Connect with us

3V->15V 15W switcher topology?

Discussion in 'Electronic Design' started by Joel Kolstad, Jun 22, 2007.

Scroll to continue with content
  1. Joel Kolstad

    Joel Kolstad Guest

    I'm contemplating building a step-up converter from about 3V (it's a lithium
    cell and realistically it's almost completely discharged at 3V) to something a
    little shy of 15V, where I need 15W plus a few more to cover losses. Starting
    with such a low voltage obviously requires some pretty hefty currents, so I'm
    thinking that a switcher topology of either a half-forward converter (nice,
    simple transformer and low parts count -- peak switch current around 12A) or a
    push-pull topology (peak switch currents around 6A, but need a center-tapped
    transformer, and controller ICs become quite uncommon given the very low input
    voltage since most push-pull applications seem to be in the area of much
    higher power). Flybacks or a straight boost converter seem highly undesirable
    due to peak currents in the ballpark of 27.5A -- ouch! Since this is battery
    powered, I care about efficiency, but I do have plenty of room, so I figured
    I'd just use a low (couple hundred kHz) switching frequency and big magnetics
    to achieve that end.

    Does this sound like a decent strategy?

    Also... anyone know of off-the-shelf modules that can perform this job? TI
    has one that's close -- -- but it's only
    12W output power. (It is amazingly small!) This is a low-volume application,
    so price isn't a huge concern.

    ---Joel Kolstad
  2. Tim Wescott

    Tim Wescott Guest

    I'm not entirely clear on what you mean by "half forward converter", but
    you should avoid conduction diodes on the 3V side, they'll eat your
    efficiency for lunch and dinner.

    Given that, a controller that will implement any two-transistor solution
    will probably be suitable for a push-pull with two N-channel FETs, so
    why not go that way from the start? If you can't find a controller that
    will work adequately from 3V, think about using something light-duty and
    cheesy to make the rail for the controller (Jorge should chime in here
    with how to do it with a '2222 and less than four passives), or ask
    yourself if you can get by with serious inefficiencies and misbehavior
    at startup as long as the controller works well enough to get the rail
    up so that it can do _something_ right.


    Tim Wescott
    Wescott Design Services

    Do you need to implement control loops in software?
    "Applied Control Theory for Embedded Systems" gives you just what it says.
    See details at
  3. colin

    colin Guest

    I dont know how you arrive at >27amps for a boost, my quick calcs show a lot
    less than this,
    more like just over 10 amps, and Low voltage mosfets have amazingly low
    rdson in small devices.

    or consider a 2 phase boost.
    im not sure what controllers go down to 3v though,
    but maybe tghe battery has a little more volts at switch on enough to
    operate a more avialable controller long enough to boot strap itself before
    the current draw makes it drop ?

    Colin =^.^=
  4. Nico Coesel

    Nico Coesel Guest

    I'd use a microcontroller to implement the controller in software. In
    this way you have a charge and discharge controller almost for free.
    Lithium is very sensitive to overcharge and overdischarge. I recently
    saw what happened with a big lithium battery being overcharged: it
    simply becomes a blow torche.
  5. Joel Kolstad

    Joel Kolstad Guest

    Hi Colin,

    I'm using peak current ~= 5.5*Pout/Vin from Marty Brown's book.
    There are a surprising number these days, actually!

  6. Joel Kolstad

    Joel Kolstad Guest

    Hi Nico,

    Yeah, but my time isn't free. :) I like the idea though; thanks.

  7. colin

    colin Guest

    I dont know that book myself, that sounds far too generalized,
    but :-
    if Power transfered via inductor =(vout-vin)*Iout = 12W
    voltage accros transformer = 12v, 3v
    power during on = Power * (12+3)/12 = 15,
    for max on time but still discontinuos
    Ipk = 2 * Pon/vin = 10amps
    yes but Im not so familiar with them.

