Connect with us

noisy differential signal

Discussion in 'General Electronics Discussion' started by foTONICS, Feb 4, 2016.

Scroll to continue with content
  1. foTONICS

    foTONICS

    332
    9
    Sep 30, 2011
    Hey everyone,

    I was probing a signal at work and noticed that it was very noisy, also that this noise had a very pronounced sinusoidal shape to it. Each spike in the noise is on an either rising or falling edge. This particular signal is the (+) half, RS422. Is it possible that the larger spikes, reaching just over 6Vp-p, could damage any sort of input? The noise didn't seem to have any pattern to it:

    upload_2016-2-4_9-18-32.png
    upload_2016-2-4_9-18-58.png

    Could controlling the rise/fall times eliminate these spikes or would that only take care of the harmonics generated by the ringing?
     
  2. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,385
    2,772
    Jan 21, 2010
    its a digital scope, right?

    What you may be seeing is ringing on the rising and falling edges combined with a sample rate below the fundamental frequency of the ringing (causing aliasing).

    The ringing may be present in the signal or caused by the signal probes (which will not be terminated correctly for this signal). Is the signal itself terminated?

    Are you measuring this signal with a differential probe, or are you measuring one side with respect to ground?

    In any case, the specs for these sorts of serial interfaces typically require the receivers to be very tolerant and my gut feeling is that they wouldn't be damaged by it. However, it would be worth looking at the datasheet for the receiver and confirming this. Note that there may be additional input protection.
     
  3. foTONICS

    foTONICS

    332
    9
    Sep 30, 2011
    Yes, Lecroy waverunner6050a

    I've noted the ringing at 43MHz on the rising edge and 49MHz on the falling edge:
    upload_2016-2-4_10-36-42.png

    Wikipedia mentioned having a sample rate greater than twice the fundamental frequency, in this case 43MHz and 49MHz, does this sound appropriate? ( Wikipedia isn't too too trusted ).

    I'm using 10:1 probes measure with respect to ground, the unit I'm measuring is enclosed in something that I can't open so we have to view it as a black box scenario. I've purchased a mating connector and terminated it with a DE9 connector. Could what I be seeing be just nasty reflections and harmonics generated by improperly terminated leads?
     

    Attached Files:

  4. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,385
    2,772
    Jan 21, 2010
    Yes, at least double the frequency, preferably higher, but you've shown a better image and it is certainly ringing.

    check this document. It gives specs that you can measure against and shows you how to measure the various characteristics.

    http://www.ti.com/lit/an/slla070d/slla070d.pdf

    The first few pages answer most of your questions, and it may be worth terminating the signal with 100Ω as this should reduce the ringing.

    note that the document above does specify the limits for voltage she unterminated.
     
  5. foTONICS

    foTONICS

    332
    9
    Sep 30, 2011
    Would this 100Ω be across the two differential signals + and - or from the half I'm measuring and ground?
     
  6. AnalogKid

    AnalogKid

    2,358
    659
    Jun 10, 2015
    Could be scope probe compensation, too long a ground clip lead, or some other inductive high-freq effect. Does the scope have a calibration waveform output? If so, how does it look?

    ak
     
  7. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,385
    2,772
    Jan 21, 2010
    100Ω between the signal pairs.

    As AK says (and I mentioned previously) it could be the probes contributing to the problems.
     
  8. foTONICS

    foTONICS

    332
    9
    Sep 30, 2011
    I tested a waveform of similar voltage levels, frequency, and it's also differential and it looks fine:

    upload_2016-2-4_15-26-49.png
     
  9. foTONICS

    foTONICS

    332
    9
    Sep 30, 2011
    Once I get to it I'll snap a picture and show you, just finishing up some more important stuff
     
  10. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,385
    2,772
    Jan 21, 2010
    Ringing is not unusual, see if terminating the signals improves it.

    Signals from different sources don't need to look identical to be within specifications.
     
  11. foTONICS

    foTONICS

    332
    9
    Sep 30, 2011
    Checked out the datasheet and they mention terminating it with a 120Ω resistor at the RS422 receiver input, I terminated it with the closest I had, 100Ω, and that took away almost all the ringing:
    upload_2016-2-5_13-26-46.png

    But this brings up something that I read the other day concerning transmission line termination. It mentioned that the signal lines should be terminated at the output of the RS422 chip to help stop emissions, but it looks like they didn't terminate it at all...
     
  12. dorke

    dorke

    2,342
    665
    Jun 20, 2015
    For RS-422 termination is needed at the receiving-end and it depends on the type of twisted wires you use(they are twisted,right?).
    What you did is the correct way to do it,
    nothing else is required.
    BTW,
    what is the rate of the signaling and what is the distance?
     
  13. foTONICS

    foTONICS

    332
    9
    Sep 30, 2011
    yes they're twisted. The rate is 200Hz and the distance is only about 10 inches or so
     
  14. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,385
    2,772
    Jan 21, 2010
    RS422 can be used un a multidrop configuration. In that case only the receiver at the end of the line should be terminated.
     
  15. dorke

    dorke

    2,342
    665
    Jun 20, 2015
    Well steve,

    Not so,
    the termination of RS-422 is always(!) at the receiving side (the end of the transmission line).
    RS-422 supports point to multi-point (one driver and many receivers)so the the "end of the transmission line" is at the farthest away receiver.
    If there are long stubs along the above line, this arrangement may have problems of reflections from the stub.
    so termination at the end of the stub is advised.

    In Multi-Drop(RS-485 multi-point to multi-point) there are many drivers/receivers working in half-duplex mode .
    In this case the "end of the transmission line" is at the two end-points of the transmission line which are both terminated.

    See here par.4
     
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

-