ZENER DIODES IN LTSPICE

[TW]

However you haven't modeled the junction capacitance, nor any snap recovery effects, so it isn't a lot better than a properly configured standard 'native' (Berkeley SPICE) diode model which is likely to be considerably faster and more stable in the simulation.

Understood. However, it seems hard to get the standard model to look much like a zener in both forward and reverse. The subckt examples posted above go to some lengths and still aren't all that convincing. Is it possible to make a .model work? I suspect most of the time it doesn't matter but it was news to me and it might be worth keeping in mind.

I've attached a file with my curve fitting and Jay's .model just to show what I'm looking at.

Tom.

[Jay_Diddy_B]

There are more params in the ltspice diode model as the usually used in the component libs.
For example, in a different thread where we messed with TC of a zener I had to add Trs1 and Trs2 parameters, and, indeed, the resulting TC was pretty different afterwards..

http://ltwiki.org/LTspiceHelp/LTspiceHelp/D_Diode.htm

The help file on diodes is useful in that it gives us the names and descriptions of the parameters within the diode model that can be adjusted.

Another useful feature of LTspice is the directive AKO: AKO means 'A Kind Of'

.model my_transistor ako: 2n2222A (BF=400)

This describes a parts with all the properties of a standard 2N2222A but with a beta= 400

We can do this with Zener diodes.

.model 1N4728A AKO: 1N750 (Rs=3.5 bv=3.3 ibv=76m ibvl=76m Nbvl=9.5)

This describes a 1N4728A as A Kind Of 1N750 with 5 parameters modified.

We can build models with the parameters stepped to explore what each of the parameters does:



Result for stepping RS




Result for stepping BV



Result for stepping IBV



Result for stepping IBVL



Result for stepping NVBL




We can see how each of these 5 parameters changes the VI curve.

Regards,
Jay_Diddy_B

[TW]

I'm still messing about with the resistor model. I'm sure it isn't a good general purpose solution. It may not be any good at all. I dunno, but think it might work for the circuit I'm working on. It is low voltage, low current DC.

Code: [Select]

.SUBCKT 1N4728A 1 2
* Node 1 -> A
* Node 2 -> K
* Curve fit to a single, used sample. YMMV
* sampled 0.5mA - 100mA. Very likely bogus outside that range.
* y=a*I**b
***************************************
VS 1 3 0 ;shunt
RZ 3 2 R ={ if( I(VS) > 0, 0.885*abs(I(VS))**-0.972 , 4.48*abs(I(VS))**-0.878) }
.ENDS

Tom.