Connect with us

Pulse transformers and RS232 (2nd try)

Discussion in 'Electronic Design' started by [email protected], Jul 25, 2006.

Scroll to continue with content
  1. Guest

    Hi Group,
    I'm looking for a relatively inexpensive way to transmit information
    from one mains switching power supply to up to two other mains
    switching power supplies a few feet away. The data rate would be 1200
    baud or less. The only catch is the common lines can differ by as much
    as 170V, depending on mode of operation. (The common line on each
    power supply is normally at about -170V with respect to earth ground)
    At first glance, this application screams for an infrared connection,
    but this has been discounted because of a stray wire or an object
    interfering with the signal. So, I have been looking at using pulse
    transformers to fully isolate the wired communications connector from
    the -170V. I'll be sending and receiving information through the
    pulse transformers using Atmel, ATTiny2313 microcontrollers. Several
    companies (Tamura, C&D technologies, etc.) make pulse transformers.
    The issue I'm running into is that I have no idea how to drive the
    pulse transformer or how to receive data from the pulse transformer.
    Does anyone have any thoughts on this? (Application notes where are
    Matt Meerian
  2. You can't send RS232 signals thru pulse transformers.

    How about optocouplers?

    They're a lot cheaper than pulse transformers, plus they can pass DC
    signals too.
  3. matt6ft9

    matt6ft9 Guest

    Yes, 'suppose I would have to use Manchester encoding or something
    like that.
    The reason that I don't believe optocouplers would work is the signal
    needs to be compeletely isolated from the -170V. (we can't have a
    pullup from the microcontroller on the external wire) Say we have an
    optocoupler (transmitter and receiver) in device A and we are wanting
    to transmit information to an optocoupler in device B. The LED can be
    energized in the optocoupler in device A. But, there is no power to
    energize the LED in the optocoupler in device B. (it would be helpful
    to have a picture)
    Matt Meerian
  4. Mark

    Mark Guest

    If there is no power in device B, how could a transformer coupled
    circuit work?
  5. and you can't borrow a milliamp from the microcontroller, or the power
    supply because....
  6. Rich Grise

    Rich Grise Guest

    Is this one-way or two-way communication? Either way, optocouplers are
    the way to go - they have isolation up to thousands of volts. Your task
    is to figure out how to arrange them. The fundamental thing is that
    the driver provides the current to the LED, and the receiver uses the PHT
    in its circuit with no electrical connection at all to the driver. The
    leads to the LED would have to be insulated to 170V, of course, but most
    wire is good for that. :)

    You could have one driver drive a bunch of receivers by putting the
    LEDs in parallel, each with its own dropping resistor, or you could
    daisy-chain them, where each one has a link to the other, and maybe
    another link the other direction.

    Then, you could use any old encoding you want to - you just have to figure
    out what topology you want.

    Good Luck!
  7. Joerg

    Joerg Guest

    Hello Matt,

    You can use any code via pulse transformers as long as there is a
    comparator with sufficient hysteresis on the other side. The hysteresis
    makes sure that if there was a high transition it stays high after the
    signal falls back to zero DC. And vice versa for low. The only issue you
    have to deal with is the arbitrary state of that comparator before the
    very first transition. IOW it needs to be "primed" after the system is
    turned on, for example by sending a dummy sign that is ignored.

    Consider some cheap transformers where you get more than one in a pack.
    I have used multi-line ISDN common mode chokes for that (only for low
    voltage differentials), for very high isolation (5kV and stuff like
    that) we wound our own using certified wire. John Larkin recently did an
    isolated power transfer via a LAN transformer. It's rather simple: You
    drive one side with a nice 50:50 duty cycle clock and then rectify and
    filter on the other side. Just like in a normal transformer-driven power
    supply except that the diodes have to be faster and the caps would be
    much smaller because you'd certainly run at tens of kHz and not at 60Hz
    (LAN transformers couldn't do that anyway).

    LAN transformers can be had cheaply four or more to the pack. You could,
    say, use two for the RS232 and the other two to create 9V, 18V or
    whatever you need on the other side from a +5V supply. Or just use one
    and live with about 4V. If your application requires some of the other
    RS232 lines you'd need another set but they cost very little because
    they are found in every LAN router.

    Make sure it's a good brand and the isolation rating is ok. I have seen
    some rather questionable varieties. With reputable manufacturers like
    Murata you would be quite safe. IIRC that's the brand I used.

    One word of caution: Mind the common mode range of the comparator.
    Devices like the LM339 or LM324 can't be driven more than 200-300mV
    below their negative rail or they'll do a tarantella dance.
  8. Matt,
    you seem to be thinking that you need to connect TWO optocouplers AMONG EACH
    OTHER, don't you? That is NEVER the case. The led side is powered by the
    transmitting party, device A in your case. The receiver side is powered by
    the the receiveing party. device B in your case. If you have two way
    communiction you need a second optocoupler for the opposite direction. Look
    at things like the HCPL-2300. They are fast and very easy to use due to
    their low led driving current of 0.5 - 1 mA.

  9. vasile

    vasile Guest

    X10 maybe ?

  10. matt6ft9

    matt6ft9 Guest

    The communications would go only from device A to device B. (one way)
    Yes, if opto isolators were to be used, there would be one in device A
    and one in device B. A rough schematic/system diagram was put up at
    All voltages shown are with respect to earth ground. Note the
    "common" line of device A is at -180V and the "common" line
    of device B is at -165V. We do not want the -165V or -180V on
    the two conductor cable.
    X-10 tends to be expensive for production units.
    Joerg, thanks for the great pointers.
  11. DJ Delorie

    DJ Delorie Guest

    How about fiberoptic? A simple LED and detector with a fiber cable
    between gives you optical isolation with no electricity at all on the
    cable. With fiber audio common in home theaters, it shouldn't be too
    much of a problem getting the parts.
  12. nospam

    nospam Guest

    So use a cheap TOSLINK fiber optic cable and something like a Toshiba
    TOTX173 TORX173 pair.

  13. Joerg

    Joerg Guest

    Hello Matt,

    That concept looks feasible, provided that each unit has it's onw power
    supply and the "Amp and Driver" section is a current source controlled
    by U2. I always thought U2 was a rock band ;-)

    It might be a good idea to reference the cable to ground. Leaving it
    floating invites charge build up and there always comes a point where
    that charge reaches a "bzzzt" limit.
  14. jasen

    jasen Guest

    Sounds like He's got live boards at both ends but doesn't want
    the input or output terminals at either end to be live.

    basically this means he needs to use passive isolation or provide
    a separate supply to power the link.

    transformers would do it, optic fibre is another option, or maybe something
    could be done using capacitors.

  15. joseph2k

    joseph2k Guest

    Capacitively couple a 1200 baud modem.

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