Connect with us

Why push/pull smps always have center tapped secondary ?

Discussion in 'Electronic Design' started by Antonio Pasini, Oct 14, 2003.

Scroll to continue with content
  1. I'm a newbie trying to learn fundamentals of switching design...

    I noticed from many sources that all push/pull diagrams uses a center tapped
    secondary.
    Nobody clearly explains why.

    I'd like to understand why a center tapped is necessary.

    Apart from lower efficiency due to doubling the diode drops, why can't I use
    a single-winding secondary followed by a full-bridge rectifier ?

    Flux balance is often guaranteed by current-control; why then the center tap
    ?


    For example, from Unitrode DN-63, by L. Balogh (from TI website) I read:

    "Figure 1 illustrates the commonly used circuit arrangement

    for full-wave rectification. For proper operation

    the secondary winding has to be

    center-tapped with one terminal connected to the

    reference potential (ground) of the circuit. The center-

    tapping splits the secondary winding into two inductors

    which are coupled strongly but not perfectly

    within one magnetic structure."



    Why ???
     
  2. CBarn24050

    CBarn24050 Guest

    you can use a single winding with a bridge rectifier. You need 4 diodes instead
    of 2, you double the conduction losses in the rectifers. During the reverse
    recovery time you have a short circuit across the winding which leads to large
    current pulses.
     
  3. Genome

    Genome Guest

    You mention about two diode drops affecting efficiency another one would be
    cost considerations. Otherwise there doesn't seem to be a reason why you
    cannot do what you are suggesting.

    What you have to consider is what's happening over the complete switching
    cycle, in particular during the time when the primary side switches are off
    and the output filter inductor current is 'free-wheeling'.

    In the case of a center tapped secondary with two diodes feeding the
    inductor with one of the primary switches on inductor current is flowing
    through a forward biased diode with the other diode reverse biased and
    non-conducting.

    When the primary switches are both off inductor current continues to flow
    and the voltage at that node in the circuit falls until both diodes begin to
    conduct and share the inductor current. This current is now flowing in the
    two center tapped secondaries.

    The important thing here is that the current flow is 'positive' out of both
    windings but the windings are in antiphase. As a result, because the
    windings are tightly coupled, there is effectively zero flux in the core and
    the windings behave as a short circuit to the center tap which is the return
    path for load current.

    You implement a full bridge feeding the output inductor and load from a
    single secondary winding. Consider the same condition when the primary side
    switches are off. Originally inductor current was flowing through one pair
    of diodes in opposite legs of the bridge.

    When the switches turn off inductor current flows in both sets of diodes
    with each pair of diodes in the same leg of the bridge being effectively
    connected in parallel. The transformer secondary ends up with zero volts
    across it and current no longer flows in the winding.

    On the basis of that description your proposal works. Wether the following
    is true is a different matter.....

    The disadvantage appears to be added cost and power loss.

    However;

    1) The transformer construction is simplified. One less winding and
    termination to deal with. That should make it cheaper.
    2) It would seem that the number of turns required is halved. Copper losses
    would be expected to be halved as well.
    3) Furthermore, proximity effect in a center tapped secondary is worsened
    because the non conducting secondary acts as a passive layer which still
    suffers corresponding losses. The single winding will be more efficient.
    4) Additionally the winding is not active during the free-wheeling phase
    which will again save power.
    5) You might also claim that primary to secondary coupling is improved but
    that might be alleviated in the center tapped case by proper winding order.
    6) Voltage stress on the diodes is reduced from 2VOUT to VOUT allowing lower
    voltage, more efficient, lower cost devices.

    The problem you have is arguing whether the above points are sufficient to
    justify the implementation. The balance may well be tipped when considering
    high voltage outputs.

    Once again the above assumes I've thought it through properly.

    Oh, a quick one about flux balance. This generally relates to the primary
    magnetising inductance. Bear in mind that your primary is a coil of wire on
    a core and is therefore an inductor, the magnetising inductance. In circuit
    it appears in parallel with the reflected output filter inductance. Current
    mode control maintains a zero net magnetising current as long as both
    magnetising and reflected output filter inductor current is sensed.

    There are some topologies where this isn't the case but people still use
    them and then wonder why things blow up.

