From: Winfield Hill
Subject: Re: Transistor amplifiers v. frequency in SPICE
Date: 12 Jan 2003 10:22:58 -0800
Organization: Rowland Institute
X-Newsreader: Direct Read News 4.20
Paul Burridge wrote...
> I'm now in a position to post the circuit concerned (see deck ...)
> The problem is that voltage gain appears to be falling off far too
> quickly with frequency for the active device employed.
You're biasing the transistor at about 80uA, far too low for most
older-style BJTs to work to 40MHz, let alone 300MHz. It's always
wise to begin a design by studying the data sheets. For a BC337
type the dc beta drops from above 100 down to 40 at such a low
current according to ON Semi, and the f_T is only about 6MHz
according to Siemens (extrapolating the curve down from 1mA).
You can also examine your transistor's model parameters, or you
can give spice simple circuits to tell you the model performance.
Here's my spice's BC337 model.
.MODEL BC338 NPN BF=520 BR=4 CJC=14.4P CJE=44.7P IKF=.3
+ IKR=.45 IS=162F ISE=17.4P MJC=.3 MJE=.5 NE=2 NF=1 NR=1
+ RB=.852 RC=65.2M RE=.363 TF=454P TR=170N VAF=90 VAR=20
+ VJC=.3 VJE=1.1 XTB=1.5
It shows higher capacitances than data sheets I've examined.
> ... this time I'm using the good old 2N2222a ... supposed to
> have a transition frequency better than 300MHz.
> However, my findings don't even come close to this:-
Once again, although the 2n2222 is better than the BC337 at low
currents, it isn't so great at 80uA. E.g., Fairchild specs h_FE
at only 35 min at 0.1mA, and we can expect an f_T of under 15 at
80uA (most curves show it dropping about 5x per decade). Keep
in mind that the f_T parameter is only useful to determine the
maximum available gain with equal source and load impedances,
such as 50 ohms. Furthermore, in high-impedance circuits such
as you have chosen, the transistor's capacitances become _much_
more significant than f_T in determining gain at high frequency.
For example, a Cob of 5pF (the 10V value) loads down your 150k
collector resistor by -3dB at only 210kHz, and -6dB per octave
> At 500Khz, the output signal had dropped to 0.4v p-p.
Yep, down by 8dB at 500kHz, pretty close to the above estimate.
That's entirely a collector-capacitance effect.
. +24V | 1nF
. | +----||--- out
. 150k |
. 1nF | C BC337
. --||--+---B 2n2222
. | E
. 150k |
. | +-----,
. gnd | |
. 140k 1nF
. | |
. gnd gnd
Another problem with your circuit (did I get it right?), you
have a very low collector-emitter voltage, nearly saturated.
Better check it. Assuming beta = 50, I calculate VE ~= 11V,
78uA emitter current, collector IC = 76.5uA, so VC = 12.5V,
for Vce = 1.5 volts! No wonder you see distortion! :>)
Actually it appears your spice's 2n2222 model under-estimates
the part's capacitances for low Vce (or fails to adjust it at
all?), giving you too much gain at high frequencies.