I am using the SynchroMaster processorexpert component to talk to a simple 4-wire SPI component attached to a TWR-K60F120m via the primary elevator. My settings are included at the bottom. I am using simple code to test (below): check buffers are empty, send the characters, wait for the transmit buffer to empty and the receive buffer to fill up, read out all the characters.
My problem seems to be that I always get back 0xff bytes no matter what should be coming back. I know that the SPI slave device works correctly as it has been tested on a different microcontroller. I am going to try to get my hands on a logic analyzer, but for now does anyone have any idea what might be causing this?
Thanks
int spi_readwrite(unsigned char *sendBuf, unsigned char *recvBuf, int size) { if (size <= 0) { return -1; } int retval; unsigned char discard; ana_Enable(); if (ana_GetCharsInTxBuf() != 0) { printf("spi_readwrite called with non-empty transmit buffer\n"); return -11; } if (ana_GetCharsInRxBuf() != 0) { printf("spi_readwrite called with non-empty receive buffer\n"); return -12; } // Send first for (int i=0; i<size; i++) { // Send character retval = ana_SendChar(sendBuf[i]); switch (retval) { case ERR_OK: break; // move on to the next character case ERR_SPEED: printf("sendChar(%d): ERR_SPEED\n", i); return -2; case ERR_DISABLED: printf("sendChar(%d): ERR_DISABLED\n", i); return -3; case ERR_TXFULL: printf("sendChar(%d): ERR_TXFULL\n", i); return -4; default: printf("sendChar(%d): Unknown response code %d\n", i, retval); return -5; } } // Wait for transmit buffer to empty int tx_bufsize; while (1) { tx_bufsize = ana_GetCharsInTxBuf(); if (tx_bufsize == 0) { break; } } // Wait for receive buffer to fill up int rx_bufsize; while (1) { rx_bufsize = ana_GetCharsInRxBuf(); if (rx_bufsize == size) { break; } } // Receive for (int i=0; i<size; i++) { if (recvBuf == NULL) { retval = ana_RecvChar(&discard); } else { retval = ana_RecvChar(&recvBuf[i]); } switch (retval) { case ERR_OK: break; case ERR_SPEED: printf("recvChar(%d): ERR_SPEED\n", i); return -6; case ERR_RXEMPTY: printf("recvChar(%d): ERR_RX_EMPTY\n", i); return -7; case ERR_OVERRUN: printf("recvChar(%d): ERR_OVERRUN\n", i); return -8; case ERR_FAULT: printf("recvChar(%d): ERR_FAULT\n", i); return -9; default: printf("recvChar(%d): Unknown response code %d\n", i, retval); return -10; } } ana_Disable(); return 0; }
Hi Alice,
I have now changed my code to use SendBlock/RecvBlock instead of SendChar/RecvChar.
I was able to get a logic analyzer trace of an SPI master-write transaction (top to bottom: Clock, MISO, MOSI, Chip Select). From this I can see that the MISO signal is always high and thus the 0xff value is 'correct'. However I think this is happening because the chip-select is toggling between bytes, whereas the slave device requires the chip-select to be held low for the duration of the transaction (usually 5, 8 or 11 bytes).
If I look at the SMasterLdd1 subcomponent, I can see that 'Chip select toggling' is set to 'no'. Do you know why the chip select appears to be toggling and is there a way I can hold it low for the duration of the multi-byte transfer?
Thanks
