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.

TMS320C6424 32 bit PCI interface to Freescale MPC8313E

Other Parts Discussed in Thread: TMS320C6424

I interfaced a DSP (TMS320C6424) with a Freescale CPU (MPC8313E) using the 32 bit PCI Interface.

There is actually no problem. Everything work perfectly fine. I probed some signals on the PCI interface to check the signal quality and the signals are mostly very badly shaped: Down to -1.05V undershoot and up to +4.24V overshoot. It certainly don't respect the absolute maximum rating.

I did not use any serial resistors on the PCI signals as everybody seams to say that they are not required. There are not many examples of using the parallel PCI bus and some are contradictory.

The nets are about 4 inches long with impedance controlled to 50 ohms +/- 10%. The MPC8313 is configured as PCI master and the C6424 as PCI slave. I think that +4.24V overshoot may be harmful for the internal PCI clamping diode and some precautions need to be considered. A 33 ohms series resistors may be required, at least on one of the devices (The TMS320 or the MPC8313), or maybe on both.

I may like to have your opinion about it.

  • Hello Denis,

    Welcome to TI E2E Forum!

    The reflection in the PCI traces is wholly depends on PCB design. I would suggest you to perform simulation and find the optimal series resistance to minimize the reflections. In general, you can use 33 ohm series resistor and check the overshoot and undershoot level.

    Regards,
    Senthil
  • Thanks Senthil,
    I was suspecting something like this.

    In the PCI V2.2 specifications, a PCI bridge usually has relatively weak output driver and assume a trace impedance of 65 ohms with no termination. In this case, series resistors are not required.

    In the MPC8313E Hardware Specification, Freescale indicates that the AC test load used for the PCI signals is formed of a 50 ohms trace and 50 ohms termination to NVcc/2. This does not correspond to PCI specifications and will also required series resistors to achieve current limiting required for Reflected wave switching.

    This answers my question as if the PCI driver is current limited for the Reflected Wave Switching operation of the PCI interface. The answer is No and external resistors must be used to match the trace impedance on both devices.

    Thanks,
    Regards,
  • In the event that we don't have enough space on the PCB to add 54 series resistors on both end of the PCI interface to reshape the waveform to match the PCI specifications, can we assume that the clamping diode on the receiver will be sufficient to absorb the energy of the overshoot and undershoot without the risk of damaging the chip on a long term?

    In other words, the serial damping resistors are they mandatory on the C6424 PCI interface?

    Thanks,

  • Hello Denis,

    I am not sure the receiver would have the capability to absorb the reflected energy. Please check with the receiver device vendor to clarify this.

    To my knowledge, PCI spec does not call for any series termination. You can perform simulation to find out optimal series resistor if required.

    Regards,
    Senthil
  • Hi Senthil,

    The MPC8313 is the host and the TMS320C6424 is the device. So, the C6424 is the receiver.

    You are correct, the PCI spec does not call for any series termination in the case were the chips (Host and device) are PCI compliant.
    In the PCI V2.2 specifications, a PCI bridge usually has relatively weak output driver and assume a trace impedance of 65 ohms with no termination. In this case, series resistors are not required. But apparently, neither of the chips MPC8313 or TMS320 are fully PCI compliant.

    For example, on AD31, when a data is transfered from the TMS320 to the MPC8313. There is an overshoot of +3.78V on the receiver (MPC8313) and an overshoot of +4.36V on the TMS320 resulting from the echo return from the unterminated net.

    In the PCI specifications, the driver must drive the signal half way to the high level until the echo return from the unterminated net and pull the signal to the high logical level (+3.3V). This is called "Reflected Wave Switching". In this case, there is no overshoot on the signal and no termination is required.

    Regards,
    Denis
  • Hello Senthil,

    By the way, I am evaluating simulators to avoid the net termination problem.
    I am thinking about HyperLynx but I am not sure it is the best software for the job.

    Do you have any suggestion?

    Thanks,
    Denis
  • Hello Denis,

    Yes, you can use Hyperlynx tool for the simulation.

    Regards,
    Senthil
  • Great,
    Thank you very much,
  • Hi Senthil,

    About the PCI interface. Freescale have some overshoot tolerance specifications in addition to the "Absolute Maximum Rating".
    The absolute maximum rating for the PCI interface is : -0.3V to Vdd +0.3V
    The overshoot tolerance for the PCI interface is : -0.7V to Vdd +20% for a maximum period of 3ns.

    Is there a similar tolerance for the TMS320C6424?

    regards,
    Denis
  • Hello Denis,

    TI has provided the absolute maximum rating for the PCI interface as –0.5 V to DVDD33 + 0.5 V for C6424. However there is no specification on overshoot tolerance for PCI interface.

    Regards,
    Senthil