There seems to be some problems in deadtime config of mcpwm (IDFGH-5600)

[Ncerzzk]

Environment

• Development Kit: ESP32-PICO-Kit
• Kit version v1
• Module or chip used: ESP32-PICO-D4
• IDF version (run git describe --tags to find it):
  4.3
• Build System: idf.py
• Compiler version (run xtensa-esp32-elf-gcc --version to find it):
  // 1.22.0-80-g6c4433a
• Operating System: Windows
• (Windows only) environment type: [MSYS2 mingw32|ESP Command Prompt|Plain Command Prompt|PowerShell].
• Using an IDE?: No
• Power Supply: [USB|external 5V|external 3.3V|Battery]

Problem Description

I want to add deadtime for my H bridge driver. I produce 4 pwm signals because my H bridge is juse very simple(there is not a nmos gate driver, and just 4 sot23 Nmos(ao3400)).
//Detailed problem description goes here.

Expected Behavior

what I want is that there are proper deadtime between the AH and AL signals, if possible , just using 2 singals is better( like AL is the reversed phase of AH).

my problem is just like the guy's:
https://hackaday.io/project/169905-esp-32-bldc-robot-actuator-controller/log/184464-mcpwm-with-deadband

Actual Behavior

I set the deadtime mode to MCPWM_ACTIVE_HIGH_COMPLIMENT_MODE, which said "MCPWMXA = rising edge delay, MCPWMXB = compliment of falling edge delay" in comment in code.

and set pwma(AH) = 10, pwmb(AL) = 100 (just close the low bridge) , there seems to produce short circuit.
and I measured the voltage(I have no scope) of AH pin and AL pin, the AH is 0.33v, and AL is 3.3V, which is not I want.
what I want is AH is 0.33V and AL is 0V.

Steps to reproduce

1. step1
2. ...

// If possible, attach a picture of your setup/wiring here.

Code to reproduce this issue

void  pwm_init(){
    
    // AH 4     BH 38
    // AL 2     BL 37
    mcpwm_gpio_init(MCPWM_UNIT_0, MCPWM0A, 4);
    mcpwm_gpio_init(MCPWM_UNIT_0,MCPWM0B,2);
    mcpwm_gpio_init(MCPWM_UNIT_0, MCPWM1A, 38);
    mcpwm_gpio_init(MCPWM_UNIT_0,MCPWM1B,37);    

    mcpwm_config_t pwm_config;
    pwm_config.frequency = 20000;    //frequency = 1000Hz
    pwm_config.cmpr_a = 0;      
    pwm_config.cmpr_b = 0;
    pwm_config.counter_mode = MCPWM_UP_COUNTER;
    pwm_config.duty_mode = MCPWM_DUTY_MODE_0;

    mcpwm_deadtime_enable(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_ACTIVE_HIGH_COMPLIMENT_MODE,10,10);
    mcpwm_deadtime_enable(MCPWM_UNIT_0,MCPWM_TIMER_1,MCPWM_ACTIVE_HIGH_COMPLIMENT_MODE,10,10);

    mcpwm_init(MCPWM_UNIT_0, MCPWM_TIMER_0, &pwm_config);   //Configure PWM0A & PWM0B with above settings
    mcpwm_init(MCPWM_UNIT_0, MCPWM_TIMER_1, &pwm_config);   //Configure PWM0A & PWM0B with above settings
}

void set_duty(float duty){
    if(duty>0){
            mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_GEN_B,100);    
            mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_GEN_A,duty);
            mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_1,MCPWM_GEN_A,0);
            mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_1,MCPWM_GEN_B,0);  
    }else{
            mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_1,MCPWM_GEN_B,100);    
            mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_1,MCPWM_GEN_A,duty);
            mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_GEN_A,0);
            mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_GEN_B,0);      
    }
}


and I called set_duty(10) in app_main()

When I search the info in the tech document, It shows that the config of MCPWM_ACTIVE_HIGH_COMPLIMENT_MODE (0b001010000 <<8 ) is just close the S3 and S5 of

It should have worked, but I don't know where the problem is in.

