The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: "John S. Dyson"
References: <3DF6537E.B0FC9DBE@worldonline.fr> <3DF90203.C9691374@worldonline.fr> <3DF94AE1.firstname.lastname@example.org> <7V3L9.email@example.com>
Subject: Re: Determining bjt noise parameters for Spice models?
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
Date: Sun, 15 Dec 2002 14:46:30 -0500
NNTP-Posting-Date: Sun, 15 Dec 2002 14:38:49 EST
"Kevin Aylward" wrote in message news:7V3L9.firstname.lastname@example.org...
> JD wrote:
> > "Kevin Aylward" wrote in message
> > news:E3XK9.162$cy2.14496@newsfep2-gui...
> >> John S. Dyson wrote:
> >>> "Kevin Aylward" wrote in message
> >>> news:MZOK9.1335$TY1.email@example.com...
> >>>> It is the fact that transistors have a large spread in rbb' from
> >>>> device to device that actually differentiates low noise devices
> >>>> from any othe device. Its inherent that if one tries to chose a
> >>>> low noise device, then one is selecting principly for *low* rbb.
> >>>> Nothing else matters much.
> >>>> Summary:
> >>>> A low noise transistor is essentially a transistor with low rbb,
> >>>> and low 1/f noise. End of story.
> >>> I'd agree with you given the following qualification: you are
> >>> speaking of low/middle freqs, and also hfe(Beta) does count for high
> >>> source resistance applications, especially in cases where the Ic
> >>> might be chosen to be high for bandwidth or other reasons.
> >> Ahmmm. Did you read my post? I did address the hfe issue, to wit:
> > I did, but you said in an unqualified way that rbb (and the obvious
> > low 1/f noise) as being the key... Remember your 'end of story'
> > statement?
> But there was an epilogue to the story.
It is probably best to consider both the en and in type effects, and not
prejudicially pronounce 'rbb and 1/f' as being the key.
Rather, I tend to prefer these ideas for midband:
en^2 = 4 * k * T * rbb(eff) + 2 * q * Ic * re^2
in^2 = 2 * q * Ib
The input noise voltage is then limited by the
en > re * sqrt(2*q*Ic) and en > sqrt(4*k*T*rbb(eff)) terms.
The input noise current is purely defined (assuming no parasitic
noise sources) by the operating current and Beta. (Again,
So, the input noise power is inversely related to the sqrt(B)
and contributions between rbb(eff) and the operating point
physical limitations are of the same kind (often of the same
order in discrete parts, at normal 100ua to 10ma bias.)
NF is kind of 'bogus', but if properly designed, it is useful for
getting a qualitative view of 'low noise'. The noise factor
for an ideal transistor is about:
nfactor (ideal rsource) = 1 + sqrt(1 + 2 * rbb / re) / sqrt (B).
Once the rbb drops much below re, then the only thing that you
can do is to increase beta so as to decrease total noise. rbb
is of diminishing returns.
When using rbb as a criteria for selecting a low noise device,
even for low impedances where the noise is still significant, choosing
increased Beta can still decrease the noise. This is still important
when trying to decrease the en by running at high Ic for a low
rbb device. Increased beta will help to counter the effect of
high Ic on the base noise current.
I don't mean to claim that low rbb components aren't often desirable.
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup