[Help-gnucap] Modeling a neon bulb

[Evan Daniel]

I'm attempting to model a neon bulb as a circuit element.  Bulbs
behave oddly in that they have a region of negative differential
resistance.  I have (approximate) curves of voltage vs. current, so it
seems appropriate to model the bulb as a resistor using the .model
TABLE spline fitting.  However, I have trouble getting the results to
converge in some circuit configurations.

For example, the following circuit converges:

I1 0 1 10u
R1 1 2 100k
R2 2 0 NE
.model NE TABLE 0,0 100p,10 1n,60 100n,68 1u,70 10u,60 100u,55 1m,60
100m,65 1,68 order=1

DC results are V(2) = 60V, I(R2) = 10uA.  The circuit will happily
converge within the entire range, including the negative resistance
portion.  However, attempting to build a relaxation oscillator by
adding a capacitor across the bulb produces a convergence failure,
even with relatively small internal time steps.  Adding a parallel
resistor across the bulb, or a resistor in series with the cap, does
not help (at least with values that should still produce oscillation.

Any suggestions, either on how to get this circuit to converge or an
alternate modeling technique that will produce convergence, would be
appreciated.

(I saw one suggestion online of using a comparator with hysteresis to
switch the bulb between high and low resistance values; I'd prefer
something more sophisticated, capable of modeling the nonlinear nature
of the problem.)

Thanks!

Evan Daniel