Subject: Re: Determining bjt noise parameters for Spice models?
Date: Fri, 13 Dec 2002 23:46:03 +0100
Organization: Wanadoo, l'internet avec France Telecom
References: <3DF6537E.B0FC9DBE@worldonline.fr> <3DF90203.C9691374@worldonline.fr> <3DF98E5B.D86102A5@worldonline.fr> <3DFA392F.7FEE50B6@worldonline.fr>
NNTP-Posting-Date: 13 Dec 2002 22:46:03 GMT
X-Mailer: Mozilla 4.79 [en] (X11; U; Linux 2.4.20 i686)
Kevin Aylward wrote:
> A.Iakovlev wrote:
> > Kevin Aylward wrote:
> >> What does the spice V(INOISE) give?
> > A) the initial, stock model: has RC=5.9, RB and RE missing (so =0);
> > with collector current of about 10mA, collector voltage about 9V,
> > collector load of 1kOhm, V(INOISE) is about 0.14 nV/rtHz.
> > In the noise analysis printout, Spice gives "RC=1,800e-22
> > IC=2,938e-15".
> > B) RB=10 added to the model, in the same conditions, V(INOISE) is
> > about
> > 0.43 nV/rtHz.
> > BTW, the datasheet is located here:
> > It says, in particular, "low base spreading resistance: rbb ' = 2.0
> > Ohm (typ.)".
> I don't believe it. It can't possible be 2 ohms. The NF graphs show
> around 4 db at 10ma at ~12 ohms. This means rbb' has to be greater then
> 12 ohms.
> Just set rbb' in the model to give the 0.65nv.
I am yet far from mastering the Spice bjt model, but I suspect that the RB
parameter is not exactly rbb'....
But I've adjusted RB, then KF and checked for several Ic, Ri - the updated
model 2SC3329 showed a behaviour ressemling that in the datasheet! I needed
to set RB=24, KF=0.38e-16 (AF=1).
I've then done the same thing for 2SC1775 ; obtained RB=200, KF=1e-17.
But the third stock model, for 2SA876, behaved differently: even without RB
in the model, it exhibited the thermal noise slightly exceeding the
datasheet specs... In order to set up the flicker noise I had to add some
RB=6, and large KF=1e-10.
> Kevin Aylward