[SalimTerryLi]

About actual behavior

the AH is 0.33v, and AL is 3.3V is what esp32 should output because of your code:

mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_GEN_B,100);    
mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_GEN_A,duty);

that drives AL to full output and AH about 10% percentage when duty=10.0

About the link to one on hackaday.io

That is a pretty well example of Operater, Sync and Deadtime submodule. From my perspective all parts function as expected (as figure indicates).

But it is not suitable for full bridge driver.

About your code

It must be runtime error when mcpwm_deadtime_enable(). Also 20k frequency is just too high for ao3400 to properly operate.

About circuit

Usually a H-bridge driver only requires two control wires that each controls a pair of MOSFETs. If it is consist of 4 NMOS then extra circuit is required and must cannot be a "very simple" design. If possible please provide your schematic so that the problem can be addressed quickly.

[Ncerzzk]

@SalimTerryLi Thanks for your comment!

This is my circuit, I think it's enough simple.
Generally, there should be a bootstrap circuit in the high bridge, while because my load is enough small and this is just a test circuit, so I never design the bootstrap circuit.

However, what I want is that I don't want to control the PWMxA and PWMxB, I just want to control one of them, and the other just become inverted signal.

The document describe the AHC Deadtime Mode is very similar to what I want.( page 431 in esp32_technical_reference_manual_en)
When I test, the result shows that the PWMxb is not the invert of PWMxA.

And then I check the picture below,

I found that in AHC Deadtime Mode, it just set S3 and S5 =1 ,and other switch set to 0. It means that the source of PMWxB is PWMxb, not PWMxA.

Well, so it 's right the result shows that the PWMxb is not the invert of PWMxA, but there is still a problem.

According to this picture , because there is a inverter in the out route of PWMxB, so the output of PWMxB should be inverter. When I set the duty 0, it should be high, and when I set duty 100, it should low(duty mode is DUTY_MODE_0).

[SalimTerryLi]

Thanks for the explanation now I get the idea.

AHC doesn't behavior as the API described, that's true. We will find a solution.

But I cannot reproduce your condition that PWMxB not being inverted. Here's the captured waveform:

PWMxA is continuously outputting 10% duty and PWMxB outputs low when the generator actually started. There is some delay between startup and outputting so PWMxB will output high for about 2ms because of the inverter. RED behaviors correctly and FED cannot be detected due to 100% duty cycle of PWMxB

Codes below already make startup time in consideration so no high output will appear on PWMxB:

#include <freertos/FreeRTOS.h>
#include <freertos/task.h>
#include "driver/mcpwm.h"

#define H_BRIDGE_A_PIN GPIO_NUM_2
#define H_BRIDGE_B_PIN GPIO_NUM_4

void set_duty(float duty){
    if(duty>0){
        mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_GEN_B,100);
        mcpwm_set_duty(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_GEN_A,duty);
    }
}

void pwm_init(){

    mcpwm_config_t pwm_config;
    pwm_config.frequency = 1000;    //frequency = 1000Hz
    pwm_config.cmpr_a = 0;
    pwm_config.cmpr_b = 0;
    pwm_config.counter_mode = MCPWM_UP_COUNTER;
    pwm_config.duty_mode = MCPWM_DUTY_MODE_0;
    mcpwm_init(MCPWM_UNIT_0, MCPWM_TIMER_0, &pwm_config);

    mcpwm_deadtime_enable(MCPWM_UNIT_0,MCPWM_TIMER_0,MCPWM_ACTIVE_HIGH_COMPLIMENT_MODE,10,10);
    set_duty(10);

    mcpwm_stop(MCPWM_UNIT_0, MCPWM_TIMER_0);
    vTaskDelay(pdMS_TO_TICKS(10));
    mcpwm_start(MCPWM_UNIT_0, MCPWM_TIMER_0);
    mcpwm_gpio_init(MCPWM_UNIT_0, MCPWM0A, H_BRIDGE_A_PIN);
    mcpwm_gpio_init(MCPWM_UNIT_0,MCPWM0B,H_BRIDGE_B_PIN);
}

void app_main(void)
{
    pwm_init

[Ncerzzk]

@SalimTerryLi Thank you for your time to test. I would try your code later(while because my esp32 board is broken a fews days ago, I could not try it recently).

Well, since the inverter behave correctly in yours, some of what I did must be wrong, I would find them.