Connect with us

Ribbon cable termination

Discussion in 'Electronic Design' started by Noway2, Jan 19, 2006.

Scroll to continue with content
  1. Noway2

    Noway2 Guest

    In a project, I am planning to use a flat ribbon cable, Z0 ~ 100 Ohms,
    with a length of about 3 feet to connect a couple of logic boards. I
    am wondering what is the best way to terminate the cable. From my
    readings, I am thinking that I will want to terminate near the
    receiving connector with a 100 Ohm resistor and a small capacitor for a
    typical AC termination scheme. From that point, my plan would be to
    pick up the signal with a schmitt trigger buffer / inverter to drive
    the logic that receives this signal. This goes against the concept of
    terminating the line past the last receiver, but it does terminate it
    at the point where the 100 Ohm cable hits the PCB traces. From the
    point of ribbon / pcb interface, the connection length to the schmitt
    trigger will be short, under a couple of inches.

    I am also thinking that it should be suitable to drive the circuit with
    standard logic gates as they have a much lower output impedance than
    the cable.

    Does this approach sound correct or is there a better way?

    In case it matters, there will be various address, data, control and a
    clk signal corresponding to the addresses and data transmitted by
    ribbon cable.
  2. Jon

    Jon Guest

    The best way to terminate the cable is to use a differential line
    receiver. Most of them give you the option of using a buit-in
    termination resistor (usually approximately 125 Ohms) or using your own
    termination resistor.
  3. Noway2

    Noway2 Guest

    Would I then need a differential transmitter too or is the differential
    input simply the signal and ground?
  4. Jon

    Jon Guest

    You can use a differential reciever in a single ended fashion, with one
    input connected to ground. A differential transmitter-receiver pair
    will give you the best common mode noise immunity.
  5. Noway2

    Noway2 Guest

    Thank you for the suggestion. Initially I was thinking about using a
    34 pin ribbon connection and running a parallel bus. I am starting to
    consider using a smaller, perhaps 10 pin, ribbon connection and running
    a few lines of signal. I want to use a mass termination (ribbon or
    similar) cable because of manufacturing considerations.
  6. John B

    John B Guest

    The 'old-fashioned' way of terminating logic lines was 220R to Ground and 330R to +5V. I think that
    SIL terminators are still available.
  7. Rich Grise

    Rich Grise Guest

    If you're terminating TTL, you've got it upside down. Also, an "active
    termination" is one where you have that 330/220 voltage divider connected
    to the base of an NPN transistor, with an additional 330R (or so) to each

  8. Joerg

    Joerg Guest

    Hello John,
    We called them space heaters. I usually prefer AC termination, much
    lower dissipation.

    Regards, Joerg
  9. Paul Burke

    Paul Burke Guest

    Yes, frequency dependent of course, but 82 ohms in series with 220pF is
    quite common.

    Paul Burke
  10. John B

    John B Guest

    On 20/01/2006 the venerable Rich Grise etched in runes:

    or ars* over t*t as some might say but I prefer head over heels.

    Thanks Rich. Too much Merlot again (me not you).
  11. Whatever standard logic gates mean to you.
    If a common mode range of almost zero is acceptable,
    then I'd use LVDS, dissipating just 3mA into 100 Ohms
    is not much.

  12. Joerg

    Joerg Guest

    Hello Paul,
    220pF seems a bit high for a short cable. I try to use a value large
    enough to muffle reflection spikes but small enough not to mess with the
    load characteristics in the frequency range of the data.

    Regards, Joerg
  13. Noway2

    Noway2 Guest

    I think I have talked myself into a varient of SPI using LVDS
    transcievers over an 8 wire ribbon cable. I am thinking that I will
    use a master clock, from the main controller, with a data in and data
    out channel, like SPI. There are no addresses or enabled to worry
    about since it is a single point to point channel. The varient comes
    in to play in that I am going to have one line driven by the slave
    board that is a INT to the computer board. That way, I shouldn't even
    have to poll the board to see if it is ready. I am also planning on
    implementing CRC in the messages as this is easy to do in hardware with
    a serial transmission, though the electrical properties of the medium
    should make errror occurances very low. Another nice thing about this
    arrangement is that I will have two drivers and two receivers at both
    ends, which is a common configuration for differential transcievers and
    I can then use the same part in both places.

    Again, thanks to everybody for their help,
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