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

O.T. "One wire" bus

Expand Messages
  • Peter Balch
    I want to send both power and half-duplex data along a pair of wires - rather like the Maxim iButton. However, I need to send more current down the line -
    Message 1 of 8 , Aug 6, 2012
    View Source
    • 0 Attachment
      I want to send both power and half-duplex data along a pair of wires -
      rather like the Maxim iButton. However, I need to send more current down the
      line - probably 20mA to 30mA.

      I'd like to see how other people have done it but I can't find anyone else's
      designs on the web.

      iButtons allow either the master or slave to transmit a bit by pulling the
      "power" line down to 0V for a few uSec. But an iButton only needs a few uA
      for power.

      An alternative would be to send, say, 12V down the wires and allow either
      end to pull that down to 7V (so there's always enough voltage to power a
      7805).

      Or perhaps send regulated 5V (or 3.3V or whatever) but superimpose an RF
      signal on it when sending a '1' bit.

      The overall idea is to have an unpowered sensor which can be read via a
      handheld "reader". The reader pokes two spikes (rather like a cattle prod)
      onto the device. The device wakes up, makes a measurement and sends a few
      dozen bytes to the reader.

      Any thoughts?

      Peter
    • D. Daniel McGlothin
      ... prod) ... few ... Maybe something like the bias tee . http://en.wikipedia.org/wiki/Bias_tee Think of power over ethernet . Used all the time in ham radio
      Message 2 of 8 , Aug 6, 2012
      View Source
      • 0 Attachment
        > The overall idea is to have an unpowered sensor which can be read via a
        > handheld "reader". The reader pokes two spikes (rather like a cattle
        prod)
        > onto the device. The device wakes up, makes a measurement and sends a
        few
        > dozen bytes to the reader.


        Maybe something like the 'bias tee'.
        http://en.wikipedia.org/wiki/Bias_tee

        Think of 'power over ethernet'.

        Used all the time in ham radio to deliver power to remote antenna
        devices, e.g.,
        http://www.ad5x.com/images/Articles/AntSwitch.pdf
        http://w4ti.net/switch/

        HTH,

        Daniel
      • Randy M. Dumse
        Peter Balch said: Monday, August 06, 2012 10:55 AM ... You know of the 4-20mA current loops used in industry analog signalling? Do a similar thing, but make
        Message 3 of 8 , Aug 6, 2012
        View Source
        • 0 Attachment
          Peter Balch
          said: Monday, August 06, 2012 10:55 AM
          > power and half-duplex data along a pair of wires

          You know of the 4-20mA current loops used in industry analog
          signalling?

          Do a similar thing, but make your zeros and ones extremes of
          current rather than voltage.

          If your equipment will operate on 4mA then just switch in a
          small resistor parallel across the power line for a 1, and leave
          it open for a 0. Resistor size depends on voltage in the loop,
          which should stay constant. For instance, send 5V down your
          probes at 4 mills. Gives you 20mw to operate on. Then use a NPN
          switching a ~320 ohm resistor across the power. That will be
          your 1.

          At the other end, use a sense resistor to detect how much
          current is being drawn. You can recover your digital signal with
          an op amp.

          Randy
        • David Buckley
          Peter Since you are wanting to transfer just a few bytes and you no doubt have a micro at each end it seem ridiculously simple. - Prod - Dump 100ma via an
          Message 4 of 8 , Aug 6, 2012
          View Source
          • 0 Attachment
            Peter
            Since you are wanting to transfer just a few bytes and you no doubt have a micro at each end it seem ridiculously simple.
            - Prod
            - Dump 100ma via an emitter follower into a capacitor via a diode.
            It sounds like it is a human doing the prodding so you have at least 250ms before the user gets bored,
            - Turn off the current source.
            - make the line 0v so the Sensor-unit knows charging has finished.
            - Sensor-unit sends a sign on byte
             - the Reader could beep to indicate communication had been made which would give you another half second to measure and report back.
            - exchange data back and forth over the one wire.
            on the off-chance the Sensor-unit runs out of power it sends a 'need more power message' (byte) and the procedure repeats.
            - Reader could beep again when communication finished.
            - unprod (after second beep).
             
