topic Re: Serial / USB not sending bytes with MSB on in Kinetis Microcontrollers

Serial / USB not sending bytes with MSB on

robertpoor — Thu, 04 May 2017 14:35:31 GMT

I have a FRDM-KL27Z dev board communicating with my Mac OSX via the OpenSDA USB port.

It appears that if I write a char to the LPUART with the MSB turned on (i.e. 0x80 - 0xff), the serial line goes idle for one byte period instead of transmitting the character, so if I send the pattern:

putchar(0x55);

putchar(0xa5);

putchar(0x5a);

... on the 'scope (J1-04, aka TX), I see 0x55 followed 10 bit periods of 'mark' followed by 0x5a .

Any idea what's going on?

Some details:

I'm enabling LPUART0 like this:

  lpuart_config_t config;
  LPUART_GetDefaultConfig(&config);
  config.baudRate_Bps = 9600;
  config.enableTx = true;
  config.enableRx = true;
  CLOCK_SetLpuart0Clock(0x03);
  uint32_t uartClkSrcFreq = CLOCK_GetFreq(kCLOCK_McgInternalRefClk);
  LPUART_Init(LPUART0, &config, uartClkSrcFreq);

And my putchar() routine looks like this:

void putchar(char ch) {
  while (!(LPUART0->STAT & LPUART_STAT_TDRE_MASK)) {
    // busy wait until TX buffer ready
  }
  LPUART0->DATA = ch;
}

Re: Serial / USB not sending bytes with MSB on

mjbcswitzerland — Thu, 04 May 2017 15:13:24 GMT

Hi Robert

Change

void putchar(char ch)

void putchar(unsigned char ch)

and it will 'magically' start working.

Regards

Mark

P.S. I think that there was a thread about 4 weeks ago where the same mistake was made, including explanation

Re: Serial / USB not sending bytes with MSB on

robertpoor — Thu, 04 May 2017 15:40:57 GMT

Hah! I just figured it out independently. :smileyhappy:

By declaring it a char (rather than uint8_t), a byte with the MSB set gets sign extended to the full 32 bit LPUART0->DATA register. Bit 13 in the DATA register is FRETSC, or Frame Error / Transmit Special Character. And from the docs: "For writes, indicates a break or idle character is to be transmitted instead of the contents in DATA[T9:T0]. T9 is used to indicate a break character when 0 and a idle character when 1, he contents of DATA[T8:T0] should be zero." Which is exactly what I'm seeing.

So, yep, using a uint8_t will fix all that.

Thanks.