TMS320F28388D: Configuration of F28388D' SPI

Hello,

Could anybody help me on that? We are still facing the problem.

Thanks,

Lin

Hello,

Can anybody help me on the SPI setting?

Thanks,

Lin

Hi Jason,

It's fine.

I have found the example links and imported the spi_ex1_loopback project. I tried to configure the SPI to control TI's LMX2595 PLL chipset.

The current status is that I can configure LMX2595 by sending SPI write command with SPI_writeDataBlockingNonFIFO() function. However, when I use SPI read command with SPI_readDataBlockingNonFIFO() function, I can only read back the lowest 8-bit data on the MISO line. I hope to get 16-bit data on the MISO line. BTW, I set SPI data width to 8 bit for LMX2595 configuration.

Could you help me on the SPI readback issue? If you need more information, I can send you the main source code, .syscfg files and Saleae capture file.

Looking forward to your further reply.

Best regards,

Lin

Hi Jason,

Thanks for your quick reply and suggestion.

Yes, I did also try using the SPI_writeDataBlockingFIFO( ) and (SPI_readDataBlockingFIFO( ). I enabled the FIFO by checking the box of Use FIFO in the SysConfig GUI. But this time I get all 1's or 0xFF from the readback.

Did I miss any setting for the FIFO before I used it? 

To provide more information for you, I attached my SPI_Read function code. Could you please check it out?

uint16_t lmx2595_read_register(uint8_t regAddr)
{
    uint8_t tx[3] = { (0x80 | regAddr), 0x00, 0x00 };
    uint8_t rx[3];

    // Clear RX FIFO first
    SPI_clearRxFIFO(SPIA_BASE);

    // --- Send 24-bit read command (3 writes = 24 clocks) ---
    SPI_writeDataBlockingFIFO(SPIA_BASE, tx[0] << 8);  // Send MSB
    SPI_writeDataBlockingFIFO(SPIA_BASE, tx[1] << 8);  // Send mid
    SPI_writeDataBlockingFIFO(SPIA_BASE, tx[2] << 8);  // Send LSB

    // Wait for TX complete
    while(SPI_isBusy(SPIA_BASE));

    // Wait until 3 bytes are received in RX FIFO
    while(SPI_getRxFIFOStatus(SPIA_BASE) < 3);

    // --- Read the 3 response bytes from RX FIFO ---
    rx[0] = (uint8_t)(SPI_readDataBlockingFIFO(SPIA_BASE) >> 8);
    rx[1] = (uint8_t)(SPI_readDataBlockingFIFO(SPIA_BASE) >> 8);
    rx[2] = (uint8_t)(SPI_readDataBlockingFIFO(SPIA_BASE) >> 8);

    // Combine middle and last bytes to get 16-bit register value
    uint16_t regVal = ((uint16_t)rx[1] << 8) | rx[2];

    return regVal;
}

Best regards,

Lin

I'd say first, the SPI_isBusy() loop probably isn't necessary.

Second, and more importantly, the bit shifting of the SPI_read/write function calls isn't necessary when using the FIFO.

Regards,
Jason Osborn

Hi Jason,

Thanks for your reviewing my code and pointed some key factors.

I commented out the SPI_isBusy() line and took away the bit shifting in the read function. One more thing I need to clarify is that in my running code actually no SPI_clearRxFIFO(SPIA_BASE); line. (if adding that line, will get a link error: undefined symbol).

So the latest status is that with the bit shifting in the read function, I will get all 0's data; without bit shifting in the read function, I will get all 1's data. Both cases are not correct.

Do you have any further code modification?

Thanks,

Lin

I didn't notice that- the correct function you're looking for is SPI_resetRxFIFO(), not clearRxFIFO.

Are you able to monitor the data lines with an oscilloscope or data analyzer? I'd suggest doing so, if possible, to verify that data is actually being transmitted/received.

Other than this, are you seeing all 1/0 in the regVal return variable, or in the rx array? Put a breakpoint just before the return variable in that function and take a look at the full contents of the rx array. If what you're seeing doesn't make any sense, also use the CCS register view to look at the contents of the SPI Rx emulation register at the same breakpoint, and let me know what all you see in the array and register.

Regards,
Jason Osborn

Hi Jason,

Thank you very much for your correct function info and all the debugging methods.

After replacing SPI_clearRxFIFO() with SPI_resetRxFIFO(), the SPI read function works. I verified results by comparing the rx array or regVal data with the data on MISO line (captured with Saleae Logic Analyzer). Now I can get the whole 16-bit readback data.

The the contents of the SPI Rx emulation register are as below for 4 cases (0x0488, 0x005B, 0x0164, 0x2518):

SPIRXEMU         0x0088/0x005B/0x0064/0x0018           0x00006106@data
       ERXBn         0x0088/0x005B/0x0064/0x0018

Here is one question about the SPI Rx emulation register: why does it only show the lowest 8-bit, not 16-bit or not one by one 8-bit?

Best regards,

Lin

Hi Lin,

I will be taking over support on this thread. Let me familiarize myself with your setup and questions and get back to you.

Best Regards,

Delaney

Hi Lin,

It sounds like your original issue was resolved based on your latest response. For your question about the emulation register: can you verify that "Continuous Refresh" is toggled on in the Register viewer? I believe all 16-bits should show up in the register. Another thing to try would be calling the SPI_readRxEmulationBuffer() function to read the emulation buffer - does this return the full 16-bit value?

Best Regards,

Delaney

Hi Delaney,

Thanks for your following up.

Yes, the issued has been resolved. Only thing is that the register contents do not match with the readback values. I tried that ways, but still not matching.

However, as the main issued is resolved. The left is minor. So if you want we can close this case.

Regards,

Lin