Connect with us

FPGA or shift registers?

Discussion in 'Electronic Basics' started by [email protected], May 24, 2007.

Scroll to continue with content
  1. Guest

    hi,

    I have a device with 34 digital (5V or 0V) inputs. Only one of the 34
    can be 5V at any one time. I'd like to know the number of input which
    is at 5V. I can think of two ways of doing this:

    a) 5 eight-bit shift registers. Each input is connected to a shift
    register and the shift registers are read and outputted one bit at a
    time periodically (left shifting the bits one bit at a time).
    b) programming a FPGA. It'll have 34 inputs and report the number of
    input that was tripped as output.

    The disadvantage of method (a) is that it has more parts and also I
    was told that during the reading of the shift registers there might be
    an error where one bit is skipped, which will result in getting the
    wrong number for the tripped input.
    The disadvantage of method (b) is that I have no idea how to program a
    FPGA.

    Can anyone recommend either of these methods (or a simpler one if it
    exists)? Is it likely that the skipping error in reading the shift
    registers will occur?

    thanks,
    Jon.
     
  2. Jan Nielsen

    Jan Nielsen Guest

    skrev:
    Analog multiplexors is simpler, but still more parts if you stay with
    the inexpensive 8ch ones, and it will use up more pins.
    3for control, 1for in and 1 for enable per chip.
     
  3. Chris

    Chris Guest

    Hi, Jon. Use 5 HC595s if you just want to get it done -- it's
    straightforward, and they work well.

    The whole business about spurious data coming from shift registers is
    mostly people who've either done really sloppy layout, or are actually
    seeing noise at the input of the shift registers.

    Keep all the ICs close to the controlling uC (or whatever you're
    using), and keep the control lines short and direct. Use common sense
    to provide power supply bypass caps for the HC595s.

    If noisy inputs are a problem, you may want to use buffered inputs
    with schmitt trigger inputs. A small R-C at the input will ensure
    that noise is filtered out. Note that layout becomes important if
    you're filtering high frequency noise -- use a ground plane.

    If you want additional safety factor, try latching and reading the
    inputs twice to see that they agree.

    http://www.onsemi.com/pub/Collateral/MC74HC595A-D.PDF

    Good luck
    Chris
     
  4. Rich Grise

    Rich Grise Guest

    You could do it with 6X 74HC148 ($0.55 at DigiKey), if you can make your
    inputs active low. The data sheet
    http://rocky.digikey.com/WebLib/Texas Instruments/Web data/SN74HC148.pdf

    will show you how to cascade them.

    Good Luck!
    Rich
     
  5. Noway2

    Noway2 Guest

    I would consider using a small CPLD as an FPGA would be mass overkill
    for your application. A CPLD would have other advantages over discrete
    logic in that it allows easy modification to the logic structure and can
    be tested and verified before you even put it on the board.

    You mentioned being unsure how to program one. They are really quite
    simple. The programming can be done either with simple schematic
    capture, much like creating a board schematic, or through a text
    language like VHDL.

    If you are still unsure, you can download a free set of development
    tools from either Xilinx or Altera, and experiment with them to see if
    you are comfortable. You could implement your design and then verify it
    with the simulators.

    The actual programming process is done through the tool and utilizes a
    JTAG port. What this means is that on the board you would need to add
    an extra connector (about 10 pins, usually in an inline or .100"x.100"
    header to connect the programmer. Your only real expense would be a
    couple of hundred dollars for a programmer. If you are only going to do
    this one design, then I would consider the discrete logic as it is more
    cost effective. If you think this is something you would use in the
    future, I would suggest that the CPLD may be a good way to go.
     
  6. Jasen

    Jasen Guest

    no, not if it's constructed correctly,

    OTOH

    if only one of the inputs will be high at a time, you could just use a
    bunch of diodes and a few resistors to make a simple encoder.

    another thing to consider is using some "priority encoder" chips.

    Bye.
    Jasen
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-