            Royalties gratefully accepted
            David
             
             
            ----- Original Message -----
            Sent: Monday, August 06, 2012 4:55 PM
            Subject: [SeattleRobotics] O.T. "One wire" bus

             

            I want to send both power and half-duplex data along a pair of wires -
            rather like the Maxim iButton. However, I need to send more current down the
            line - probably 20mA to 30mA.

            I'd like to see how other people have done it but I can't find anyone else's
            designs on the web.

            iButtons allow either the master or slave to transmit a bit by pulling the
            "power" line down to 0V for a few uSec. But an iButton only needs a few uA
            for power.

            An alternative would be to send, say, 12V down the wires and allow either
            end to pull that down to 7V (so there's always enough voltage to power a
            7805).

            Or perhaps send regulated 5V (or 3.3V or whatever) but superimpose an RF
            signal on it when sending a '1' bit.

            The overall idea is to have an unpowered sensor which can be read via a
            handheld "reader". The reader pokes two spikes (rather like a cattle prod)
            onto the device. The device wakes up, makes a measurement and sends a few
            dozen bytes to the reader.

            Any thoughts?

            Peter

          • Peter Balch
            Daniel ... That s OK for a system that s _meant_ to be radiating RF but I d be worried about it in, say, an office. Let s say I want to send 100 bytes in a
            Message 5 of 8 , Aug 6, 2012
            View Source
            • 0 Attachment
              Daniel

              > Used all the time in ham radio to deliver power to remote antenna
              > devices, e.g.,
              > http://www.ad5x.com/images/Articles/AntSwitch.pdf
              > http://w4ti.net/switch/

              That's OK for a system that's _meant_ to be radiating RF but I'd be worried
              about it in, say, an office.

              Let's say I want to send 100 bytes in a fraction of a second - that's maybe
              50kbps. How many cycles of RF do I need to encode 1 bit? Maybe 10 cycles. So
              that's 0.5MHz. Even if I was careful to make sure that each cycle was a sine
              wave, there are the square corners as I start and stop the 10 cycles.

              I guess it might be possible to get past emc regulations if one was very
              careful. I don't have enough experience to know.

              What sort of power would you expect the RF to be? Wouldn't I need fairly
              chunky inductors to filter out the RF and get 30mA at 5V to power the sensor
              and uP?


              Randy
              > If your equipment will operate on 4mA then just switch in a
              > small resistor parallel across the power line for a 1,

              I suspect I'm going to need more like 20mA or 30mA. Which makes the resistor
              rather power-hungry. Plus, I'll have a processor of some sort which will
              need a reasonably constant supply.

              > You know of the 4-20mA current loops used in industry analog
              > signalling?
              > Do a similar thing, but make your zeros and ones extremes of
              > current rather than voltage.

              Good idea. When you describe it like that, it sounds rather like a
              telephone. A telephone exchange supplies roughly 48V but, if you're going to
              power equipent from it, it's best to think of it as more like a
              constant-current source. Both people on a phone can talk and listen
              simultaneously. The traditional way is to keep the converstaions separate
              with a "Hybrid" - often a transformer-hybrid:
              http://sound.westhost.com/appnotes/an010.htm

              I only have a very vague idea of how an op-amp hybrid works. How does a
              modern phone get a constant voltage to power its electronics and yet send
              and receive two conversations? All over two wires! Very cunning.

              I was trying to convert a rotary-dial "candlestick" phone to generate DTMF
              and have a memory (for often-used numbers). I got it to work using a PIC but
              the psu for the PIC was too high impedance - the current it takes from the
              phone company is within the right limits but its AC impedance is too low.
              The result is that I can hear other people fine but the transmitted audio is
              way too low. I still haven't worked out how to get a smooth 5V out of the
              phone line without a low AC impedance.

