Connect with us

SMPS frequency responce measurement

Discussion in 'Electronic Design' started by Michael, Oct 29, 2003.

Scroll to continue with content
  1. Michael

    Michael Guest

    I need to measure switching power supply frequency responce (or at
    least evaluate it) to get an idea which values need to be used for
    compensation network.
    I am planning on usind this method: 02 feature.pdf
    without network analyser. I will try to get by using high end sound
    card with the CoolEdit s/w.
    Here is the question I have: what do I need for the transformer? What
    are the caveats I need to look for when choosing one?
    I ordered MT4149-ND from Digi-Key (1k-50Ohm) transformer.
    Any clue will be greatly appreciated.
  2. You should be able to use any transformer you have handy. The measurement is
    setup so the transformer does not enter the equation. One thing to remember
    though, is that the impedance of the xformer needs to be low compared to the
    node it is applied to. That is a foundation when testing loops with the
    gain/phase and testgenerator insertion method. Also you need a source to be
    powerfull enough to drive the xformer in order to have sufficient drive....


  3. I like time domain measurements of control loops better than
    sinusoidal measurements. I often make a load bank with selectable
    resistors for the minimum load, and connect other load resistors to
    the supply with a big fet that is switched on and off at a rate low
    enough for the loop response to damp out after each transient. Any
    frequency peaking in the loop response shows up as a ring (damped
    sinusoidal response) following each of the transients.

    It is important to perform this test at various load conditions,
    because SMPS loop response is often far from linear.
  4. Paul Mathews

    Paul Mathews Guest

    It's true that many types of transformers can be used for frequency
    response measurements, however, you can expect to encounter problems
    relating to the dynamic range required for switchmode power supplies
    and the large amount of noise present at the output of the switcher.
    As Dr. Ridley correctly points out, you need to measure the power
    supply output ONLY at the sweep frequency, and this might be
    microvolts in the presence of volts of ripple. Good luck using a
    sound card. Let us know how it goes.

    In theory, you do a calibration run with the transformer alone in the
    loop. You use the results of this run to correct for the frequency
    and phase response of the transformer. However, particularly at the
    low end of the sweep, you'll find that it can take a fairly large core
    and/or lots of turns to prevent core saturation. At the high end, the
    large core and those turns diminish sensitivity where you most need

    Having said all that, I use a 2" diam Siemens toroid with 100 turns
    bifilar wound. It's good from ~70Hz to 10MHz.

    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