Post Go back to editing

LTC2512-24 SPI Clock Polarity and Phase

Category: Datasheet/Specs
Product Number: LTC2512-24
Software Version: NA

Hello.  We are interfacing an LTC2512-24 in an application where we do not need to run at very high speed.  We don't expect to need a SCKA to go beyond 32 MHz, however, in order to sample the first bit of data after DRL goes LOW, we would like to setup the SPI bus with the CLK polarity ACTIVE LOW.   That then allows us to sample on the leading (Falling) edge of the SCKA to ensure we can read the first bit BEFORE the rising edge of the clock triggers the LTC2512 to move to the next bit.

From the LTC2512-24 spec diagrams, it seems that the 'normal' configuration is to have the SCKA polarity set to ACTIVE HIGH, and unfortunately in that configuration sampling on the leading (Rising) edge of the clock isn't meeting the hold time requirement of our processor, which needs around 18ns.

Although there doesn't seem to be a clear statement that the SCKA polarity must be ACTIVE HIGH, will it work correctly with the polarity set to ACTIVE LOW ?

Thanks in advance !

  • Hi  ,

    No. Base on its datasheet. There is no reverse polarity mode. So, it is best advice to configure what is on the datasheet since that would be its optimal performance. And it is already tested. Please refer to page 12 for the timing diagram.

    Regards,
    Jo

  • Thank you very much for the prompt reply, I do appreciate it.  Yes, we did see the timing diagram on page 12 and many of the other pages in the spec and they show the SCKA as inactive LOW.

    However, the very detailed timing diagram for the serial interface, which exactly matches our usage is shown on page 30.  In this diagram, admittedly the only one in the spec that shows this, the SCKA is actually shown as a "don't care" effectively during the conversion phase and until the serial interface is activated.  In this diagram, if the SCKA is inactive HIGH, then there would be an initial low (active) half-cycle of the clock to coincide with the DA31 bit, which is exactly what we need.

    This is all consistent with other areas in the spec that indicate the serial interface should not be used until DRL is asserted and not when a conversion is taking place.

    We have tested this configuration (SCKA active LOW, sampling on falling edge of the SCKA) and we are getting the expected results, in other words, it is 'working'.   However, before committing to our design, we just want to be sure there is nothing harmful to the LTC2512 in having the SCKA HIGH during the inactive periods.

    Again, we understand that the majority of the diagrams show the SCKA as 'low' in the inactive periods, but our application will not function that way due to timing constraints.

  • Hi  ,
     

    The rising edge signifies the shifting of the data. Then after it goes high, SCKA becomes low.
    And if you will be reading your data at low, what you will be getting is the previous data. 
    Thus, it won't harm your LTC2512. Also, errors may occur since the timing of the
    data set and the sclk is not in sync.


    Regards,
    Jo