This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

FSI at 50MHz

Part Number: TMS320F28388D
Other Parts Discussed in Thread: TMDSCNCD28388D, C2000WARE

My customer uses two F28388D boards for FSI communication with DMA.

Communication works normally at 25MHz clock, but CLK and DATA signals are broken at 50MHz.

CLK = 50MHz

at 50MHz

CLK = 25MHz

at 25MHz

Both boards were connected to the same 3.3V and GND,

the master's TX was connected to the slave's RX through a buffer, and Master's RX was directly connected to the TX without a buffer.

And there is a 0 ohm resistor on the signal line.

Buffer is NC7WZ125 (OnSemi), and has propagation delay as below.

Are there any cautions when increasing the Clock?

Please let me know what to check.

  • Hi SY Yoon,

    What software example are you testing with and do you see errors being reported while testing in CCS? It's possible the clock signal is shown gaps at 50 MHz because of the signal analyzer sampling bandwidth. Can you verify the analyzer's sampling speed or use a oscilloscope instead?

    Best,

    Kevin

  • Hi, 

    The analyzer's sampling speed is 100M/s and he saw errors in CCS.

    A CRC error occurs when the master receives data.

    Failure rate is about 25%. 

    He used a mixture of several examples.

    dma_ex1_gsram_transfer

    fsi_ex16_daisy_handshake_lead

    fsi_ex16_daisy_handshake_node

    fsi_ex3_loopback_dmacontrol

    multi_axis_lead_ctrl_f2838x_cpu1

  • Hi SY Yoon,

    The analyzer's sampling speed is 100M/s and he saw errors in CCS.

    A CRC error occurs when the master receives data.

    OK, sounds like it must be a signal integrity or hardware issue then. Have to keep in mind that these are 50MHz signals which is fairly high frequency.

    Can you further explain the hardware and connections that are being used? Are two TMDSCNCD28388D boards being used and connected together in some way? If so, are they using the J2 connector on the board or making connections from the docking station?

    Note if using J2 connector specifically then there are some resistors that should be removed.

    Best,

    Kevin

  • Hi, Kevin,

    My customer doesn't use the TMDSCNCD28388D board. He uses his custom board.

    The captured image has 0 ohm resistors. According to you, should I remove this?

    My customer's circuit also has a 0 ohm resistors.

  • Hi SY Yoon,

    The analyzer's sampling speed is 100M/s and he saw errors in CCS.

    It would be better to measure with a higher bandwidth oscilloscope. This is just at Nyquist frequency for the 50 MHz FSI clock.

    Ok understand. The issue is likely hardware related and they should further analyze the FSI signals on their boards. Some things to consider below:

    • There could be interference / noise on the signals
    • The length of the signals could be too long, or there could be stubs
    • Too much skew could be getting introduced between the clock and data signals

    The captured image has 0 ohm resistors. According to you, should I remove this?

    My customer's circuit also has a 0 ohm resistors.

    Not necessarily. These resistors are relevant to the FSI signals on TMDSCNCD28388D, but I don't know what they are for on your customer's board.

    Best,

    Kevin

  • Hi, 

    My customer tried the FSI example (fsi_ex16_daisy_handshake_lead and fsi_ex16_daisy_handshake_node) in the C2000Ware.

    These examples worked at 25 MHz, but not at 50 MHz.

    I think it's a HW problem.

    Could you review the schematic and PCB layout?

  • Hi SY Yoon,

    Sure, I can take a quick look at their FSI signals in their HW design. Do you prefer to send it to me directly?

    Best,

    Kevin