Problems with SPI between TMS570 and PIC18

Alex,

Based on the waveform you send to me I will recommend using the following setup:

PIC: CKP=0, CKE=0, SMP=0.

TMS570 POL=0, PHASE=0.

In your spi_code() function, I did a modification.
This routine in called when the PIC toggles GIOA0.

1] First you prepare the data to be send to the PIC (TMS570 is slave)
    The spiSetData does only access the MIBSPI data memory, setting the data field and a flag bit to specify that the buffer is valid.
2] At that point you can call the spiTransfer routine. This will make the transfer group 0 enable and ready to transmit.
     Because the TMS570 is configured as slave, nothing will happen until the master starts generating the clock.
3] At that point, you can set the GIOA1 to signify to the PIC (master) that the transmission can start.
    I assume that at the point the PIC is ready to start the transfer, and it will start the clock.
4] On the TMS570 side, we have to wait for the completion of Transfer Group 0.
5]  When this is done, it is necessary to read from the MIBSPI received RAM the data.
6] Than it is time to signify the PIC that we are done by toggling GIOA1 to 0.

#3 was happening before #2 in your original code.

There is still something not correct in this sequence.
After the wait for the end of transfer group 0, it is necessary to clear the flag.

Here is the modified version of spi_code()

void spi_code (void)
    {
        /* Write rubbish to ram for transfer */
        spiSetData(spiREG1,SPI1group,tele_rubbish_out);
        y=2;
        
        /* Once PIC starts to send data, transfer rubbish */
        spiTransfer(spiREG1,SPI1group);
        y=3;

        gioSetBit(gioPORTA,1,1);                        //Tell PIC that the TMS570 is now ready (GPIOA1)

        /* Wait until transfer is complete - returns 1 when complete. */
        while (!(spiIsTransferComplete(spiREG1,SPI1group)));

        spiREG1->INTFLGRDY = 0x0001;                                             // Clear flag for End of Transfer Group 0
        y=4;
           
        /* Transfer has completed, received data placed in tele_data_in */
        spiGetData(spiREG1,SPI1group,tele_data_in);   
        y=5;
        
        SPI++;
        
        gioToggleBit(gioPORTA,1);                        //Toggles GPIOA1, in this case to 0
    }

I have checked this code using 2 TMS570 one in Master (to emulate the PIC) and one in Slave.
For clarification, here is the code for the MASTER.

    while(1)
    {
   spiSetData(spiREG1,SPI1group,data_out);           // Write the data to be send to Slave

     gioSetBit(gioPORTA,0,1);
     gioSetBit(gioPORTA,0,0);                                                  // This pulse will trigger an interrupt on the Slave side.
                                                                                            // spi_code() will be executed

     while (gioGetBit(gioPORTA,1)==0);                                     // Wait for handshake from Slave (Slave is ready, data are in buffer, waiting for the clock)

     spiTransfer(spiREG1,SPI1group);                                       // Master can start transfer data.
       do
     {
      result = spiIsTransferComplete(spiREG1,SPI1group);            // Wait until transfer completed.
      count++;
     } while(result==0);
       spiREG1->INTFLGRDY = 0x0001;                                             // Clear the flag.
     while (gioGetBit(gioPORTA,1)==1);                                              // wait for handshake from Slave
    }

Can you have a try and let me know.

Regards,

Jean-Marc

Hi Jean-Marc,

Thank you for looking into this, it turns out that it was a problem with the development boards (I've not looked into exactly what caused the problem), since my last post I received the engineering model of my design and my original code works perfectly on it. 

To confirm I have the waveform registers set up as

PIC: CKP=0, CKE=0, SMP=0.

TMS570 POL=0, PHASE=0.

The code also works without the GIOA handshakes, though I will keep them in for greater control.

Thanks again for looking into my problem.

Kind regards,

Alex