              Anyway, the 48V used by the phone company is way more than I want for my
              reader. The reader is handheld and should be powered with AAs.

              I think that something more like DCC (used by model railroaders) is what I
              need. That sends "AC power" down the track except that the AC is
              Manchester-encoded (more or less). But DCC is strictly one-way data. I need
              half duplex.

              Peter
            • Kevin Ross
              Use a differential line transceiver. The signal is the polarity of the two lines rather than a high/low relationship. You can tap a fair amount of current from
              Message 6 of 8 , Aug 6, 2012
              View Source
              • 0 Attachment
                Use a differential line transceiver. The signal is the polarity of the two
                lines rather than a high/low relationship.

                You can tap a fair amount of current from a device like this. Just need to
                have a circuit that does a full bridge rectification at the end to provide
                constant power.

                -----Original Message-----
                From: Peter Balch
                Sent: Monday, August 06, 2012 3:58 PM
                To: SeattleRobotics@yahoogroups.com
                Subject: Re: [SeattleRobotics] O.T. "One wire" bus

                Daniel

                > Used all the time in ham radio to deliver power to remote antenna
                > devices, e.g.,
                > http://www.ad5x.com/images/Articles/AntSwitch.pdf
                > http://w4ti.net/switch/

                That's OK for a system that's _meant_ to be radiating RF but I'd be worried
                about it in, say, an office.

                Let's say I want to send 100 bytes in a fraction of a second - that's maybe
                50kbps. How many cycles of RF do I need to encode 1 bit? Maybe 10 cycles. So
                that's 0.5MHz. Even if I was careful to make sure that each cycle was a sine
                wave, there are the square corners as I start and stop the 10 cycles.

                I guess it might be possible to get past emc regulations if one was very
                careful. I don't have enough experience to know.

                What sort of power would you expect the RF to be? Wouldn't I need fairly
                chunky inductors to filter out the RF and get 30mA at 5V to power the sensor
                and uP?


                Randy
                > If your equipment will operate on 4mA then just switch in a
                > small resistor parallel across the power line for a 1,

                I suspect I'm going to need more like 20mA or 30mA. Which makes the resistor
                rather power-hungry. Plus, I'll have a processor of some sort which will
                need a reasonably constant supply.

                > You know of the 4-20mA current loops used in industry analog
                > signalling?
                > Do a similar thing, but make your zeros and ones extremes of
                > current rather than voltage.

                Good idea. When you describe it like that, it sounds rather like a
                telephone. A telephone exchange supplies roughly 48V but, if you're going to
                power equipent from it, it's best to think of it as more like a
                constant-current source. Both people on a phone can talk and listen
                simultaneously. The traditional way is to keep the converstaions separate
                with a "Hybrid" - often a transformer-hybrid:
                http://sound.westhost.com/appnotes/an010.htm

                I only have a very vague idea of how an op-amp hybrid works. How does a
                modern phone get a constant voltage to power its electronics and yet send
                and receive two conversations? All over two wires! Very cunning.

                I was trying to convert a rotary-dial "candlestick" phone to generate DTMF
                and have a memory (for often-used numbers). I got it to work using a PIC but
                the psu for the PIC was too high impedance - the current it takes from the
                phone company is within the right limits but its AC impedance is too low.
                The result is that I can hear other people fine but the transmitted audio is
                way too low. I still haven't worked out how to get a smooth 5V out of the
                phone line without a low AC impedance.

                Anyway, the 48V used by the phone company is way more than I want for my
                reader. The reader is handheld and should be powered with AAs.

                I think that something more like DCC (used by model railroaders) is what I
                need. That sends "AC power" down the track except that the AC is
                Manchester-encoded (more or less). But DCC is strictly one-way data. I need
                half duplex.

