Loading ...
Sorry, an error occurred while loading the content.

Lengthy explaination of noise problems that I posted on another forum.

Expand Messages
  • wuhu_software
    Conversation about the WS3600 and the handshaking lines for communications... ... Well, I definetly agree that it is extremely odd to implement any sort of
    Message 1 of 4 , Feb 3, 2006
    • 0 Attachment
      Conversation about the WS3600 and the handshaking lines for
      communications...

      ---

      Well, I definetly agree that it is extremely odd to implement any
      sort of protocol using the handshake lines.

      The only reason I could see for doing so is so that you could reduce
      your production costs by eliminating the serial tranceiver chip
      (UART). I am 99% sure there is a real UART in the WS2310 stations.

      I am guessing that is what they did. In this way, you could have the
      software (embedded on the station) control the lines, albeit at a
      much slower rate.

      I think that is a common problem with companys trying save a buck or
      two, they tend to go off the deep end with cost savings. This
      interface reeks of such a cost saving measure.

      Now as far as your situation goes, I am suspecting that it may suffer
      a common problem with the WS2310.

      On the WS2310, they are apparently using the DTR and RTS as not only
      a virtual ground, but also as a source of power for some circuit they
      have in the unit (maybe powering the transceiver circuit).

      This is all fine and dandy if it is done correctly. I believe that La
      Crosse has not really looked closely at this problem.

      In a previous post, I alluded to the fact that I have found another
      possible noise issue (besides the wiring from the thermo unit to the
      anonemeter).

      When I was monitoring communications between the unit and
      Heavyweather, I happened to notice that a bad checksum byte was
      returned by the unit to the software. Assuming you have any sort of
      decent quality in your communication lines, this should be a very
      rare occurrence.

      I decided to pull out the oscilliscope and check out the quality of
      the signals.

      After a while, I noticed that there were high frequency voltage
      spikes on the DTR and RTS lines.

      As I mentioned, the DTR and RTS are sent to +12vdc and -12vdc to
      provide a virtual ground (you have to have some sort of reference
      ground when using serial communications).

      It also appears they may be using these lines to provide parasitic
      power to their circuit(s).

      When the data is sent back to the PC, it is not in the normal range
      of +12vdc, -12vdc. Like many newer serial ports, the voltage is more
      like +4vdc,-4vdc. This range is also typical for laptop serial ports
      where they are trying to save $$$. (Em, most cost saving measures...)

      Now watching the DTR and RTS on a scope, I noticed 1vdc high
      frequency spikes on the lines.

      This translated in to 2vdc high frequency spikes on the PC RX line.
      Because the RX line is only +/- 4vdc, this presents a problem and
      could generate false data on the PC line (your in the 2vdc range when
      you have the spikes).

      The software may not see bad framing errors and other garbage because
      NT filters out the noise unless you are looking for it. One way it
      can show up is a bad checksum byte. If you have a corrupt bit in the
      checksum itself, or a bad bit in the data returned, the checksum will
      fail and this is what I have seen (within 30 seconds on my unit).

      So this noise is clearly on the PC Rx line and I suspect may be
      traveling through other circuits of the weather station. In either
      case, this is very undesirable.

      To rectify this situation, you can take a .01 microfarid capacitor
      (non-directional, determined .01 by trial and error), and tie one end
      to your PCs RX line and the other to signal ground (or the PC ground
      if easier).

      This removes the high frequency noise generated by the weather
      station from the PC RX line.

      You can also add a 1uF capacitor to the RTS and DTR lines (to signal
      ground), smoothing out the high frequency spikes there as well. This
      may or may not improve the situation on the weather station circuits,
      we would have to perform a live disection to determine if this is
      true.

      Now going back to your problem, I am suspecting that they used a
      similar approach with the parasitic power (and virual ground) and are
      creating spikes on the lines the PC is monitoring.

      The only way to know for sure is to hook up a scope and look at the
      quality of the data.

      Finally, if anyone has a WS2310 or WS3600, and you have a
      oscilliscope, it would be helpful if you could check this out
      yourself.

      You can buy a two-jack telephone adapter and radio shack and create a
      break out box for monitoring the signals (make sure you cross the
      lines so the box is a pass-through, I learned the hard way and had to
      re-sync my station and outdoor sensor! Ugh).

      Once you have the breakout box, you can also add the capacitors to
      see the effect.

      Eventually the idea is to create a small 9-pin to 9-pin converter
      that will pass through the signals as well as hold the capacitors
      (and tie them to signal ground).

      You would just put this between your PC serial port and the WS 9-pin
      cable.

      So I will leave it there, if anyone wants to lend a hand, let me know.

      Thanks.
    • stevech
      The open3600 software, in C, talks to the 3600 directly via a serial port. Studying that code, you can see the protocol.
      Message 2 of 4 , Feb 4, 2006
      • 0 Attachment
        The "open3600" software, in C, talks to the 3600 directly via a serial port.
        Studying that code, you can see the protocol.

        http://open3600.fast-mail.nl/tiki-index.php


        -----Original Message-----
        From: wuhu_software_group@yahoogroups.com
        [mailto:wuhu_software_group@yahoogroups.com] On Behalf Of wuhu_software
        Sent: Friday, February 03, 2006 1:10 PM
        To: wuhu_software_group@yahoogroups.com
        Subject: [wuhu_software_group] Lengthy explaination of noise problems that I
        posted on another forum.


