[Prev][Next][Index][Thread]

Re: PSpice modeling of spark gaps



Hi Bart,

> Original Poster: "Barton B. Anderson" <mopar-at-uswest-dot-net> 

<snip>

> > If Rbreak is sufficiently smaller (<0.1) than Zpri,
> > will its non-linear action noticably affect operation?
> >
> > Has anyone added an op-amp and current shunt to Sbreak
> > to more closely model the quenching?
> > --
> 
> Greg, I haven't tried an op amp or current shunt to Sbreak, but I think
> modeling quenching more closely is a good idea. In Gary's post regarding
> behavior spice models, I think the site offered good information and
> sparked some ideas to those who took the time to read through it.
> 
> I think it can be modeled by adding in behaviors derived from
> current/voltage levels, avalanche thresholds, and other associations in
> the gap. ("includes negative arc resistance, arc sustaining current, and
> accounts for variable dV/dt applied to the device") as stated by Gary.
> These parameters can be "modeled".  Has anyone scoped out these or other
> parameters (real world measurements)?
> 
> Bart

Closest I've come has been scoping primaries and deducing the basic 
characteristics from the waveforms, then applying a bit of math to it 
to reach some interesting conclusions regarding Q etc.  I have also 
tried using a bank of MOSFETs in place of the gap to investigate 
quench issues and resonator action. Additionally, after forming and 
opinion as to the nature of the real gap model, I did setup a 
benchtop LC circuit and stuck in both back-to-back zeners and anti-
parallel diodes (low drop relative to the peak cap voltage). The 
zeners were unrepresentative at the low voltages I was working with 
but the diodes did a nice job *provided that the operating point was 
carefully picked*. I found that I could obtain log, linear and 
antilog responses depending on which circuit values I used so it pays 
to be careful when using diodes of any sort as a substitute for the 
real thing.  
     The most amazing finding the math revealed was that Q for a 
tuned circuit with a gap in it couldn't be quantified as a fixed 
value the way it can if there is normal resistance only. The math 
showed that Q climbed with voltage and absolute losses increased with 
current. The reason for this is that Vgap is not proportional to Igap.
Neither voltage nor current is a fixed quantity but continuously 
varying in normal circuit operation.

Malcolm