Connect with us

air core transformer efficiency

Discussion in 'Electronic Design' started by Fuxue Jin, Oct 7, 2003.

Scroll to continue with content
  1. manifold_1

    manifold_1 Guest

    How fast does the strong magnetic field change? You may get some
    unexpected behavior if the surrounding magnetic circuit induces
    current in the air core transformer. It may jump out of it's fasteners
    or induced high voltages may cause damage to other components in the
    circuit.

    I think the torriodal design would help solve the problem; induced
    currents in one direction should be cancelled out by induced currents
    in the other direction. Oh, unless the ambient field is normal to the
    plane of the toroid, then it looks like one big loop.

    It will not be easy to wind and get a 5KV rating. A transformer house
    may be able to help you with a nice design that will meet the 5KV
    rating and save your hands in the process. Proto rates used to be a
    few hundred dollars for 2 to five pieces. Get the design worked out
    first and then shop the design around and talk to them about what you
    are doing. They are usually very knowledgable about designs.

    There are products that use piezo electric crystals to transfer power
    with very high isolation voltage ratings. They are primarily for
    isolation of 20kV and higher. The efficiency is not so good but they
    are immune to magnetic effects.
     
  2. Tom Bruhns

    Tom Bruhns Guest

    ....
    If you think of it as an RF coupling device using a resonant tank,
    it's quite easy to drive with a single-ended class C amplifier at high
    efficiency. Once you think of things in those terms, you realize that
    people have been using single-ended amplifiers at power levels to tens
    of kilowatts, and often at plate efficiencies around 80%, for many
    decades. You can use the same principles at the milliwatt level if
    you wish. Careful design with solid state devices should get you to
    modestly better efficiencies.

    Cheers,
    Tom
     
  3. Yes, I was browsing through an old book last week,
    "Electron Tube Circuits", by Samuel Seely. As you
    say, power transfer efficiencies (across the tank)
    of around 80% were routinely achieved.

    The interesting twist described by Seely is the use
    of a double-tuned tank.

    +-----+-----+ +------||---+
    | | M | C2 |
    | )|||( |
    | Lp)|||(Ls |
    | )|||( \
    C1=== | | /Rload
    | \ \ \
    | Rp/ /Rs |
    | \ \ |
    | | | |
    +-----+-----+ +-----------+

    The primary is parallel-resonated and the secondary
    is series-resonated. In this way the mutual inductance
    (M) reflects Rload (plus Rs) back into the primary as a
    pure resistance (effectively in series with Rp). The
    driver sees an overall resistive load.

    Seely gives an estimate of the overall power transfer
    efficiency in terms of the pri/sec loaded/unloaded Q's.

    Transfer Efficiency = (1 - Qlp/Qup)(1 - Qls/Qus).

    Where Qlp = unloaded primary Q, etc.

    The loaded Q's, Qlp and Qls, cannot be less than about
    10 to 20 for proper harmonic suppression. So if the
    ratio of both loaded/unloaded Q's is about 1/10, the
    the transfer efficiency is about 0.9x0.9.... 81%.

    Note though that that circuit has to be driven with current
    pulses, presumably optimised for operation with tubes. It
    might be useful to speculate on series-resonating the primary
    as well, for voltage-driven operation.
     
  4. Boris Mohar

    Boris Mohar Guest

    Can you wind another set of turns to buck out the external magnetic field?

    --

    Regards,

    Boris Mohar

    Got Knock? - see:
    Viatrack Printed Circuit Designs http://www3.sympatico.ca/borism/
    Aurora, Ontario
     
  5. R.Legg

    R.Legg Guest

    A two-transistor forward, or a forward,flyback or asymetrical circuit
    with a switched snubber also recover magnetizing energy. The latter
    versions can recover this energy to the output, saving you the trouble
    of reprocessing it.

    RL
     
  6. Tom Bruhns

    Tom Bruhns Guest

    Thanks for the additional clarification, Tony!

    Yes, that circuit is exactly what's used in a "link-coupled"
    transmitter output. And if you have room for a coil 1 inch in
    diameter and an inch or so long, at 20MHz it can have an unloaded Q in
    excess of 400. It's not clear to me that harmonic suppression is
    important in this application, so the loaded Q might even be on the
    low side of 10, but 10 is a good figure to shoot for. Then you can
    have a tank circuit efficiency around 95%. The broadcast transmitter
    I worked with back in the 60's had an overall plate circuit efficiency
    of nearly 80%, and that was DC plate input power to power fed to the
    transmission line, so that efficiency includes substantial plate
    dissipation in the tubes. It was a push-pull primary version of the
    same circuit.