    Good luck in your new journey of discovery. If you are lucky you'll avoid
    turning into an obnoxious alcoholic like me.

    DNA
     
  4. R.Legg

    R.Legg Guest

    It's more common in low voltage circuits or circuits with split supply
    rails.

    Obviously a push-pull or full wave source will work most comfortably
    if the energy from both cycles is processed in a similarly efficient
    manner.

    A single winding could also be used in a full wave voltage doubler.
    Why what?

    The output current in the topology mentioned uses one output inductor.
    The switches therefore have to provide this same reflected inductor
    current when 'on' as inductor winding current can not suddenly jump
    from one value to another, for successive switch closures, without
    difference energy difference going somewhere.

    When this current has to switch from one winding to another in the
    transformer (also an 'inductor'), or reverse direction in the same
    winding - the problem of this sudden energy change is experienced.
    Because it's not a perfect world, the transfer is not immediate or
    perfect due to practical coupling imperfections.

    The article you quote also goes on to list an alternative Hybridge
    current-doubler circuit (a la 1914). This requires no centertap, but
    uses two output inductors. The current in these inductors is
    independent, with one switch current reflected in one inductor and the
    other switch current reflected in the other inductor.

    Basically if a circuit does what it needs to do to justify it's
    application, reliably and repeatably, under budget and within the box
    size, wherever and whenever it's supposed to, it's doing all that is
    'neccessary'.

    Keep reading. The TI Unitrode articles discuss leakage inductance
    elsewhere.

    RL
     
  5. default

    default Guest

    If you mean why do Push Pull AMPS (smp?) have center tapped power
    supply transformers . . .

    If you are talking about TUBE P/P amplifiers it is because it is easy
    to put two diode rectifiers in a single glass envelope with a common
    cathode. (the way most tube amps were originally designed)

    Four diodes would take three or four isolated filament windings.

    With low voltage high current supplies (for semiconductor use) it
    eliminates a source of waste power. Most CT transformers (but not
    all) are bifilar wound (wound with two wires and then connected one
    start to one finish to provide the CT - perfectly balanced that way).
     
  6. Thanks a lot to all of you for your clear explanations.

    Your answers make me more comfortable.

    I need to draw more current from the auxiliary winding of a push/pull
    design, and I have no more pins to add a center tap.
    Instead, I have no efficiency problems.
    So, using a full bridge (drawing power in both cycles) will keep the
    transformer more balanced, I think.

    Thank you for your suggestions!
     
  7. Genome

    Genome Guest

    Wherein lies the nastiness of the schmelly poo . With a leaky bit sitting
    about doing nothing that is suddenly asked to do something by a bigger bit
    you end up with lots of volty bits all over the other bits, and sometimes
    otherwise. Other thingies get redirected to the other places and there is
    much complicatedness going on. All of which affects the reliablenessness
    unless some added bits are judy applied to make the problem invisible smoke
    for no reasonable.

    It's Twu, It's Twu, It's Twu.

    I would personally recommend....

    "An Introductory Guide to 'Cultural Theory and Popular Culture" by John
    Storey.

    ISBN 0-7450-1316-3 (hard back)
    ISBN 0-7450-1317-1 (paper back)

    AKA "A personal insight to the 'thinking' mind."

    Now let me psychoanalize you. And while we're at it I'd like to know why I
    told you how to design the ideal video recorder/human interface and I still
    can't use the fucking thing.

    Woop wooop, I'm a teapot.

    DNA
     
  8. Paul Mathews

    Paul Mathews Guest

    It's worth mentioning that flux balancing is not actually "guaranteed
    by current control", and there are a number of reasons why. Current
    mode control is usually implemented by feedback, and there are many
    practical problems with feedback loops. For example, during both
    startup and overshoot conditions, feedback can be very non-linear, as
    various clamps limit the range of control. During such conditions,
    cycle-by-cycle current control does not actually occur, and
    substantial flux-walking can occur. This is why some designs oversize
    the transformer, include an air gap, add a series capacitor, and other
    remedies. Another problem with relying on on CM control to prevent
    flux walking is the consequent need for a DC-coupled current sensing
    device. Look for a recent related thread that covers this subject
    more thoroughly.
    Paul Mathews
     
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

-