SPI CS glitch

[accacca]

I am working with SPI interface in master mode
This is my peripheral configuration

Code: Select all

#define PIN_NUM_MISO 19
#define PIN_NUM_MOSI 23
#define PIN_NUM_CLK  18
#define PIN_NUM_CS   5

  spi_bus_config_t buscfg={
      .miso_io_num=PIN_NUM_MISO,
      .mosi_io_num=PIN_NUM_MOSI,
      .sclk_io_num=PIN_NUM_CLK,
      .quadwp_io_num=-1,
      .quadhd_io_num=-1
  };

  spi_device_interface_config_t devcfg={
    .clock_speed_hz = 10*100*1000,           //Clock out at 1 MHz
    .mode           = 0,                     //SPI mode 0
    .spics_io_num   = PIN_NUM_CS,            //CS pin
    .queue_size     = 3,                     //We want to be able to queue 7 transactions at a time
    .pre_cb         = 0,
  };
  
    //Initialize the SPI bus
  ret=spi_bus_initialize(VSPI_HOST, &buscfg, 0);
  assert(ret==ESP_OK);

  ret=spi_bus_add_device(VSPI_HOST, &devcfg, &mcp23x17.spi);
  assert(ret==ESP_OK);
  

My application transmit only four bytes periodically to slave device (device_addr+reg_addr+reg_data+reg_data)

Code: Select all

  // wirte out
  memset (&t, 0, sizeof(t)) ;
  t.tx_data[0] = (MCP23x17_ADDRESS << 1) + 0x00 ;
  t.tx_data[1] = MCP23x17_OLATA ;
  t.tx_data[2] = outA ;
  t.tx_data[3] = outB ;
  t.flags      = SPI_TRANS_USE_TXDATA ;
  t.length     = 8*(1+1+1+1) ;

  ret = spi_device_transmit(mcp23x17.spi, &t);  //Transmit!
  assert(ret==ESP_OK);            //Should have had no issues.

I check my code with EPS32 devboard without any device connected to SPI pins I have connect only the analyzer probes for checking the message format and sometimes I see a little glitch in a random postion in the SPI_CS pin (I attach two images) The glitch is 20ns wide. When the SPI slave peripheral detect a CS change state reset the communication in progress and tx message fail. I dont know if there are more wide glitch on SPI_CS but I am worried.

My salve peripheral (disconnected at the moment) is an I/O expander MCP23S17 and datasheet say min CS hold time 50ns I think the 20ns glitch can't be detect but if sometime the glitch is more wide...

[tomekk4]

Hi accacca,

Have you ever managed to solve this problem?
I've experienced the same issue with 20ns glitches on CS pin.
It happened after esp-idf upgrade from v3.2.2 to 4.1.

Regards,

[ESP_Sprite]

20ns sounds pretty small... are you sure the glitch is generated by the ESP32, and isn't something that is inducted to that line? Putting a scope instead of a LA on there may tell.