Digital isolator: enable pin?

Discussion in 'The Projects Forum' started by exscape, Feb 23, 2012.

  1. exscape

    Thread Starter New Member

    Jan 19, 2012
    I'm looking to connect an ADC and a microcontroller (Atmega*) that have no shared ground (they're on different rails in a dual-floating rail power supply).
    To do this, I've been looking at digital isolators, namely the Silicon Labs Si8441 which appears to be perfect for the purpose; 3/1 channels (3 one way, 1 the other) which fits SPI great.

    However, I ran in to some confusion when reading the data sheet. The part has two "enable" pins, but I'm unsure of exactly why. Here's the relevant stuff from the data sheet (full data sheet):
    (These caps are the relevant pages (20 and 21) from the data sheet above.)

    My goal is simply to isolate the ADC from the µC; all 4 ports will be used. Does this mean I should simply wire the PCB so that +5V goes not only to the Vdd pins, but also the enable pins?

    Also, bonus question: is there anything more to it than to wire up all the IO pins, the 2 power supplies and enable pins? That is, does it take anything else, or will it work exactly the same as ADC -> µC directly after that? My understanding is that it'll "just work", since it's not an optoisolator (with the diode voltage troubles and all), and since the bandwidth it can handle is far above what's required.
  2. praondevou

    AAC Fanatic!

    Jul 9, 2011
    EN enables the outputs , this is why you have one for A4 and one for B1, B2, and B3.

    Tie them to VCC on their respective side of the isolation barrier. Read page 21 through 26 of the datasheet, there is also an errata sheet.

    There are fast digital optoisolators too, don't know why it would HAVE TO be the one you proposed.
  3. exscape

    Thread Starter New Member

    Jan 19, 2012
    Thanks, that's what I thought; I just didn't (and don't) see why you'd ever want to disable them, so I also don't see why the pins even exist.
    I did read the relevant sections in the data sheet multiple times before asking, of course.

    Also, I don't think I said it HAS to be this one. ;)
    It's cheap ($2.5, compared to the larger solutions I was looking at earlier) and easy (no need for opamps at each input, etc).

    Edit: Hmm, I read your post as "THESE are fast", not "there are" - surely 150 Mbps should be enough for some basic SPI communication?! :)