Connect with us

Make an inductor look bigger

Discussion in 'Electronic Design' started by markp, Oct 1, 2003.

Scroll to continue with content
  1. markp

    markp Guest

    Hi All,

    I have an inductor feeding a circuit from Vcc. If I put a chopper on the
    input to the inductor and use the voltage of the output side of the inductor
    to control the input pulse width, can I make the inductor look much larger
    to the rest of the circuit? The frequencies present on the output side are <
    1kHz, so my thoughts were if I were to chop at, say, 22kHz this would work.
    Any thoughts?

  2. Active8

    Active8 Guest

    if i read this correctly, you use the output to decrease the pulse width
    or increase the freq. the higher the freq, the more XL, the lower the
    output voltage, the higher the freq. an FM chirp. it'll run away.

    if lower output V lengthens the pulse width (lowers freq), you'll get
    more output which will raises freq and stability becomes an issue. it
    also defeats your purpose of which i haven't a clue, nor, i surmise, has
    anyone else reading this post, you included, perhaps ;-)

    at any rate, the inductance won't *look* bigger. you *can* however use a
    gyrator to make a big cap look like a big inductor which eliminates a
    lot of problems involeved with inductors like size and EMI.

  3. John Jardine

    John Jardine Guest

    It'll actually make it look *smaller* pro-rata with the pulse width.
    You're making a (sort of) voltage controlled inductor.
    Eg, switch closed 100% of the time L=L
    switch closed 10% of time then L=L/10
  4. A E

    A E Guest

    Yeah, use one of these:
  5. markp

    markp Guest

    No, the other way around. I didn't make my post very clear, I'll try again!

    I'm using a Baxandall oscillator like the one below:

    ============================= T1
    +--///--+--///--+ +--//////////--+
    L1 | W2a | W2b | | W1 |
    5Vdc--+--///------|-------+ | | |
    | | | +------||------+
    +---/\/\----|-------+ | Cload
    R1 | | |
    \| | |/
    e/| W3a W3b |\e
    | |

    I'm trying to reduce the physical size of L1, but doing that
    increases the winding resistance and decreases effeciency.

    The voltage at the W2a-W2b side of L1 is a full wave rectified
    sinewave. Current will stabilize (i.e. not increase or decrease)
    through L1 when the average voltage of this rectified signal
    equals Vcc (ignoring resistive losses). My thoughts were that if
    the Vcc side of L1 were chopped with FETs, then I could provide
    an average voltage on that side that 'follows' the voltage on
    the other. This could be done using a PWM whose mark/space was
    dependent on the instantaneous voltage on the W2a-W2b side. I
    think in this scenario I would need to make the gain of this
    slightly less than 1 and provide a minimum PWM value, but it
    looks like the circuit will see an inductance in L1 that has
    been amplified, rather like a bootstrapped opamp amplifies the
    input resistance. This is because the rate of change of current
    through L1 depends on the voltage difference, and the voltage
    difference will be smaller due to the PWM feedback, therefore
    the system behaves as if the inductance L1 has increased.

    Any thoughts?

  6. Active8

    Active8 Guest

    out of curiosity, how big is it? what values and freq are we talking
    that actually makes sense. i'm trying to think of another way. you
    *could* just control the dc voltage with a series pass, but we're back
    to lowered efficiency again.

    i think you should try it. especially if you have the parts on hand. you
    could control the voltage with a switcher chip and use an external FET
    if you need to. that way you'd have the error amp and all that in one
    chip rather than a bunch of discrete circuitry.

    let us know what happens.

  7. markp

    markp Guest

    L1 needs to be around 500mH, and the fundamental frequency of the full wave
    rectified waveform is around 1kHz (i.e. 500Hz sine wave).
    Yes, I thought maybe of using a gyrator, but to do that may require a
    chopping amplifier of some type so that the voltage drop is not resistive. I
    still think that idea has merit, although I've not seen a circuit to do it.
    OK, I'm going to implement something like this but it may take several
    months before I get an answer as there are other parts to this design that
    need doing first. I'll let you all know when I do!


  8. markp

    markp Guest

  9. Active8

    Active8 Guest

    i was really wondering about physical size, but inductance, too.
    that's what had me stumped. never saw it before, but the explaination
    seems to hold a little water. i always wonder whether something has been
    thought of before, assume it has been in a lot of cases, and just didn't
    pan out for whatever reason. That must be why the gyrator is a classic.
    but that resistance is there and it's significant.
    oh, ok. just run off for a few months and come back whenever ;-)

    if i get bored, i'll play with that circuit and have the answer.

    have fun,
  10. Bill Sloman has also used the Baxandall osc, but afair he used
    a transistor constant current source instead of L1, (presumably
    with a voltage compliance equal to pi*Vcc/2).

    For your circuit maybe a 5:10v charge pump ic to get enough
    headroom for a const-I source with 8v compliance. Efficiency
    would go down, but the thing could be much smaller.

