NE-2 I-V Curve ??

Oh, ye of little faith...

A single line equation, 91 characters long, counting every ()*/+- and alpha-numeric... Algebraic/TANH ;-) ...Jim Thompson

It's difficult because neons vary wildly in their behavior. So there is no really good model.

But why make a big deal, and spend a lot of time, to model an NE-2? It's not likely that anybody would care much, and LT Spice has one already.

Feather in one's cap? Where do you get these archaic expressions?

Missed adding, when I posted my simulation curve: Gas discharge devices don't have hysteresis, except perhaps a very small amount in the ionizing/quenching process (and, of course, temperature), but they _do_ have a negative resistance region... see my other post...

Subject: Re: NE-2 I-V Curve ?? Date: Mon, 17 Mar 2014 16:47:19 -0700 Message-ID:

Sounds like you let LTspice mislead you with its switches-with-hysteresis model?

Looks like there's also some confusion here about TANH curves and TANH-based switch models. TANH-based switch models converge much better, but have absolutely nothing to do with neon gas discharge behavior. ...Jim Thompson

And it's amazingly, stunningly, stupidly wrong.

We'll maybe not amazingly; we have grown to expect wrong from you.

OK, you win. Who would know more about parrots and feathers than an old hen?

This is krw-level criticism. If you don't say where and how it's stupidly wrong, you are being more than a little stupid yourself.

I can't say that I find the first part of the V/I curve - from 0 to 20uA

- to be anything like physically realistic, since there's no low voltage way to get tiny currents through a miniature neon, but this error might make the model more tractable.

Not reliably, and certainly not reliably enough to justify crying "wrong" without being specific about the nature and extent of the error.

just noticed this thread. [plus, Aioe only responds on three groups.]

is the NE-2 the little glass bulb with two electrodes inside, that's put inside those AC mains testors [along with a 100k to 1MEG resistor]?

If so, years ago I characterized a few. Plotted their active region I-V curve. Did I write it down? no, no need to, because I'll remember, right?

From memory, turned on around 90V. Stayed on until 60V, or so.

There was a kind of constant current sink into the discharge. Once the arc fired, the voltage was 'kind of' a constant, plus a rather very small sloped current term ==> very low positive resistance. The arc, once fired seemed to sit near that 'turn off' voltage so as the current shut off, and the arc became 'starved' went like negative resistance and would 'pop' off. Also, from memory, pretty much a 1 Watt device, got hotter than ?? when too much current went through, didn't run them there, so did not explore.

I think I had the luxury of an old Tektronix Curve Tracer, too. Trying to remember the series ON resistance, arrrrggg! vaguely remember in the 10 to

Sustaining voltages? on the samples I had they were all very close: 60,

So active ON is something like a 60Vdc battery in series with 50 ohms, or such.

Watch out for photons to 'precharge' the gas tube and cause it to fire at lower voltages in your lab lighting versus firing in a dark room. We had those early version of German made 400V gas discharge suppression arc tubes easily pass incoming inspection, yet NOT fire until almost 600 volts in the dark! I think they now do something to 'pretrigger' them so they fire whether in light or dark.

For what it's worth, the best way to power a gas tube [fluorescent bulb] is with a constant power source, not a contant current source, and definitely not a voltage source.

I'll bet Jeff Leibermann has some URLs that actually plot the 'ON' curve of I-V.

The post was just my first pass to demonstrate a negative resistance region can be done in a way that doesn't choke any Spice with convergence issues. Right now there are too many coefficients to twiddle to get to a specific curve, so I'm in the process of massaging the math to make it definable with peak and valley parameters, VP, IP and VV, IV. Watch this space. Listen to John "Math and simulation are crutches" Larkin whine like a 6-year-old brat >:-}

Device models you wanted are on the way. ...Jim Thompson

Well, it's got the right number of bumps, anyway, but greatly overestimates the current in the low-voltage region--that curve should be in the picoamps or below until the bulb actually strikes. There's no way you can push 20 uA through an unlit NE2.

Cheers

Phil Hobbs

That was just a first-pass to demonstrate the shape... manipulation/simplification of coefficients underway. ...Jim Thompson

It's cheating, though, because the slope is so gentle in the low current range. An accurate model might not converge at all well.

See e.g.

Cheers

Phil Hobbs

The slope is not the problem... it's discontinuities, like "IF" statements and switches that give Spice's heartburn. ...Jim Thompson

There isn't that much difference between the actual I-V behaviour and a switch, though. Smoothing it out by 100 fA worth between zero and 75V won't help modified trap by very much.

Cheers

Phil Hobbs

ired

It's a glow discharge, sustained by positive ion bombardment of the cathode , which knocks off enough electrons to sustained the current. The business of sustaining the glow depends on an erratic process. Not every ion hitting the cathode knocks off an electron, and some knock off more than one. If t he current is low enough, there can be so few charge carriers involved that there's a significant chance that the statistical fluctuations in gain can kill off the process, and you have to wait for natural radiation to genera te a new charge carrier to restart the process.

For an arc discharge, the cathode has to get a lot hotter, to the point whe re it can deform into sharply pointed spikes which emit electrons by therma lly assisted field emission. NE-2 bulbs don't last long at that kind of cur rent.

It's been done.