nRF52832 SPI with MCP2515: nrf_drv_spi_transfer seems to send properly, but not receiving in all cases.

RE: nRF52832 SPI with MCP2515: nrf_drv_spi_transfer seems to send properly, but not receiving in all cases.

BendaEng — Thu, 12 Sep 2019 15:29:48 GMT

Presently I'm using SES. But if I move away from it in the future, I'll have to look at it. Thanks!

RE: nRF52832 SPI with MCP2515: nrf_drv_spi_transfer seems to send properly, but not receiving in all cases.

Nguyen Hoan Hoang — Thu, 12 Sep 2019 02:25:48 GMT

If you are not usisng SES. This is a much better driver. Works with 100s of devices and sensors. https://github.com/IOsonata/IOsonata/blob/master/ARM/Nordic/src/spi_nrf5x.cpp

RE: nRF52832 SPI with MCP2515: nrf_drv_spi_transfer seems to send properly, but not receiving in all cases.

BendaEng — Wed, 11 Sep 2019 17:15:18 GMT

On another note I just tried your newest response with the read command being 1 byte short. That was exactly the problem for the initialization!

Much appreciated, now I can work on getting some messages flowing.

Thanks!

RE: nRF52832 SPI with MCP2515: nrf_drv_spi_transfer seems to send properly, but not receiving in all cases.

BendaEng — Wed, 11 Sep 2019 17:13:24 GMT

Hi,

Thanks for the response. I have tried modes 0 through 3 earlier with no results. I just tried again, but the initialization still fails. Interestingly enough regardless of SPI_MODE, the response I get for mcp2515_readStatus() is always 0x92.

Good point on the reset capacitor. I'll have to test that in a moment.

RE: nRF52832 SPI with MCP2515: nrf_drv_spi_transfer seems to send properly, but not receiving in all cases.

hmolesworth — Wed, 11 Sep 2019 16:57:21 GMT

SPI Modes 0,0 (Mode 0) or 1,1 (Mode 3) both work on the MCP2515, but the initialisation code above has Mode 2 (NRF_DRV_SPI_MODE_2). I would suggest first try changing this to NRF_DRV_SPI_MODE_0 or NRF_DRV_SPI_MODE_3.

The Reset capacitor seems very large at 10uF. given that only a minimum of 2uSec reset is required, although no max is quoted; unless the VDD rise time is extremely slow it might be wise to reduce that somewhat. 100nF might be a better choice. However that's unlikely to be an issue with the SPI.

The read command (if Read Instruction) seems to be 1 byte short, though it's not clear which command is in the code. The 3-byte transfer is <instruction><address><response>. If that was meant to be Read Rx Buffer, only 2 bytes <instruction><response>, then the address would not be part of the tx.

uint8_t mcp2515_readRegister(const uint8_t address)
{
    uint8_t m_tx_buf[] = {MCP_READ, address};
    uint8_t m_rx_buf[3]; // not sure what MCP_READ_STATUS is: sizeof(MCP_READ_STATUS) + 1]; 
    uint8_t m_tx_length = sizeof(m_tx_buf);
    uint8_t m_rx_length = sizeof(m_rx_buf);

    memset(m_rx_buf, 0, m_rx_length);
    spi_xfer_done = false;
    APP_ERROR_CHECK(nrf_drv_spi_transfer(&can_spi, m_tx_buf, m_tx_length, m_rx_buf, m_rx_length));

     while (!spi_xfer_done)
        {
            __WFE();
        }
    return m_rx_buf[2]; // was 1, require 3rd byte
}
STATIC_ASSERT(m_rx_length == 3);