Connect with us

Need Equation for PWM

Discussion in 'General Electronics' started by jack, Apr 12, 2005.

Scroll to continue with content
  1. jack

    jack Guest

    Hi,

    I'm not a math or electronics genius but I was wondering if anyone
    could provide an equation that gives amplitude (voltage) as a function
    of time and duty cycle for a square wave:

    symbolically:

    A (V) = f(t (sec),duty cycle (%))

    perhaps also the function would include something about the initial
    phase (phi)......

    A simple equation that just gave back 1 or 0, as a function of time and
    duty cycle would work good too....I presume any answer would involve a
    trig function someway....

    I have already tried searching the web for the answer and the closest I
    found was this link:

    http://www.dattalo.com/technical/theory/sqwave.html

    However, I'm not sure exactly how LaPlace transforms figure into it, or
    if any of these are usable equations for what I'm looking for.

    Please help....if you can.

    Thank you.
     
  2. Google on "Fourier series" plus "square wave". Here's a nice piece
    from Addison-Wesley:
    http://www.aw-bc.com/ide/Media/JavaTools/srfoursq.html
    If you're into electronics, you might want to study the use of
    the Laplace transform in solving second-order ordinary
    differential equations. It is especially handy when the equation
    contains a discontinuous "forcing term", which it frequently
    does in electronics.
    LH
     
  3. Bob Eldred

    Bob Eldred Guest

    I assume you mean: Vave = VpD the average voltage equals the peak voltage
    times the duty cycle. The duty cycle, D = Ton/(Ton + Toff) = Ton/(period) =
    f Ton. The RMS voltage is more complicated.
    Bob
     
  4. I am not quite sure what your question is. The terms *step function* and
    *Heavyside function* are sometimes used to represent pulses. Such a function
    is merely H(t)=1 for t>0 and 0 for t<0. A pulse of width T would be
    represwented by f(t)= H(t)-H(t-T). These are not the nice continuous
    functions with smooth derivatives that old-time mathematicians like, but
    they do have nice Laplace transforms.

    Bill
     
  5. jack

    jack Guest

    Thank you very much for your reply.....but, it didn't anwser my
    question....

    I did the Google search you suggested and I came on this site:

    http://mathworld.wolfram.com/SquareWave.html

    At first glance, this website seems to answer my query but upon further
    inspection it is not any good.

    First, EXCEL has a problem raising negative numbers (in this case -1)
    to a fractional exponent. So the first equation is no good.

    Secondly, it does allow for varied duty cycle.

    Thirdly, the other 2 formulas are not conducive to inclusion in
    EXCEL......

    Also, any "Fourier series" solution to my problem is also not conducive
    to EXCEL since it requires summation of an infinite series.....

    Also, I really don't care about the LaPlace transform of the equation
    I'm looking for, just the actual equation itself.....
     
  6. jack

    jack Guest

    No...I'm not talking about sine waves but square waves....also, I'm not
    talking about Vave but Vinst (e.g.. +1 and -1, or Vhi and Vlo or Vhi
    and 0)...I want to know what Vinst is at any time (time) for a given
    frequency/period, amplitude and duty cycle.....

    see the links I mentioned to get a better idea.

    thank you...
     
  7. jack

    jack Guest

    without using conditional statements (like "IF"), how would you write a
    formula in EXCEL (or a formula that could be translated into EXCEL)
    that would give you amplitude (e.g. 1 or -1) at a particular value of x
    (or time) for a particular frequency/period, and dudty cycle? In other
    words, a formula showing amplitude as a funtion of time, frequency, and
    duty cycle....???
     
  8. Joel Kolstad

    Joel Kolstad Guest

    Jack,

    The 'formula' for a PWM square wave is exactly what you think it ought to
    be; something like:

    T=period
    p=percentage 'on' time
    t=current time
    O=output

    Set t'=t mod T (this gets the time into the current period)
    Set O=1 if t'>=p*T
    else set O=0

    Although this notation wouldn't please a mathematician, it can be 'prettied
    up' such that it if perfectly legitimate. Excel knows the modulus operator
    and if/then/else operators if you want to implement it.

    The stuff about Fourier transforms relates to taking a square wave and
    expressing it as a sum of some other set of 'fundamental' (aka, 'basis')
    functions; sines and cosines turn out to be particularly useful in this
    endeavor, but you can typically express any function as some (often
    infinite) sum of some other set of basis functions so long as those basis
    functions have certain special properties (e.g., if you like, you can
    express sine in terms of bunch of square wave-like functions). There's lot
    of math behind this that you can dig up if you're so inclined.

    Here's a good web page with examples:
    http://cnx.rice.edu/content/m0041/latest/ Note the equation at the bottom
    of the page -- you could use that summation using, e.g., the first 5 terms
    and then checking to see if the result is greater than or less than 0, but
    this is a lot more effort than just using the definition of a PWM square
    wave directly.

    ---Joel Kolstad
     
  9. Joel Kolstad

    Joel Kolstad Guest

    What don't you like about IF/THEN/ELSE?

    In your case, you can cheat a little and use various rounding functions such
    as TRUNC or perhaps CEILING/FLOOR... but realize that all you're doing is
    shifting the conditional statement to being 'hidden' (Excel does it
    internally) rather than your doing it explicitly.

    There is no finite series closed-form expression in terms of "familiar"
    functions that I'm aware of for a square wave. Even if you have an infinite
    number of terms, there's still an error in the approximation of ~9%, if I
    recall correctly, due to something called the Gibbs phoenomenom. You might
    be able to remove it with another infinite series that represents an ideal
    brick wall filter, however, but this is really starting to get silly...

    In signals and systems work, it's common to get around most of these
    functions by _defining_ a function called, e.g., "rect" that defines a
    single rectangular pulse of a given duration.

    ---Joel Kolstad
     
  10. jack

    jack Guest

    Thank you for your response...

    Here's my retort:
    Hard to explain, but basically I want to do further operations on the
    equation which makes the use of IF/THEN/ELSE problematic...
    See Formula 1 of the following link:

    http://mathworld.wolfram.com/SquareWave.html

    Problems with this formula are 1.) that EXCEL will not allow (-1) to be
    raised to a fractional exponent and 2.) duty cycle is not accounted
    for...
     
  11. Bob Eldred

    Bob Eldred Guest


    I've read everybodies repy and your answers to them and clearly you don't
    know what you are asking for and neither do we. If you don't know what you
    are asking, why are you asking it? PWM signals are used to drive motors,
    vary voltages, powers or currents in power supplies, class D amplifiers,
    motor controllers and the like. The relationships I gave you above are what
    you need if you are working on a power supply or amplifier and want to
    relate the output to the duty cycle. You responded with some quip about sine
    waves? Clearly you don't understand what I told you and I have no clue what
    you are asking. Instead of chasing URL's, try to state your problem in
    english, maybe we can help. The URL you refer to is highly mathematical and
    forms the underlying basis for harmonic cancellation to suppress harmonics
    from PWM signals.Is this what you want to put into practice? It seems like a
    big jump from your original question.
    Bob
     
  12. jack

    jack Guest

    O.K., fair enough....let me respond by point:
    Simply stated, I want a mathematical equation which shows
    voltage/amplitude of a square wave (+1 or -1) as a function of time,
    frequency/period, and duty cycle. However, this equation must be
    "neat" - no conditional statements (like "IF"), or piecewise statements
    (like Heaviside), or infinite series (like Fourier). The desire for a
    "neat" equation is for easy incorporating and manipulating in EXCEL.

    If this is not clear enough, then please let me know.

    My "quip" about sine waves was in response to your misinterpreting my
    question as referring to RMS/sine waves as opposed to square waves. You
    had stated:

    Here is my "quip" back to you, making it clear I was talking about
    square waves, not sine waves...I was not confounding the issue, just
    clearing it up:
    I think I do on both accounts. You had responded to me using the
    concepts of RMS, sine waves, and Vave, whereas I am interested in
    square waves and Vinst. I think the record shows your reply was
    off-target, albeit well-intentioned.

    See above.

    Here is the URL again:

    http://mathworld.wolfram.com/SquareWave.html

    I agree this URL eventually becomes "highly mathematical" but Eaquation
    1 is not. Do you believe Equation 1 is "highly mathematical"? If so,
    how so? As previously stated, the formula would work great if 1.)
    EXCEL allowed rasing a negative number to a fractional exponent and 2.)
    duty cycle was included.
     
  13. Joel Kolstad

    Joel Kolstad Guest

    Hi Jack,

    OK, equation (1) is certainly simple enough. :) Equation (2) is probably
    your best bet for implementation in Excel; it has the "signum" functin
    (SIGN).
    With equation (2), let's see... you could account for duty cycling by
    subtracting two sine waves that are out of phase with one another and taking
    the SIGN of the result. The amount that they're out of phase will determine
    the duty cycle (albeit non-linearly) -- 0 degrees out of phase is 0% PWM,
    180 degrees out of phase is 100% PWM.

    ---Joel
     
  14. jack

    jack Guest

    Thank you for your response.

    You're right that equation 2 will "work" in EXCEL although it is still
    not really mathematically "pure" or very "elegant".

    However, I challenge you to try to get Equation 1 to work in EXCEL.
    Raising a negative number to a fractional exponent causes an error in
    EXCEL. Try it. I agree Equation 1 would be great if it worked in
    EXCEL. :-(
     
  15. I read in sci.electronics.design that jack <>
    A negative number to a fractional exponent is a complex number. Excel
    2003 understands them, but I've never used that feature. Earlier
    versions (e.g. Excel 97) didn't understand them.

    I think it is irresponsible to allow accountants to use a spreadsheet
    that can cope with imaginary numbers. (;-)
     
  16. Joel Kolstad

    Joel Kolstad Guest

    Hi Jack,

    I suppose that's in the eye of the beholder?
    Yeah, I see that you're right. This isn't too surprising, though, in that
    typically raising negative numbers to fractional exponents ends up with an
    imaginary result, and Excel doesn't support complex numbers 'natively' (it
    can be done with them with the Analysis ToolPak, but it's still a little
    clunky). In fact, looking at equation (1) on the MathWorld site, I believe
    you have to look strictly at the real part of the result to get the output
    (the square wave) that you're after (and just discard the imaginary part).

    The Windows calculator, regular scientific calculators without native
    complex number support, and many other applications that you might try to
    use with eqn (1) will fail... those guys over at MathWorld aren't bothered
    by this because their product, Mathematica, pretty much never fails. :)
    (Mathematica will produce results from expressions that almost any
    'everyday' calculator like Excel will barf at, since Mathematica 'knows'
    many extensions of 'familiar' functions that make them 'work' in uncommon
    domains that high-powered math guys are familiar with... for instance, the
    factorial function can be extended to produce results even when its argument
    is negative or fractional, even though for the vast majority of people this
    is unexpected behavior.)

    An implementation using equation (2) is going to be a lot more efficient
    than one using eqn (1). An implementation using a boring 'if/then/else'
    approach is going to be the most efficient of all. Often the most 'elegant'
    mathematical expressions are the most inefficient if you perform the
    'straightforward' implementation; this is understandable in that by
    'elegant' one oftens means they're trying to hide lots of ugly details in a
    few nice, all-powerful functions -- which are inherently slow to evaluate.

    ---Joel
     
  17. HorneTD

    HorneTD Guest

    John Woodgate wrote:
    Snip
    Thanks for the laugh.
     
  18. Rich Grise

    Rich Grise Guest

    Do you even know what a "square wave" is?

    Do you have any idea at all why asking for an "equation that gives voltage"
    is stupider than a joke?

    You'd better ask your teacher for remedial instruction on the difference
    between "pressure" and "flow". Or maybe the difference between
    "measuring" and "fantasizing".

    Good Luck!
    Rich
     
  19. Rich Grise

    Rich Grise Guest

    This is the problem. Your problem statement doesn't make any sense.

    There Is NO Equation that can give you voltage based on any of those
    other parameters.

    You cannot write an equation that derives height from speed. They're
    orthogonal.

    You really need to go back to your fundamentals and learn the difference
    between things like voltage, resistance, power, frequency, and so on.

    Good Luck!
    Rich
     
  20. Life is truly tough. It is made up of a series of trick problems. solve them
    for yourself. Don't ask others to do it for you, especially if you pay no
    attention to the replies.

    Bill
     
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

-