Connect with us

flyback converter driving

Discussion in 'Electronic Design' started by Jamie Morken, Oct 19, 2007.

Scroll to continue with content
  1. Jamie Morken

    Jamie Morken Guest

    I was using the LT1619. Datasheet shows 40uA max current draw in
    shutdown mode. Simulator shows about 265uA :)
    I think the current draw will need to be greater in order to charge the
    capacitor up within ~100ms max too.
    Thanks! I put in that new circuit, I think it is working fine, but
    the simulation gets stuck at 147ms for some reason doing "transient
    analysis" so I don't know for sure. I think I might have to increase
    the inductance of the L2 coil so that Vcc gets to at least the zener
    voltage (20V) to turn off Q2, there isn't much draw off of the L2 coil,
    is there a way to estimate what inductance to use, since the simulator
    won't run so I can't trial and error it like usual!

    Here are the current schematic and ltspice files:

    I guess a simpler solution could involve using a PTC thermistor, but
    a restart would have to wait for the resistor to cool down.

    Also if there was an optocoupler that is normally on, then turns off
    when the voltage on the input increases that may work too. I don't
    know if they have optocouplers that work this way? Maybe one built
    with a depletion mode n channel fet would work?
    I want to put the circuit into my SMPS, it is to generate a 12V rail
    to run the uC etc.

  2. Jamie Morken

    Jamie Morken Guest


    By changing C5 to 20uF the simulation runs, it prevents Vcc from going
    below a threshold which was causing the simulator to stall - thanks
    Helmut from ltspice yahoo group :)

    Q2 is still not shutting off though, the zener voltage on Q2's base is
    staying higher than the emitter voltage so that is keeping it turned

  3. Tom Bruhns

    Tom Bruhns Guest

    OK about the part number. I'm having a hard time keeping up. ;-)
    I've trimmed a whole lot the last post off to keep from using quite so
    much net bandwidth...
    Why are you using a 20 volt zener?? I think you'll find that a 10
    volt zener will have a sharp enough knee that it will work well with
    just a few microamps. It's been years since I looked carefully at
    one, though, so I could be wrong. Surely you don't need 20 volts into
    the controller chip to get it going.

    Re-read what I wrote about the flyback voltages. If you have turns
    ratios pri:sec1:sec2, the voltages across the windings (assuming
    perfect coupling) will be always in that ratio. During flyback, the
    output side goes to the regulated output voltage (or gets there after
    charging the cap, anyway), plus a diode drop. That tells you the
    flyback voltage on the other windings. And inductance ratios are the
    square of turns ratios: with perfect coupling of the turns, the
    inductance goes as the square of the number of turns.

  4. Tom Bruhns

    Tom Bruhns Guest

    Yes, you are absolutely right. Thanks for noticing that. Either
    Jamie changed the polarity between the time I looked at it and when
    you looked at it, or I had my head in a dark, smelly place when I
    looked at it. Most likely the latter.

  5. Jamie Morken

    Jamie Morken Guest

    I was worried that the high voltage rated fet would need to have a large
    initial gate drive voltage to avoid it overheating, but looking at the
    datasheet of the 900V fet, 2SK3565 it seems to be ok with 7 volt gate

    Also the start-up threshold voltage for the UC3842/LT1246 is 16Volts, so
    the zener has to be higher than that. The UC3843/LT1247 has a 8.4Volt
    start-up threshold, but its minimum operating voltage is only 7.6Volts
    which doesn't give much room for the circuit to start unless the startup
    resistor is small enough to keep the Vcc rail above 7.6Volts during
    startup, still should work though either way I guess.
    Thanks that works now!