    For the OP: if you go this way, then you will maximize the efficiency
    by maximizing the ratio of Qu/Ql. Qu increases with frequency for a
    given coil size, and with coil size. For a coil that's about as long
    as its diameter, a useful first estimate of Qu is
    100*diam(inches)*sqrt(f(MHz)). Qu is lowered by metal in the area:
    best to keep the coil in a volume with a coil-diameter's space on all
    sides. You get to do some circuit analysis to determine Ql, but in
    general the heavier your loading, the lower Ql.

    Cheers,
    Tom
     
  7. A high L/C ratio is usually associated with high
    power transfers.

    Tom, do you know if there is an optimum format for a
    a resonant air-cored transformer?

    eg, Is it solenoids alongside, solenoids within each
    other, or something?

    My instincts would be toroidal, but the book dismisses
    toroidal resonant inductors as "having excessive winding
    resistance".
     
  8. Tony Williams wrote...
    Do you remember my experiments with large litz-wire
    Murgatroyd D inductors? Toroidal air-coil inductors
    also suffer from high proximity-effect losses at the
    inner region where the windings compress together.

    Thanks,
    - Win

    whill_at_picovolt-dot-com
     
  9. Only vaguely. What I remember most about that thread
    is not really understanding it.
     
  10. John Larkin

    John Larkin Guest


    A torroidal air-core coax-wound transmission line transformer would be
    fun. Zero leakage inductance, high Q, zero external field, excellent
    insulation.

    John
     
  11. R.Legg

    R.Legg Guest

    My instincts would be toroidal, but the book dismisses
    This averages out. I'd be interested to know what the book reference
    actually says - whether there are practical references cited.

    Toroids are avoided usually in any commercial application because they
    are difficult to fabricate in a controlled manner. They are more
    common in accellerator designs, where magnetic field problems are
    unique in amplitude - this application is in itself somewhat obscure.
    They are also more common in military and space environments. (or
    TEMPEST devices that are difficult to enclose for one reason or
    another)

    I'm still curious as to how this circuit is intended to be applied
    without infecting the system with EMI.

    The 'large magnetic field' is likely present precisely because it is
    needed to measure very small EM effects. Introducing extraneous noise
    is likely to be detrimental.

    RL
     
  12. Tom Bruhns

    Tom Bruhns Guest

    Hi Tony (and others),

    I think it depends on what you mean by "optimum." In things I've
    dealt with, it hasn't been an issue most of the time. That is, losses
    elsewhere dominate, and there has generally been plenty of room for
    large inductors. Of course there are practical issues, too: what's
    optimal from a performance standpoint may not be from a construction
    standpoint. I do note that if you couple the coils very closely, they
    no longer tune separately. I haven't analyzed things in detail, but I
    guess with perfect coupling, they behave like one inductor.

    When I worked on quadrupole mass spectrometers, I used to visit with
    the fellow who was doing the RF drive work. We wanted high Q; as I
    recall, we got something on the order of 300 at 1.00MHz with coils
    that were about 1 inch diameter, by placing two coils with axes
    parallel and separated by perhaps 1.25 inches. Each was roughly an
    inch long. They were connected in series to form one inductor. The
    shield around them was more fun: the engineer measured the direction
    of the field from the coils, and cut slots in the shield oriented to
    break up the current. It seems to me that makes the shield less
    effective, but I wasn't going to tell him how to do his job, and the
    proof is in how it all works.

    A high L/C ratio presumably means low loaded Q, if the load is shunt.
    For a series load, it would be a high C/L ratio for low loaded Q.
    Guess I just think in terms of Qu and Ql and let the L and C fall
    where they may...and if something is impractical, I look for ways to
    do impedance transformations or for other topologies.

    Cheers,
    Tom
     
  13. R.Legg

    R.Legg Guest

    But very large, when using 5Kv hipot-able coax.

    Biomedical-safety-grade toroids can be achieved using simple
    three-layer film wrap between primary and secondary, providing the
    lead-outs are suitably dressed.

