Artifacts in simulations

[F6FLT]

I just came across this spice diagram provided by Vasik ( https://www.overunityresearch.com/index.php?topic=4227.msg97375#msg97375 ).




If you know electromagnetism, you should be very very surprised.
How could a generator feeding linear elements like capacitors or inductors, with a single sinusoidal frequency, give rise to signals of different frequencies? Since a model is based on a mathematical treatment and this is mathematically impossible in the formalism of electromagnetism, it is therefore a problem of the model. The real realization of this setup will not show any beats.

This kind of anomaly is due to the choice of the time step. The default time step in LTspice is often too large when a resonance frequency of a circuit is close to, but different from, the generator frequency. I already had this case in my simulations. But I thought about it rather than publishing the miracle picture of the fairy tale because I prefer understanding to grandstanding. And by reducing the time step to 0.1 µs, the beat disappears.
The only difference between the left and right simulation is the time step.




There is a total correspondence between the representation of a signal in the time domain and in the frequency domain. If your time step is 10% of the signal period, you cannot differentiate frequencies that are less than 10% apart. As LTspice introduces noise, it introduces signals with different frequencies which become very visible if the time step is badly chosen because it is too large.

Those who are satisfied with a mediocre knowledge of electromagnetism because they think they are smarter than the physicists who found the laws of electromagnetism, are condemned to see anomalies and free energy everywhere they are not, and conversely when there is really some in their setup, they will not see it.  "Chance favors only the prepared mind". If only they would remain discreet, but the less they know, the more they want to impress us with their pseudo-discoveries.

[partzman]

F6FLT,

Ah yes I totally agree with your analysis above as I have fallen victim to these circumstances in my own feeble work many times!

However in this case, below is the sim posted by Vasik with a maximum timestep of 20ns!  I have no idea of what he is proposing here and the results he shows are only possible with proper coil resistances.

Regards,
PM

[F6FLT]

Hi Partzman,

I see I'm not the only one who got screwed    !
Everything depends on the Q factor. On Vasik's schematic, the inductances are of zero resistance except L1 which is of very small value (0.01 ohm). Q is therefore very high, not realistic in practice, its time step of 20 ns is too long.
In my schematic, Q was much lower because of the resistances (R0 0.1 ohm, R1/R2 1 ohm, and 1 ohm more for each of the 3 inductors, Rser not displayed).
I have just redone the test with 0 ohm everywhere except 0.01 for R0, and of course I find the beat again, even with 20 ns time step.

I have attached my initial asc file, for those who would like to check.