To Whom It May Concern,
When simulating a design using a 2n4117 I found my noise figure was much worse than expected. I traced the issue back to the pspice library component included with LTSpice. In particular, it looks like at some point there was a translation error from exponential notation to using prefixes in the model.
In particular, for the 2n4117 the original value for kf was 45.61e-18. In the current version of LTspice kf=45610f=45610e-15=45.610e-12. Basically it looks like someone shifted the decimal place in the wrong direction.
.model 2N4117 NJF(Beta=0.033m Betatce=-0.5 Vto=-1.2 Vtotc=-2.5m Lambda=13m Is=5.261f Xti=3 Isr=51.03f Nr=2 N=1 Rd=1 Rs=1 Cgd=3.94p Cgs=4.93p Fc=0.5 Vk=90.45 M=435.9m Pb=1 Kf=45610f Af=1 Mfg=Linear_Systems)
Old Version I found on GitHub"
.model 2N4117_E NJF(Beta=33.07u Betatce=-.5 Rd=1 Rs=1 Lambda=13m Vto=-1.2 Vtotc=-2.5m Is=5.261f Isr=51.03f N=1 Nr=2 Xti=3 Vk=90.45 Cgd=3.94p M=.4335 Pb=1 Fc=.5 Cgs=4.93p Kf=45.61E-18 Af=1 mfg=Fairchild)