x1 (POWERON_RESET),boot:0x12 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0xee
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:1
load:0x3fff0018,len:4
load:0x3fff001c,len:1216
ho 0 tail 12 room 4
load:0x40078000,len:10864
load:0x40080400,len:6432
entry 0x400806b8
MOT: Monitor enabled!
MOT: Init pins.
MOT: PWM config.
MOT: Enable.
MOT: Align sensor.
MOT: Absolute zero align.
MOT: Success!
MOT: Motor ready.
Motor ready.
Set the target angle using serial terminal:
Iām starting to think that it could be just some PID setting problemā¦
now I lowered the I value to 0.1 (from 1.0) while K=20.0. Movement is precise, and it seems the motor doesnāt run away anymoreā¦
so:
P=0.01
I=0.1
K=20.0
R=300
L=50
F=0.01 (I really donāt understand what this changesā¦)
So now Iām going to try to implement my own up/down acceleration ramps although I think that should be a VERY interesting addon for this libraryā¦
Do you have any suggestion on my numbers, or on the esp32 pins I used?
Thanks again, Davide.
2300KV motors do not like to run slowly! Theyāll pull a lot of current. Iām using 100-200KV gimbals which have a resistance of 10ohms. I reckon your motors resistance will be about 0.5ohms - hence the high current. Itās not a problem if your power supply and board can handle it.
By the way - Iād recommend tuning in velocity mode and getting that sweet before moving on to angle mode as the angle PID relies on a well tuned velocity PID and itās hard to do both at the same time.
L=50
F=0.01 (I really donāt understand what this changesā¦)
What is L? - that isnāt a config value Iām aware of? F is probably Tf - which is the amount of smoothing done on velocity. Youāve only got 512cpr which is quite low so if you ask for instantaneous speed itāll come out like this 839, 0, 0, 0, 842, 0, 0, 0, 0, 845 as most āloopsā in program happen without your encoder ticking over. With this smoothing the speed will look more like 242, 220, 200, 180, 246, 220, 200, 180, 243. The PID loop finds it easier to deal with this smoothed velocity but it is adding lag so you donāt want it to high.
Thanks for your suggestions.
Those letters are for changing values on the fly. Taken from an example:
List of commands:
- P: velocity PID controller P gain
- I: velocity PID controller I gain
- D: velocity PID controller D gain
- R: velocity PID controller voltage ramp
- F: velocity Low pass filter time constant
- K: angle P controller P gain
- N: angle P controller velocity limit
- L: system voltage limit
- C: control loop
- 0: voltage
- 1: velocity
- 2: angle
- V: get motor variables
- 0: currently set voltage
- 1: current velocity
- 2: current angle
- 3: current target valueAll your pins seem fine. GPIO 5 will output a PWM signal on boot, but this shouldnāt cause you any problems.
Your L (voltage limit) is too high. Specially for motors with a really small winding resistance you want to set this to nice limit as to not draw to much current. I set a value that I know my motor will be able to handle and not get too hot.
You can modify the acceleration ramp by changing R, itās in termos of voltage over seconds. You can play changing this value and see the different behaviors.
Ops, my fault, Iāve confused L with Nā¦
Iāve set L (voltage limit) to 1.0 (with 12V input) to make all the tests and N=50 which is the āangle velocity limitā.
In Velocity mode, my P and I settings seem to work ok, sudden speed changes are done as expected. the only bed thing happening is sometime, without any special reason, the motor run away, asking for exagerate current, even with the voltage limit set at 1.0
This happens with the INDEX. if I donāt use the index everything is okā¦
Hey @dab
What you are describing now might be really a software bug.
Does this happen only when you set the position to 0?
Iāll have to investigate this behavior but it might be that the index handling routine is the one which is responsible for this.
I had tested it extensively but I did not use a motor which is as fast as yours and maybe the vibrations around the 0 (where index is located) are causing it to skip count. I am sure but this might be it.
To verify this:
Could you please include motor.monitor() in your arduino loop() and show us the serial output when this happens.
What should happen is a sudden jump in motor position.
I understand your intuition. Iām trying that right now, and after some movements in which the error doesnāt happen while going back and forth to zero (I can see the first number, which I guess is the current position, slowly oscillating from -0.0xx to +0.0xx), i started it again and made a video, in which at the first movement after calibration, giving 5 as the destination, the motor suddenly run away asking more than 5 Amps from the power supply.
You can see it here, and you can hear the motor when it runs away.
and this is my setup:
Then, another strange thing happens with this setup:
depending on the motor position when starting to calibrate, the ESP32 goes into a reset infinite loop like this:
MOT: Monitor enabled!
MOT: Init pins.
MOT: PWM config.
MOT: Enable.
MOT: Align sensor.
