Simulator error? Or the manifestation of a surge?

Discussion in 'Power Electronics' started by Vinyasi Qx, May 10, 2017.

Are simulated surges realistic?

  1. Yes, they are realistic.

    0 vote(s)
    0.0%
  2. No, they're not. But they are still interesting since they exhibit numerical surges in simulators.

    100.0%
  3. I don't know. That's weird.

    0 vote(s)
    0.0%
  1. Vinyasi Qx

    Thread Starter New Member

    May 8, 2017
    19
    0
    I'm testing out an idea I had developed on Paul Falstad's simulator of power surges and how realistic are they within the context of a simulation? So, my first successful surge on LTSpice also expands its output towards infinity only if the simulation's time step is no greater than 1 micro second. At 100 nano seconds, it escalates to infinite outcome even faster than at 1 micro second.

    [​IMG]

    [​IMG]

    For comparison, here's the power supply plus its inductive load all by itself without the amplifying subcircuit...

    [​IMG]

    BTW, my best simulation on Paul Falstad's simulator dies the instant trapezoidal approximations are turned off on all inductors and capacitors. Naturally, I did not reset anything within Control Panel under Tools. I just left it set at its default of "modified trap' which produces the same results as does trapezoidal. But with 'Gear' Default Integration Method, I get the following results...

    [​IMG]
    At first, I thought I was making a distinct discovery of parasitic transients which used to plague DC transmission lines of the Edison Power Company in New York city a century ago in which a fellow by the name of Steinmetz had to be called in to fix the problem. But now, I don't know if I am studying electrical surges or numerical surges?
     
    Last edited: May 10, 2017
  2. Bordodynov

    Well-Known Member

    May 20, 2015
    2,046
    622
    Hacks! Always default inductors to Rser=0
    and
    Spice> Solver=Alternate
     
    Vinyasi Qx likes this.
  3. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    9,210
    2,176
    Interesting, but demonstrates the fallibility of simulation, especially when using impractical components. Where will you find an air-core coil with zero resistance and an inductance of 1700 Henries?
     
    Vinyasi Qx likes this.
  4. Bordodynov

    Well-Known Member

    May 20, 2015
    2,046
    622
    For L1-L4 Rser=0. R9=1.5 +1mOhm or +0Ohm difference measly
    I have offered the zero resistance of the inductances (by default) to possible was correct to compare different Spice.
     
    Vinyasi Qx likes this.
  5. Vinyasi Qx

    Thread Starter New Member

    May 8, 2017
    19
    0
    Well, after careful research into this particular suggestion, I've come to no definitive conclusion. I'm still ambiguous as to whether or not my simulations are accurately depicting electrical surges or not. But I have managed to gather an opinion that we may not know enough whether to choose one Solver or another, or choose a non-Trapezoidal Integration Method or not.

    This fellow talks about the naiveté of using certain methods of solution for ill-conditioned problems, in section 2.12 entitled: Condition Number, which will result in 'roundoff error' due to the use of inverse matrix ...
    http://ufdc.ufl.edu/UFE0011721/00001/37j
    http://ufdc.ufl.edu/UFE0011721/00001/38j
    http://ufdc.ufl.edu/UFE0011721/00001/39j

    This led me to Wolfram Mathworld's entries on several topics the most significant of which suggested that invertible matrices yielding similar results are at fault ...
    http://mathworld.wolfram.com/ConditionNumber.html
    http://mathworld.wolfram.com/Ill-ConditionedMatrix.html
    http://mathworld.wolfram.com/SingularMatrix.html
    http://mathworld.wolfram.com/MatrixInverse.html
    http://mathworld.wolfram.com/InvertibleMatrixTheorem.html
    http://mathworld.wolfram.com/Invertible.html
    http://mathworld.wolfram.com/InvertibleKnot.html
    http://mathworld.wolfram.com/AmphichiralKnot.html
    http://mathworld.wolfram.com/ChiralKnot.html
    http://mathworld.wolfram.com/MirrorImage.html

    For some unknown reason at this point in my search, I decided to search for the term 'Infinite Mirror' which led me to the Wikipedia entry on Infinity Mirrors and from there to this man's webpage. Now, thinking over what he says there concerning Gabriel's Horn, I may conclude that what we have here is a problem of discernment, namely: deciding which method/s shall be employed to solve my circuit based on any presumption we may have (or prejudice) in assuming that what we are simulating is a finite or an infinite progression? Like Gabriel's Horn, are we projecting an infinite series via analyzing its surface area? Or, are we projecting a finite series via analyzing its volume?

