RE: nRF51 SPIM transfer

zhwzh308 — Thu, 19 May 2016 15:12:39 GMT

From my experience, the buffer is filled when you call nrf_drv_spi_transfer. On the application side you can make two fixed size buffers for both tx and rx.

When you call nrf_drv_spi_transfer(&m_spi_m, tx_buffer, tx_bytes, rx_buffer, rx_bytes), specify value for tx_bytes, it is the number of bytes you want to put on the SPI master-out pin; rx_bytes refers to the number of bytes you want to fill for the rx_buffer, beginning at the start of your transaction.

For example, a register read on a SPI peripheral requires first writing the address of register you want to read, SPI peripheral will return the value of that register as soon as you finish writing the address. In this case tx_bytes is 1, rx_bytes is 2. When the call completes with call back, you will get the read from rx_buffer[1].

In general, if you want to tx x bytes, and receive y bytes after x bytes has been transferred, rx_buffer needs to be the size of (x + y). A valid read back value will start at rx_buffer[x] for the length of y.

Back to your question, I think you can make a rx_buffer large enough to contain the read back from SPI peripheral; write a call back function after the transaction is done and examine the value of rx_buffer. Write a callback in the form of typedef void (*nrf_drv_spi_handler_t)(nrf_drv_spi_evt_t const * p_event); and provide it when you call nrf_drv_spi_init(&m_spi_m, &spi_config, spi_event_handler); to deal with the information you have received.

Hope this helps!