Guru Meditation Error: Core 1 panicāed (Coprocessor exception)
Core 1 register dump:
PC : 0x400d27e9 PS : 0x00060031 A0 : 0x800d10cd A1 : 0x3ffbeb50
A2 : 0x3ffc1444 A3 : 0x00000001 A4 : 0xfffffffb A5 : 0x00000000
A6 : 0x00000008 A7 : 0x00000001 A8 : 0xfffffffd A9 : 0x3ffbebf0
A10 : 0xfffffffb A11 : 0x00000000 A12 : 0x00000000 A13 : 0x3ffc1de8
A14 : 0x00000003 A15 : 0x00000000 SAR : 0x00000005 EXCCAUSE: 0x00000004
EXCVADDR: 0x00000000 LBEG : 0x00000000 LEND : 0x00000000 LCOUNT : 0x00000000
Core 1 was running in ISR context:
EPC1 : 0x400d27e9 EPC2 : 0x00000000 EPC3 : 0x00000000 EPC4 : 0x40083835
ELF file SHA256: 0000000000000000000000000000000000000000000000000000000000000000
Backtrace: 0x400d27e9:0x3ffbeb50 0x400d10ca:0x3ffbec40 0x40080f79:0x3ffbec60 0x40081715:0x3ffbec80 0x401059af:0x3ffbc5a0 0x400d87c7:0x3ffbc5c0 0x400915f5:0x3ffbc5e0 0x40090079:0x3ffbc600
Rebooting...
ets Jun 8 2016 00:22:57
rst:0xc (SW_CPU_RESET),boot:0x12 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0xee
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:1
load:0x3fff0018,len:4
load:0x3fff001c,len:1216
ho 0 tail 12 room 4
load:0x40078000,len:10864
load:0x40080400,len:6432
entry 0x400806b8
MOT: Monitor enabled!
MOT: Init pins.
MOT: PWM config.
MOT: Enable.
MOT: Align sensor.
Guru Meditation Error: Core 1 panic'ed (Coprocessor exception)
Core 1 register dump:
PC : 0x400d27e9 PS : 0x00060031 A0 : 0x800d10cd A1 : 0x3ffbeb50
A2 : 0x3ffc1444 A3 : 0x00000001 A4 : 0xfffffffb A5 : 0x00000000
A6 : 0x00000008 A7 : 0x00000001 A8 : 0xfffffffd A9 : 0x3ffbebf0
A10 : 0xfffffffb A11 : 0x00000000 A12 : 0x00000000 A13 : 0x3ffc1de8
A14 : 0x00000003 A15 : 0x00000000 SAR : 0x00000005 EXCCAUSE: 0x00000004
EXCVADDR: 0x00000000 LBEG : 0x00000000 LEND : 0x00000000 LCOUNT : 0x00000000
Core 1 was running in ISR context:
EPC1 : 0x400d27e9 EPC2 : 0x00000000 EPC3 : 0x00000000 EPC4 : 0x40083835
ELF file SHA256: 0000000000000000000000000000000000000000000000000000000000000000
Backtrace: 0x400d27e9:0x3ffbeb50 0x400d10ca:0x3ffbec40 0x40080f79:0x3ffbec60 0x40081715:0x3ffbec80 0x401059af:0x3ffbc5a0 0x400d87c7:0x3ffbc5c0 0x400915f5:0x3ffbc5e0 0x40090079:0x3ffbc600
Rebooting...
ets Jun 8 2016 00:22:57
rst:0xc (SW_CPU_RESET),boot:0x12 (SPI_FAST_FLASH_BOOT)
configsip: 0, SPIWP:0xee
clk_drv:0x00,q_drv:0x00,d_drv:0x00,cs0_drv:0x00,hd_drv:0x00,wp_drv:0x00
mode:DIO, clock div:1
load:0x3fff0018,len:4
load:0x3fff001c,len:1216
ho 0 tail 12 room 4
load:0x40078000,len:10864
load:0x40080400,len:6432
entry 0x400806b8
MOT: Monitor enabled!
MOT: Init pins.
MOT: PWM config.
MOT: Enable.
MOT: Align sensor.
until i slightly turn the motor by hand changing the starting position, then the calibration goes as expectedā¦
hey @dab, from what I can see on the video it seems like a problem of the index handling
Here are the docs pages for monitoring and motor commands.
So from what I can see from the video it seems like an index handling problem. But it is very hard to see on the video because it happens so fast.
Could you please dump the contents of your serial monitor to a file and post it here somehow. That would be very helpful!
You could remove motor.monitor(); and put something like this in the loop.
Serial.print(motor.voltage_q);
Serial.print("\t");
Serial.print(motor.target);
Serial.print("\t");
Serial.println(motor.shaft_angle);
Serial.print("\t");
Serial.println(motor.shaft_velocity);
It would be interesting to see the velocity and angle measurement.
The coprocessor exception, from googling seems to be caused by trying to use the coprocessor (fpu) in an interrupt routine.
You canāt do float operations but can do doubles (as that is done in software)
I reckon the cause is this line in encoder.cpp
pulse_counter = round((float)pulse_counter/(float)cpr)*cpr;
Try a double cast instead of float.
Esp32 also gets upset if interrupt takes too long.
Tomorrow I should have more time to test. Thanks.