On 03/17/2014 06:19 PM, Jim Thompson wrote:
On Mon, 17 Mar 2014 18:06:37 -0400, Phil Hobbs
wrote:
On 03/17/2014 01:27 PM, Tim Wescott wrote:
On Sun, 16 Mar 2014 11:20:38 -0700, Jim Thompson wrote:
On Sun, 16 Mar 2014 13:59:29 -0400, Phil Hobbs
wrote:
On 3/16/2014 1:39 PM, Jim Thompson wrote:
Anyone have an accurate I-V curve for an NE-2 neon indicator lamp?
(I think it's time I applied TANH to it ;-)
...Jim Thompson
Of course tanh is single-valued, unlike the I-V of a NE2.
Cheers
Phil Hobbs
Unless you are clever, like me :-}
Look back over my piece-wise-linear curve-fitting posts, using TANH,
with all derivatives existing and finite.
In this particular case, as I envision it, there are not even any
break-points involved.
What I seek right now is a reasonably accurate I-V curve.
I can't tell if you're being purposely coy, or if you've missed Phil's
point.
Gas discharge devices like neon lamps have hysteresis: when they're on,
they'll conduct at a lower voltage than their turn-on voltage.
So a simple I-V curve doesn't cut it, unless you're only trying to
simulate the device in the on state, and leaving the user out to dry for
figuring out turn-on and turn-off behavior.
If you can model the off-state and on-state behaviour, then at speeds
slow compared with the recombination time, you can get by with a 1-bit
memory telling you which model to use. It's easy to patch that up with
a tanh or the equivalent to make the switch soft.
LTspice does that for you if you specify a negative hysteresis voltage.
Cheers
Phil Hobbs
LTspice (and many other Spice variants) use "IF" statements which
converge for crap.
...Jim Thompson
Sure. But if you add them together using a switch with negative
hysteresis, you'll get the nice smooth behaviour that helps convergence.
LTspice uses arctan instead of tanh, which isn't as nice but usually
works.
Asymptotic function theory uses the idea of neutralizer functions, which
are cobbled together from erf(1/x) in the same way. They're even
better theoretically since all their derivatives go to zero at the
origin, so you can patch it to a straight horizontal line with no
discontinuity in any order. Won't matter much for SPICE.
Cheers
Phil Hobbs
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net