    I used this formula, and substituted voltages for turns since they
    are equivalent:

    coil_turns(inductance) = (1 / (fullcoil_turns / coil_turns(turns))^2) *

    coil_turns(inductance) = (1 / (12v / 22v)^2) * 70uH

    Also changed the feedback voltage divider, using 2.5V reference in the
    IC the formulas are:


    R1=4.7k(22v/2.5v - 1)

    R1 = 36.6k
    R2 = 4.7k

    That gives 11.3v regulated output and ~21.3v to power the IC and shut
    off the npn transistor now. Thanks for all your help! :)

  6. Tom Bruhns

    Tom Bruhns Guest

    I'm not sure at this point which IC that is, but if it's the LT1519,
    that's more than the data sheet rating of the part, I believe. It's
    generally a good idea to run ICs well below their maximum rated
    voltage... Maybe it's for that other part; I don't know the ratings
    for that one. Also, a lot of mosfets have a maximum gate voltage of
    20V, so you'd want to be careful about running the driver on a voltage
    that could be higher than the gate voltage rating of some part you may
    pick in the future. Since the mosfet almost certainly won't turn on
    much "harder" with 20V gate drive than with 10V, the extra voltage
    mainly means that you dissipate more power than you need to in the
    gate drive.

  7. Jamie Morken

    Jamie Morken Guest


    I tested the circuit from 20V input to 450V input,
    the circuit starts to operate at about 30V (below
    that it just doesn't start up) and the startup
    time is about 550ms at 30V input. The startup time
    at 450V input is about 18ms, measured by when the
    input switch shuts off. So using an input switch
    series resistor of 20k, this resistor is dissipating
    about 10watts for 18ms, and then 0watts once the
    circuit is started. Is that safe to use a 1206 package
    or even smaller resistor for this?

    Also the fet dissipates about 1.5watts over the whole
    input voltage range, and at 450V input, there are very
    short 700watt power spikes going through the fet. Is
    this ok for this fet to handle?

    It says 5amp continuous current, but 700watt spikes seem high :)

    I think I could increase the switching frequency and/or increase
    the inductance to reduce these spikes but the low voltage operation
    may be effected by this. Right now at 30Volts input, the circuit
    is only outputting about 8.5volts, I would like to increase this
    to at least 10V if possible to do without increasing the heat
    dissipation too much at 450V.

    Are these 3 coil ferrite transformers an off the shelf part?
    I would rather modify the circuit to work with an off the
    shelf transformer than to get one custom wound to my trial
    and error "specifications" :)

  8. JosephKK

    JosephKK Guest

    Tom Bruhns posted to
    It happens to everyone who is honest with themselves. I am pretty
    darn close to making it to membership.
  9. Tom Bruhns

    Tom Bruhns Guest

    Ouch, 1.5 watts dissipation in a part in a supply that's only
    supplying a watt?? Something needs to change there, I think. I don't
    have time to run the simulation and analyze it, but you should
    probably look more at the max current in the MOSFET. If it stays
    within the rating, you should be OK.

    You're asking for a huge input voltage range. Why so large? Most
    commercial switching modules operate over a 2:1 range or a little
    more, and they are designed by people with quite a bit of experience.
    It's not that it can't be done, but you need to make tradeoffs that
    will almost certainly cause you to pick parts that will lower the
    efficiency: you have to handle higher currents at low input voltage,
    and still handle high voltages. Back when you were suggesting 100 to
    400 volts, that seemed at least feasible, but now you're talking about
    going to 30 volts on the low end?

    You MAY be able to find an off-the-shelf transformer, but especially
    since you want such a high ratio of max:min input voltages, the design
    of the transformer is going to be a lot touchier than if you are doing
    a narrow range. I've pretty much always just wound my own for protos,
    and/or contracted with a transformer manufacturer. I notice CoilCraft
    has a few stock switcher transformers, but I have no idea if any will
    be even close to what you need.

    I'm going to need to bow out of this for a while. I hope you're able
    to get to where you want with this, but I suspect you'll need to make
    some compromises to get something workable. I still think for such
    low power that one of the chips designed for use in compact off-line
    switchers, with built-in mosfet switch, will be a good solution. I
    think they've thought through many of the things you've been
    struggling with. EVEN IF you have to put a circuit in front of it to
    boost low input voltages a bit, it will likely be a simpler and more
    efficient solution.

  10. Tom Bruhns

    Tom Bruhns Guest

    ;-) Yes, and if you can't laugh at yourself for having done it, the
    membership is not worth a lot.
  11. Jamie Morken

    Jamie Morken Guest

    That was with a 3ohm load on the 12V output, with a 100ohm load the
    fet dissipation is only 33mW at 30V input, and 486mW at 450Vin.

    Thanks for all the help, just need to source the transformer now!

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