What is the maximum signal frequency that SPICE can handle ?

SPICE itself doesn't really have any fundamental frequency limitations: The limitations are in the models -- both components and SPICE's model of "the universe."

The later -- SPICE's model of the universe -- is that there is no radiation and that (roughly) all components occupy one point in space: There's no phase delay from one wire node to the next, and as such there are no reflections between wired nodes (although components such as transmission lines can create time delays and thereby reflections).

How good this model is of your real-world project is largely a function of how physically large it is relative to the frequencies it's using, in other words, how large it is in terms of wavelengths (lambda) at the maximum frequencies used. When this exceeds, oh, say, lambda/20 or thereabouts (i.e., 18 degrees) -- it may need to be considerably more conservative in some cases --, you probably want to start looking at a different type of simulator.

The former -- component models -- are more directly a user problem. I.e., you can't really obtain a perfect capacitor much less an inductor, so it's up to you to augment those perfect component models once the parasitic effects become significant. Parasitic capacitances at, e.g., op-amp terminals are a commonly-needed addition, as are lead inductances if you're trying to do any sort of high-speed package modeling.

With passive circuits such as you LC low-pass filter, even if there's some "weird numerical things" that happen at 150THz (this would tend to be simulator-specific), you could just normalize all the frequencies involved: A 150Hz LC filter looks exactly the same as a 150THz LC filter, you just have to divide the L and C values by 1e12. (And then realize that there's no means of physical building fractional femtofarad capacitors and fractional picohenry inductors :-) )

---Joel

My favorite solenoid calculator is coming up with 8.98 x 10^-17 H for a 50 picometer coil. The capacitance of a 50pm sphere is 5.56 x 10^-21 F, for a resonant frequency of 225 PHz, or somewhere in the EUV/soft x-ray band (930eV). Assuming, of course, that you could confine a hydrogen atom's electron to a hypothetical subatomic copper wire at this energy in order to measure its inductance.

I suppose, to carry the Bohr-ing analogy further, one might assert that an electron's mass increases its "inductance", slowing it down so it only takes ~13.6eV (3288 THz) to sit around a hydrogen atom.

Unfortunately, this works in the opposite direction, because a more massive particle (e.g., muon) sits much closer to a nucleus, at a higher energy and frequency. Silly quantum physics.

Tim