    Colin =^.^=
  8. kell

    kell Guest

    Look at the very bottom of this circuit by Tim Williams, in the area
    where you see Q14:
    The cap he has across the transformer primary may be a litte fat, you
    could try something smaller.
    For low power you might be able to dispense with the secondary of the
    transformer and just pick pulses off the 20 uH inductor with a
    To regulate it you can use a TL431 to shunt the transistor base to
    ground, with the tl431 feedback pin attached to a voltage divider on
    the output.
  9. kell

    kell Guest

    P.S. I was suggesting this as a low-power supply to drive a chip.
  10. colin

    colin Guest

    this would easily do what you want....,C1,C1003,C1042,P22513,D18284

    a quick sim in ltspice sees peak current of 10-11 amps at 3.3v
    they have an example sim circuit just need to change the feedback resistor,
    and the laod resistor.

    Colin =^.^=
  11. Guest

    Maybe lower - I find that around 30-50 kHz, with modern ferrites, losses are
    "invisible". If you got room.
    I think you want a push-pull inverter at this low voltage to cut down the peak
    currents. How about a current fed Royer oscillator? If you really want, you can
    regulate the secondary with a mag-amp to keep it all discrete ;-).
  12. Nico Coesel

    Nico Coesel Guest

    If you are going to use lithium cells and care about there life
    expectancy, then there are very little other options than checking the
    battery voltage droop versus time (dV/dt) to see if the battery is
    empty. The voltage across the cell may vary a lot with temperature.

    If you are going to use a boost converter you'll need a chip with a
    huge switch capable of handling 30A (16.6% duty cycle, 5A average
    input current). I think a converter with a transformer is a good
    option here to keep the primary current down. Wurth has flex
    transformers with which you can choose the turns ratio by combining 6
    individual windings. Never used them in a circuit, but I will if get
    the chance :)
  13. colin

    colin Guest

    with 3v input and 15v output that coresponds to an on time of 80%
    for edge of discontinuos mode. I ran a sim and found 10.5amps peak for 3.3v
    with continuos mode this is even lower.

    Colin =^.^=
  14. Joel Kolstad

    Joel Kolstad Guest

    Hi Frithof,

    That's a little too fancy for me, although I was fascinated some years ago
    reading Pressman's book and his all-discrete design using mag amps for the
    controller. Quite impressive, and while it's not super-high on the list, it'd
    be cool to duplicate someday. (Speaking of which... does anyone have a link
    to Pressman's obituary? Or even just when he died?)

    I've been thinking of using the LT1950 controller IC for this; it's very much
    designed to go down to 3V using a little built-in boost converter for
    bootstrapping the gate drive up to 8V. Alas, I probably won't get back to
    serious design for another week here.

    Thanks for your input,
  15. Joel Kolstad

    Joel Kolstad Guest

    Hi Nico,

    Well... I'm doing charging with an off-the-shelf TI (formerly Benchmarq) IC.
    In this application the device is supposed to spend the vast bulk of its time
    plugged-in, so other than various watchdog ICs that effectively remove the
    load below about 2.7V, there isn't any discharge monitoring circuitry that
    I've added. The battery itself comes with the usual built-in protection,
    though -- it disconnects itself at too high of a charge rate or too low of a
    voltage (2.5V, it says).
    Yep, agreed.
    Cool, thanks for the name; I'll definitely check them out.

  16. colin

    colin Guest

    has everyone gone crazy, why only work at 16.6% duty cycle?

    Colin =^.^=
  17. Nico Coesel

    Nico Coesel Guest

    You are right. I got the duty cycle the wrong way around, but the
    currents are still huge. During Toff the current through the rectifier
    diode will be 5A which leads to considerable losses even with schottky
    diodes. Synchronous rectification is something to consider to save
  18. Nico Coesel

    Nico Coesel Guest

    Panasonic has some excellent documents on-line on charging and
    discharging Lithium cells. But it seems your battery pack has this
    intelligence build into it.
    See my reply to Colin.
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