        Conversation about the WS3600 and the handshaking lines for
        communications...

        ---

        Well, I definetly agree that it is extremely odd to implement any
        sort of protocol using the handshake lines.

        The only reason I could see for doing so is so that you could reduce
        your production costs by eliminating the serial tranceiver chip
        (UART). I am 99% sure there is a real UART in the WS2310 stations.

        I am guessing that is what they did. In this way, you could have the
        software (embedded on the station) control the lines, albeit at a
        much slower rate.

        I think that is a common problem with companys trying save a buck or
        two, they tend to go off the deep end with cost savings. This
        interface reeks of such a cost saving measure.

        Now as far as your situation goes, I am suspecting that it may suffer
        a common problem with the WS2310.

        On the WS2310, they are apparently using the DTR and RTS as not only
        a virtual ground, but also as a source of power for some circuit they
        have in the unit (maybe powering the transceiver circuit).

        This is all fine and dandy if it is done correctly. I believe that La
        Crosse has not really looked closely at this problem.

        In a previous post, I alluded to the fact that I have found another
        possible noise issue (besides the wiring from the thermo unit to the
        anonemeter).

        When I was monitoring communications between the unit and
        Heavyweather, I happened to notice that a bad checksum byte was
        returned by the unit to the software. Assuming you have any sort of
        decent quality in your communication lines, this should be a very
        rare occurrence.

        I decided to pull out the oscilliscope and check out the quality of
        the signals.

        After a while, I noticed that there were high frequency voltage
        spikes on the DTR and RTS lines.

        As I mentioned, the DTR and RTS are sent to +12vdc and -12vdc to
        provide a virtual ground (you have to have some sort of reference
        ground when using serial communications).

        It also appears they may be using these lines to provide parasitic
        power to their circuit(s).

        When the data is sent back to the PC, it is not in the normal range
        of +12vdc, -12vdc. Like many newer serial ports, the voltage is more
        like +4vdc,-4vdc. This range is also typical for laptop serial ports
        where they are trying to save $$$. (Em, most cost saving measures...)

        Now watching the DTR and RTS on a scope, I noticed 1vdc high
        frequency spikes on the lines.

        This translated in to 2vdc high frequency spikes on the PC RX line.
        Because the RX line is only +/- 4vdc, this presents a problem and
        could generate false data on the PC line (your in the 2vdc range when
        you have the spikes).

        The software may not see bad framing errors and other garbage because
        NT filters out the noise unless you are looking for it. One way it
        can show up is a bad checksum byte. If you have a corrupt bit in the
        checksum itself, or a bad bit in the data returned, the checksum will
        fail and this is what I have seen (within 30 seconds on my unit).

        So this noise is clearly on the PC Rx line and I suspect may be
        traveling through other circuits of the weather station. In either
        case, this is very undesirable.

        To rectify this situation, you can take a .01 microfarid capacitor
        (non-directional, determined .01 by trial and error), and tie one end
        to your PCs RX line and the other to signal ground (or the PC ground
        if easier).

        This removes the high frequency noise generated by the weather
        station from the PC RX line.

        You can also add a 1uF capacitor to the RTS and DTR lines (to signal
        ground), smoothing out the high frequency spikes there as well. This
        may or may not improve the situation on the weather station circuits,
        we would have to perform a live disection to determine if this is
        true.

        Now going back to your problem, I am suspecting that they used a
        similar approach with the parasitic power (and virual ground) and are
        creating spikes on the lines the PC is monitoring.

        The only way to know for sure is to hook up a scope and look at the
        quality of the data.

        Finally, if anyone has a WS2310 or WS3600, and you have a
        oscilliscope, it would be helpful if you could check this out
        yourself.

        You can buy a two-jack telephone adapter and radio shack and create a
        break out box for monitoring the signals (make sure you cross the
        lines so the box is a pass-through, I learned the hard way and had to
        re-sync my station and outdoor sensor! Ugh).

        Once you have the breakout box, you can also add the capacitors to
        see the effect.

        Eventually the idea is to create a small 9-pin to 9-pin converter
        that will pass through the signals as well as hold the capacitors
        (and tie them to signal ground).

        You would just put this between your PC serial port and the WS 9-pin
        cable.

        So I will leave it there, if anyone wants to lend a hand, let me know.

        Thanks.







        Yahoo! Groups Links
      • stevech
        The open3600 software, in C, talks to the 3600 directly via a serial port. Studying that code, you can see the protocol.
        Message 3 of 4 , Feb 4, 2006
        • 0 Attachment
          The "open3600" software, in C, talks to the 3600 directly via a serial port.
          Studying that code, you can see the protocol.

          http://open3600.fast-mail.nl/tiki-index.php
        • wuhu_software
          Steve, Looked over the code, it looks fairly straight forward. If I had a ws3600, I would probably implement it. ... serial port.
          Message 4 of 4 , Feb 4, 2006
          • 0 Attachment
            Steve,

            Looked over the code, it looks fairly straight forward.

            If I had a ws3600, I would probably implement it.

            --- In wuhu_software_group@yahoogroups.com, "stevech" <stevech@...>
            wrote:
            >
            > The "open3600" software, in C, talks to the 3600 directly via a
            serial port.
            > Studying that code, you can see the protocol.
            >
            > http://open3600.fast-mail.nl/tiki-index.php
            >
            >
          Your message has been successfully submitted and would be delivered to recipients shortly.