Connect with us

Noise Reduction in Signal Cable

Discussion in 'Electronic Design' started by Rileyesi, Feb 26, 2004.

Scroll to continue with content
  1. Rileyesi

    Rileyesi Guest

    We make instrument systems where the sensors generate analog DC voltages and
    are connected to the display panel with a signal cable. We have always used a
    multi-condictor cable that has a braided wire overall shield (i.e. no
    individual shields between the condunductors). We have always used analog
    indicators. I've never had complaints about signal noise showing up on the

    However, I have a new installation where digital indicators will be used. I
    THINK that if there is any noise may show up more on the digital indicators
    than on the analog. So, I am looking for ideas on noise reduction.

    To be more specific, the current signal cable has 7 condictors that are gage 18
    in size. 2 are to power the sensors and the other 5 are signal. The range of
    singals is about 0-5 VDC. I recommend to the customer to ground only one side
    of the braided shield. One last thing, the singal cable run is anywhere from
    50 to 2,500 meters. On the 2,500 meter run, we might use an RF link.

    The problem is that our products are installed in a wide variety of
    environments. Anything from insustrial plants to remote fields with nothing
    for miles!

    Any tips or information sources on noise reduction would be appreciated.

  2. Okay, first, change to a twisted pair cable. That will reduce your
    noise tremendously, even though it doubles your number of signal wires.
    Use differential signalling of your data.

    Next suggestion... :cool:

    Edmondson Engineering
    Unique Solutions to Unusual Problems
  3. Rileyesi

    Rileyesi Guest

    Thanks for the reply.

    Just to be sure I understand...

    I would use a twisted pair for each signal. One of the pair would carry the
    signal voltage and the other would be tied to ground? In other words, for all
    signal pairs, one of all pairs would be tied together to ground.

    Question, if I use twisted pair signal cable, do I need the outer or overall
    shield too??

  4. Wellll....
    Actually, I would ground the second wire only at one end, probably the
    transmit side, and would differentially measure the voltage at the
    other, probably with a balanced termination. You ground the sheild only
    at one end to help limit ground loops. I real life, I would use digital
    signalling and optoisolation on a long run like this to protect the
    receiver from transients...

    Edmondson Engineering
    Unique Solutions to Unusual Problems
  5. Like Charles said, i would use differential signalling.

    At the TX side, convert from one wire signal, to 2 wires differential
    At the RX side, convert from 2 wires differential, to 1 wire signal.

    Connect ground only at one side.

    There are plenty of chips out there on the market to perform that job with
    ease, and high speed.

    Antonio Sergio Sena

    Antonio Sergio Sena
    (Field Applications Engineer)

    Primetec - Engenharia de Sistemas, Lda.
    Rua Porto Alegre, 9 - 1º Esq.
    2780-031 Oeiras


  6. Fred Bloggs

    Fred Bloggs Guest

    If the sensor reading was accurate before then there will be no
    difference with the digital meter if you LP filter the hell out of it-
    using the Fc of the previous analog meter would be a good start-
    absolutely nothing else changes, same cable, same sensors, same people.
  7. Mac

    Mac Guest

    I have read over a few of the posts, and I have some thoughts. I think
    Fred Bloggs is right that if it worked before, it can be made to work now
    with a suitable capacitor, with no cable changes. The only caveat is that
    the real part of the impedance on the sensing circuit must be comparable
    or higher. Another way of saying this is that the DC current flowing in
    the SIGNAL cables should be the same or lower. If it is, then adding a
    capacitor to the sensing end of the cable should do the trick. You may
    have to experiment a bit to find the right value.

    I have to admit that sensing DC values across 2.5 km sounds a bit sketchy
    to me, but if it worked before, it should still. Also, if you are sending
    power over a 2.5 km cable run, you better calculate the voltage drop in
    the wire and make sure you can live with it.

    Now, if you would consider making more changes, then you might consider
    designing a unit with one or several analog to digital converters. Then
    you can use some kind of digital communications scheme to send the
    information over the wire. This could save you a lot on wiring, because
    you could probably use a single shielded twisted pair for data. Unshielded
    twisted pair would probably work, too. Afterall, DSL works with fairly
    long wire runs, and has only a single pair of wires with virtually unknown
    impedance characteristics. And it can handle pretty high data rates.

    Good luck!

  8. Since we're talking DC, I don't see a serious issue. The digital indicators
    use an ADC, and likely will have internal signal-conditioning (low-pass
    filtering and overvoltage protection). To ensure you're not getting any
    noise at the receiving end, you might want to look at it with an

    A few other posters suggested using a differential transmission scheme (with
    twisted-pair cable), where neither conductor is ground. For that to work,
    the negative side of the source needs to NOT be at ground potential and, the
    input of the digital indicators needs to not have the low side tied to
    ground. The beauty of differential (or balanced) transmission is that
    longitudinal or common-mode currents are all-but ignored. I used to do
    carrier system maintenance for a well-known Telco, and we used that method
    in T1, T148, and Duobinary transmission. Their repeaters are typically
    placed about 5000ft/1500m apart and though the pulses (effectively, 770KHz
    rate) were pretty rounded-off, there was hardly *any* noise.

    You're right on the mark in recommending grounding the shield at only one
    end (to avoid nasty ground loops that might induce error-producing AC hum
    and/or random noise). I'm thinking simplest is bestest! :^)

  9. Tim Shoppa

    Tim Shoppa Guest

    Others have suggested differential signaling with digital or analog
    voltages... but I'm going to take a different tack:

    Send currents, not voltages. High-impedance voltage circuits by
    their very nature pick up noise. Current loops are potentially vulnerable
    to induced currents from nearby high AC currents, but twisted pairs
    really do help.

    Use one copper pair for each signal you need to send. This means that
    you probably have to go from 7 conductors to 12 conductors, which will
    have some cost.

    But at the other end of the cable you just put milliammeters. Be
    they digital or analog, it doesn't matter.

    Even better: go to the industrial standard of 4-20mA current loops!
    There's a wide variety of prepackages 0-5V and 1-5V to 4-20mA converters
    already out there.

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