    ISTR that Bill has also muttered something in the past about
    actively shaping the feed current, so I'll cc a copy of this
    post to him in case he has not seen this thread.

    BTW: In the other thread you were right about using an extra
    resonating capacitor. The EL's load of 1k in series with 50nF
    takes it's inherent Q down too low and the only way to raise
    the Q is to have more VA circulating in the resonant tank.
    This also has the fortunate effect of requiring a lower
    inductance in the tank.
  11. Bill Sloman

    Bill Sloman Guest

    I did use a transistor current source, but it wasn't sourcing a
    constant current, but rather something approximating a rectified sine
    wave, which gave me a much nicer sine wave.
    My circuit used a different approach - the source current waveform was
    derived from the voltage at the centre-tap of the transformer primary
    via a variable gain current mirror, and the gain of the current mirror
    was set by a feedback loop to stabilise the output voltage from the
    oscillator at a level that gave the transistor current source enough
    headroom to operate.

    That is, the peak voltage at the centre tap was held to several volts
    below Vcc, rather than being allowed to rise to pi*Vcc/2. I was
    actually controlling the rectified voltage detected on an auxilliary
    winding of the LVDT that the oscillator was driving, so the peak
    voltage at the centre tap could vary a little as the LVDT warmed up.

    Obviously, such a circuit can't start up,so the current source always
    sourced a certain low minimum current - the "rectified sine wave" of
    current sat on a low pedestal - but the whole thing worked very well.
    It formed part of an up-grade of the Metals Reasearch Gallium Arsenide
    crystal puller which - back in those days - was the work-horse of the
    single-crystal gallium arsenide business, producing some 95% of the
    single crystal GaAs boules made in the west.

  12. markp

    markp Guest

    Thanks, this is an option. If I use FETs instead of transistors I might even
    be able to not use a voltage doubler either.
    Yes, I've seen some of his posts. His approach was very interesting,
    basically getting rid of harmonics at source by providing a sinusoidal
    current source feed in phase with the voltage.

    Yes, this might be necessary, as long as the extra VA doesn't increase the
    core size. I calculated that at 400Hz the Q (i.e. wL/R) is about 7.2 for
    this circuit, so as long as the resistance of the output inductance is kept
    down the Q should be OK.

  13. [/QUOTE]
    It could be interesting to explore the possibilities of
    an even more drastic move away from a pure Baxandall.

    Baxandall put L1 in there to obtain a constant-current drive
    into a parallel tuned circuit, with low-loss synchronised
    current steering. But if there was some efficient method of
    providing a sinewave voltage across W2A+W2b then L1 would not
    be needed. Baxandall didn't have an easy method but you have,
    in the shape of your PWM drive to those switching transistors.

    ========================================== T1
    <---///-+ +--///--+--///--+<k>+--//////////--+
    F/B W3 | | W2a | W2b | | W1 |
    <-------+ | | | | |
    | | | +---/\/\--||---+
    | | | 1k 50nF
    5Vdc--------|-------+ |
    \| |/
    Tr1|--+ +--|Tr2
    e/| | | |\e
    | | | |
    | |
    Sinewave voltage-driving a transformer with a resonant
    secondary is a standard circuit that is detailed in many
    rf textbooks.

    Tr1/2 are switched to provide a synchronised 'sinewave' voltage
    drive to W2a+W2b, either a full PWM, or a 3-step quasi sinewave.
    W3 could be used to provide the synchronising signal to the PWM
    generator and possibly as a measure of the voltage across W1.

    W1 is still that 2.8H and the W2:W1 turns ratio would be about
    1+1:10. Note though the 'k' added in between W2 and W1. It is
    there as a reminder that the coupling has to be factored-in to
    the design. This is also the usual practice for such circuits
    in rf textbooks.

    Worth looking at?
  14. markp

    markp Guest

    Very interesting. This would, as you hinted, require a feedback circuit to
    vary the 'amplitude' of this sinewave PWM signal so as to regulate the
    output voltage. There are also going to be startup problems to solve. It is
    actually safe to switch these transistors on and off quickly like this?

  15. Bill Sloman

    Bill Sloman Guest

    Provided that the current in any inductor in series with the
    transistor has someplace else to go,it is actually safest to switch
    the transistor off as fast as you can. There is a lot more power
    dissipated in the transistor during switch-off than when it is
    carrying high-current in the saturated, fully-on state.

    Using the transistor to snub the energy stored in the stray inductance
    of a coil is not a good idea, as the voltages generated can easily
    exceed the break-down voltage of the transistor.
  16. I think one of the problems of PWM'ing a resonant network could
    perhaps be in knowing what the network looks like to the PWM
    frequencies. If the PWM frequency is high enough (10KHz+ ?) then
    the network will probably look inductive to the transistors.
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