    RL
     
  14. R.Legg

    R.Legg Guest

    Stumbled across two references to efficiency of the solenoid shape.
    The first looks at an transformer application around 100W, but I think
    they are using impractical winding methods for a 5KV hipot. The second
    is only a thermal evaluation, and there seems to be no reference to
    layering.

    "Air-Core Transformer for High Frequency Power Conversion"
    K. W. E. Cheng, H. L. Chan and D. Sutanto
    EPE'99

    http://manuales.elo.utfsm.cl/conferences/seminarios/EPFL/pc/papers/338.pdf

    "ANALYTIC THERMAL MODELLING of an AIR - CORED COIL"
    C.Coillot, Y.Patin, F.Forest, P.Chantrenne
    EPE'99

    http://manuales.elo.utfsm.cl/conferences/seminarios/EPFL/pc/papers/471.pdf

    RL
     

  15. Thanks for posting this link R. Legg. It makes for a very interesting read.

    Unfortunately the authors don't have a clue what they are jabbering about.
    I'm sure it dazzles most readers in BS however, and probably fullfilled
    their university's "research" requirements for its professors.

    There are numerous very serious problems with it, but something that really
    caught my eye was their choice of Schottky diodes for the circuit of figure
    8. They use the MBR1045 which is a 10A rated 45V schottky diode. Then they
    say a couples lines under figure 8 they have an input voltage of 60V and a
    19:19 (or 1:1) turns ratio on their transformer. This means their diodes
    get to see an abusive 60V or so during some parts of the cycle. I wonder if
    they blasted a whole bunch of those MBR1045 (and the IRF530N devices) before
    they finally found a set that didn't avalanche to destruction immediately
    and could block the 60V.
     
  16. R.Legg

    R.Legg Guest

    Thanks for posting this link R. Legg. It makes for a very interesting read.
    As with most articles, it has to be taken with a grain of salt. You
    will often find that there are problems with an author's approach to a
    particular application. I can only suggest that you make these
    observations known to the author. There are many reasons why this
    paper may have been presented at EPE'99 in it's present state that
    have nothing to do with actual research performed or results and
    conclusions obtained.

    It may be a poorly cribbed summary of more extensive work, which, if
    you do query results directly, may be offered in response.
    Unfortunately, a time lapse of four years can be a serious obstruction
    to such communications, as authors move on.
    The topology is series primary resonant, with a parallel-loaded
    series-resonant secondary rectifier. Depending on the ratios of Llk
    (they indicate that added series inductance was introduced) to Lout.
    The primary series inductor Llk may have dominated, in which case the
    output voltage will have effectively clamped the voltage across the
    rectifiers (Vout=24V) in spite of the presence of 'Lout'. The authors
    don't have to tell you this, for their research to be valid. They
    don't have know that this is what saved their devices from failing -
    if this was in fact the case.

    "Llk and Rlk are the sum of primary and secondary leakage inductances
    and resistances and additional series resonant inductor respectively"

    We also know, from their explicit statement, that magnetizing
    inductance values were low - I expect that magnetizing Lm should more
    truthfully have been drawn in the circuit, producing a further
    effective input voltage reduction.
    It is not their duty to report how many parts popped while collecting
    data, if this was indeed the case.

    What struck me as the biggest limitation in this presentation, was
    that this device was not compared to a part wound without
    superconductive media. Obviously, the windings of a normal device
    would not have had a 25DegC impedance of 180Ohms or greater, as this
    device did.

    The real proof of the value of the material change would be to test
    the superconductive and normal parts at both cryogenic and normal
    temperatures. This would give a figure of merit to:

    a) cryogenics alone

    b)superconductive winding media alone.

    At a minimum, it would have produced data that others could refer to,
    when using normal conductors at normal temperatures, which must be
    considered as a kind of baseline.

    Perhaps this data was collected. In this event the shortcoming is in
    the paper alone - not in the authors or the research. Keep them
    separate in your mind, to maintain faith in humin beens and your own
    sanity.

    RL
     
  17. There are sums in other books (related more to if transformers)
    that optimise the coupling factor for highest current in the
    secondary tank.

    [snip, thank you Tom]
    I suppose that, if the load and loaded-Q are both defined,
    then L and C are also automatically defined.
     
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

-