The post https://community.nxp.com/t5/i-MX-RT/3-line-SPI-configuration-for-send-and-receive-data-at-the-same/... has had an answer involving modification to the nxp driver. It's been 2 years and I wonder if there are updates on this matter.
I have also tried alank's answer from the post modifying my MOSI pin which is currently connected to the output device's SDIO to be kLPSPI_SdoInSdoOut but with no success.
The 1170 driver seems to have some support for half-duplex mode but it only goes as far as setting 0b1 to the config register
Hi @DCheung,
For a proper support of half-duplex mode of the LPSPI module, is by setting the correct pin configuration described on Section 70.3.1.4 Pin configuration:
Set PINCFG to either '01' or '10':
Set the RXMSK and TXMSK:
And set OUTCFG = '1':
These are the changes that the post you reference is trying to exemplify. Please make sure you are properly implementing these changes in order for 3 wire operation to work adequately.
BR,
Edwin
Hi @DCheung,
For a proper support of half-duplex mode of the LPSPI module, is by setting the correct pin configuration described on Section 70.3.1.4 Pin configuration:
Set PINCFG to either '01' or '10':
Set the RXMSK and TXMSK:
And set OUTCFG = '1':
These are the changes that the post you reference is trying to exemplify. Please make sure you are properly implementing these changes in order for 3 wire operation to work adequately.
BR,
Edwin
Hi Edwin,
Thank you for your reply. I believe what you have screen captured makes a lot of sense. I spent a bit more time to understand how the nxp driver attempts to act on half-duplex and I could see that what you have mentioned has mostly been supported by the driver itself. For example,
The driver is aware of setting the Tx/Rx masks when appropriate.
OUTCFG is also being set when it detects 3-wire mode
My code is something like below and the problem I am seeing is that the initial transmit could be clearly seen on an oscilloscope, but with the following receive, only the clock line could be seen and the MOSI line looks like it's fighting control with the master (some ripples in sync with clock near 0V). From what I can see from the nxp driver, all I have to do is to ensure PINCFG is set correctly and the rest is handled by the driver but could still not achieve the result I wanted. Please let me know if you could spot anything odd with my logic.
LPSPI_MasterGetDefaultConfig(&config);
config.pinCfg = kLPSPI_SdoInSdoOut; // This should set PINCFG to '10'
LPSPI_MasterInit(LPSPI3, &config, clock);
lpspi_transfer_t xfer;
xfer.txData = txBuffer;
xfer.rxData = NULL;
LPSPI_MasterTransferNonBlocking(LPSPI3, &handle, &xfer);
// Wait for the transfer to finish ...
xfer.txData = NULL;
xfer.rxData = rxBuffer;
LPSPI_MasterTransferNonBlocking(LPSPI3, &handle, &xfer);