    Analogously, are we focusing on the finite volume enclosed by the three dimensional shape of an electromagnetic sine wave, or are we focusing on its infinite surface defined by the sine wave itself?

    Since, if we're merely attempting to discern the finite volume of a sine wave based on what section of infinite space is enclosed by a circuit's electrodynamic behavior, then any non-trapezoidal integration method is to be employed for simulation, or the 'alternate' sparsing engine as opposed to the normal sparsing engine in LTSpice's Control Panel. Interestingly enough, this is similar to removing one capacitor from the pair of central capacitors in my circuit example (for instance, C4) which will also defeat the infinite escalation of output.

    [​IMG]

    This is why I can't make up my mind which set of methods is to be employed based on my ignorance of whether I should be searching for an answer which bounds a finite volume of space enclosed by an electromagnetic sine wave, or else search for an answer within the infinite surface area of an electromagnetic sine wave?

    Since the finite volume inside an EM wave is merely its consequence, I'm further confused as to: "What is my goal for simulation?" To look at the consequences of a circuit's behavior, or to merely focus on the behavior of its electrodynamics?
     
  6. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    9,210
    2,176
    And I'm convinced they're not. You are simulating, with finite precision, a wholly analogue real-world circuit which inherently has infinite precision, so there are bound to be rounding errors.
     
    Vinyasi Qx likes this.
  7. Vinyasi Qx

    Thread Starter New Member

    May 8, 2017
    19
    0
    Thus, could it be said that this simulation has inappropriated finite precision at 1 micro second, or less, time step for this circuit in particular while readily imposing infinite precision at 2 micro seconds, or greater, time step resolving into a surge (at the beginning) which tapers off rather than erroneously expanding to infinity?

    [​IMG]

    Or in the alternative, this rounding error is inherently resolved by removing C3, C4 or both without altering the output?
     
  8. Vinyasi Qx

    Thread Starter New Member

    May 8, 2017
    19
    0
     
  9. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    9,210
    2,176
    I think all you're seeing is a damped oscillation involving a non-physically-realisable massive ideal inductor and tiny ideal capacitances.
    Run the sim with a maximum timestep of 0.999uS and 1.001uS and see how it behaves.
     
    Vinyasi Qx likes this.
  10. Vinyasi Qx

    Thread Starter New Member

    May 8, 2017
    19
    0
    By all appearance, I am successful at simulating phase conjugation within the field of electrodynamics and outside the usual domain of optics where discontinuity of a wave - leading to divergence of numeric data points on an oscilloscope screen - is desirable and realizable within crystallography despite my success being at the expense of credible electrodynamics. But that's not my fault that electrical engineering never covers phase conjugation, wave discontinuity, retroreflection, and time dilation as a matter of course. Our power supplies are never designed around optics or acoustics where these concepts are not held to be foreign. These concepts are considered alien to electrodynamic theory. A wave is a wave. It makes no difference in whatever medium it finds itself.

    So, you're right and you've made your point well. Ours is a collective defect not circumspect to anyone.

    I'll do whatever I can to find some chink in the armor of our ignorance, because the theory from physics and optics and wave mechanics along with my own simulation experience backs me up. Yet, how do I reach you or anyone else born of standard conventional thinking? I don't know.

    Super thin capacitors are natural phase conjugators for their ability to both reflect and refract an individual EM wave. That's a natural.

    Long coaxial cables composed of numerous junctions are another opportunity for wave discontinuity to develop. And whenever a PA system is setup with too many of these coax junctions along with the magnetic pickups attached to guitars on stage (of any rock concert) accelerating these discontinuities, before you know it, a surge develops and a large speaker directly in front of the audience blows up from the excessive build up of electrical force despite the gas fired AC generator (supplying power for this imaginary micro-grid PA system) whose own meter has not varied one bit. Where does the extra power come from?

    So, my simulations are always composed of two ultra-low capacitors and at least one transformer. At its heart, this is all that is needed to develop wave discontinuity outside of its usual domain of crystallography and blow up the simulation.

    Thanks. Without my depression emanating from your logical suggestions, I could not appreciate to what level of compartmentalized, specialized depravity we have collectively sunk to along with the divide and conquer mentality rigorously guiding our educational institutions. These institutions of learning encourage radical innovation outside of the development of our power supplies. Optics is nice, but it doesn't recharge an EV.

    I hope I've done this topic, and your coverage of it, justice by bowing out to your superior understanding of what is politically and financially correct.

    https://is.gd/explainconjugate
     
Loading...