                Peter



                ------------------------------------

                Visit the SRS Website at http://www.seattlerobotics.orgYahoo! Groups Links
              • Randy M. Dumse
                Peter Balch said: Monday, August 06, 2012 5:58 PM ... Ywa. 4-to-20 was only used to suggest a standard. Any range of current can work. Use 30mA for a zero and
                Message 7 of 8 , Aug 6, 2012
                View Source
                • 0 Attachment
                  Peter Balch said: Monday, August 06, 2012 5:58 PM
                  > Good idea.

                  Ywa. 4-to-20 was only used to suggest a standard. Any range of
                  current can work. Use 30mA for a zero and 34mA for a one (or
                  vice versa). Across the sense resistor these will make two
                  different voltages due to Ohms Law, different voltages cause by
                  different currents. Sense if the current drawn is above or below
                  32 mA.

                  > I only have a very vague idea of how an op-amp hybrid works.

                  Just a voltage comparator against a reference. You look across
                  the differential inputs to see what the voltage is across the
                  sense resistor. If it is below a set point, it's a 0, If it is
                  above a set point, it's a 1 (or vice versa). You can use the
                  difference between 1mA if you like, but then your electronics in
                  the sense has to be very stable. Just be sure your target unit
                  uses less than a set-point-current, and when the extra resistor
                  is switched across the supply, more than the set-point-current.
                  Since you're doing comparators, the output is full on for one
                  rail (maybe need a pull up), or full the other. It's really
                  simple once you get the concept.

                  Randy
                • Peter Balch
                  ... Well, I ve come up with a circuit. The cattle prod prongs are powered from 4 AA cells - i.e. something over 6V - sent through a diode ( diode A ). Let s
                  Message 8 of 8 , Aug 9, 2012
                  View Source
                  • 0 Attachment
                    > I want to send both power and half-duplex data along a pair of wires -
                    > rather like the Maxim iButton. However, I need to send more current down
                    > the
                    > line - probably 20mA to 30mA. <<Actually, more like 10mA - 20mA>>
                    > The overall idea is to have an unpowered sensor which can be read via a
                    > handheld "reader". The reader pokes two spikes (rather like a cattle prod)
                    > onto the device. The device wakes up, makes a measurement and sends a few
                    > dozen bytes to the reader.

                    Well, I've come up with a circuit.

                    The "cattle prod" prongs are powered from 4 AA cells - i.e. something over
                    6V - sent through a diode ("diode A"). Let's say the diode drop is 0.6V.

                    The "sensor" unit includes a PIC with serial comms. The circuit takes the
                    prongs volts, sends it through another diode ("diode B") then smooths the
                    result with a 47uF and 100nF. The resulting voltage is what the PIC uses as
                    power - call it Vdd.

                    There's a 10uF from the PIC serial Tx pin onto the "cattle prod" prong (i.e.
                    on the prong side of diode B).

                    When the PIC transmits, the bits appear on the prong. They go from the Tx
                    pin through the 10uF onto the prong. They can do so because the diode A
                    doesn't present any impedance to soak up the bits. That's because when the
                    bit jiggles up, diode A will be reversed biased.

                    How does the PIC at the other end of the bus receive the bit? The voltage
                    across diode A varies between 0V and 0.6V. That's enough to turn on a
                    transistor (well, it needs a little extra bias current too). So it's a PNP
                    with the emitter is on the positive side of diode A; the base (with a
                    resistor) on the negative side of diode A; and the collector has a resistor
                    to ground. The collector voltage follows the original Tx pin and so can
                    connect straight into the 2nd PIC's Rx pin. (The transistor is being used in
                    common-base mode.)

                    So the reason the "one wire bus" works is because the transmitter doesn't
                    have to work against the low impedance of the battery. Diode A is hery high
                    impedance (at least for the first 0.6V).

                    Why don't the +ve going pulses on the prongs affect the power going into the
                    PIC (via diode B)? Why doesn't the 47uF on the other side of diode B present
                    a low impedance and soak up the pulses?

                    Darned if I know.

                    Peter
                  Your message has been successfully submitted and would be delivered to recipients shortly.