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.

GPIO protection of TMS320C674x

Other Parts Discussed in Thread: TMS320C6745

Dear e2e members!

Is there anywhere some recommendation on protection of GPIO of TI's DSP, in particular TMS320C6745. (I mean recommended values of current-limiter resistors an so on).

Thank You!
Vitalii

  • Hello Vitalii,

    Could you please elaborate your requirement ?

    Are you going to connect high current input to GPIO ?

    Regards,

    Senthil

  • Hello Senthil!

    Thank You very much for the answer!

    I see that maximum output current of DSP outputs is 4mA(subsection 4.4 of C6747(45) datasheet SPRS377E). How should I protect GPIO from overcurrent. I don`t know the output resistance of GPIO, so I cannot calculate the value of a current-limiting resistor.

    For example in our customer-board the TMS320C6745 DSP interfaces with another uCPU. And during a debugging there is risk that output of DSP will shorted to the output of another uCPU. (the spi sck outp of DSP will be shorted to the spi sck outp of uCPU due to a possible error of the SPI tuning ). 

    There is also a question, why in the subsection 4.4 of C6747(45) datasheet (SPRS377E) the boundary value of the high-level output current is negative (-4mA). And low-level value is positive. I have thought that first one should be positive and the second - negative.

    Thank you very much!

    Vitalii

  • Hello Vitalii,

    I see that maximum output current of DSP outputs is 4mA(subsection 4.4 of C6747(45) datasheet SPRS377E). How should I protect GPIO from overcurrent. I don`t know the output resistance of GPIO, so I cannot calculate the value of a current-limiting resistor.

    The GPIO output can source or sink up to maximum of 4mA. So you need not to be worry about the overcurrent when it is configured as output. The current limiting resistor will be calculated based on this maximum current of 4mA.

     

    For example in our customer-board the TMS320C6745 DSP interfaces with another uCPU. And during a debugging there is risk that output of DSP will shorted to the output of another uCPU. (the spi sck outp of DSP will be shorted to the spi sck outp of uCPU due to a possible error of the SPI tuning ). 

    I don't think the SPI CLK to SPI CLK connection creates problem for you. It is supposed to be connected like this. The software will take of the master slave configuration and set one as output and another as input.

    There is also a question, why in the subsection 4.4 of C6747(45) datasheet (SPRS377E) the boundary value of the high-level output current is negative (-4mA). And low-level value is positive. I have thought that first one should be positive and the second - negative.

    The high level output current is source current and low level output current is sink current. The sign convention is based on the electron flow direction from one point to another. Please refer the below link for better understanding.

    http://en.wikipedia.org/wiki/Current_sources_and_sinks

    Regards,

    Senthil

  • Dear Senthil!

    Thank You very much for the answer and gived information!

    When I speak about SPI, I suggest the case of error: both DSP and uCPU are configured as masters. In this case output will be shorted to output. ... Such errors sometimes appear during software developing or debugging. Hardware environment of DSP must protect it from such errors...

    As I understand, The value of current limiting resistor is (both CPUs are 3.3V):

    R = 3.3V / 4 mA = 825 Ohm. 

    Very big value comparatively with a similar one (22 Ohm) from the Spectrum digital OMAP L137 EVM schematics rev "G" (sheet 4 "Serial I/O").

    And another problem connected with such protection. We plan to boot DSP from uCPU in the SPI0 slave mode. To do this:

    1. Host uCPU sends reset to DSP (Reset to low);

    2. Host uCPU set proper values at the DSP BOOT pins. Note please, that at this stage the used pins of the host uCPU are configured as output;

    3. Host uCPU releases the Reset pin of the DSP. After this DSP configure its units for booting, e.g. SPI0. Begins the booting of the DSP as it is described in the http://processors.wiki.ti.com/index.php?oldid=57073. But note please, that DSP BOOT pins are multiplexed with SPI0 pins. SPI0_SOMI is multiplexed with BOOT[0]. So after the releasing of reset this pin will be reconfigured as OUTPUT. Host uCPU must simultaneously reconfigure its corresponding pin from OUTPUT to INPUT. Otherwise output of the host uCPU will be shorted to the output of the DSP. I think such simultaneous change of a pin state is impossible. It will be always a some delay between the state changing of pins (outp or inp) of host uCPU and slave DSP. So we must protect our DSP GPIOs from overcurrent...

    Dear Senthil, could You please comment my "fears"

    Thank you very much!

    Vitalii

  • Dear Senthil, could You please comment my "fears"

    Thank you very much!

    Vitalii
  • Hi Vitalii,

    This ±4mA source/sink current shown in data sheet is minimum (not maximum indeed!) output current at which the respective min/max output voltage levels (0.4V/2.4V) are guaranteed.

    There is another current limit in datasheet - ±20mA - but again this isn't really the maximum output load current. It is the absolute maximum of allowable clamp current that can flow through the I/O's internal diode protection cells when the voltage at pin extends beyond the limits of (Vdd-Vss). I suppose we may use this current level as a permissible limit for the short-term current contention - at least as a first approximation in the lack of more adequate information.

    These 22 Ohm serial resistors shown at the EVM schematics are primarily the impedance matching ones to prevent any signal integrity issues (overshoots etc).

    Simulation with IBIS models for C6745 shows that the shoot-through current between two output pins of the same type in opposite logic states would be ±45mA without any current limiting resistors, and in the case of a short circuit to Vss or  Vdd the current would be as large as +87/-83mA respectively. The 100 Ohm resistor added between two outputs of C6745 lowers that shoot-through current to the relatively safe level of 20mA max (17mA typ). The value of 22 Ohm limits that current to 35mA max.

    It should be noted that addition of such relatively high resistance as 100 Ohm in series with clock pins can lead to some signal integrity issues and should be simulated beforehand.

    BR,
       Denis

  • Dear Denis!

    Thank You very much for answer and simulation with IBIS models!!!
    So, probably, 22 ... 100 Ohms would protect DSP outputs from overcurrent during DSP software debugging.

    Thank You very much!
    Best Regards
    Vitalii