8/7/2019 Freq Resp Etc
1/38
Physics116A,12/4/06
Draft Rev. 1, 12/12/06
D. Pellett
Amplifier Frequency Response,Feedback, Oscillations;
Op-Amp Block Diagram and
Gain-Bandwidth Product
8/7/2019 Freq Resp Etc
2/38
2
8/7/2019 Freq Resp Etc
3/38
Negative Feedback and Voltage Amplifier
(see solutions to Prob. 10.35 for proofs)
AB is called the loop gain.
3
8/7/2019 Freq Resp Etc
4/38
Voltage Amplifier: AF dependence on A
B = 1/20, AF = A/(1 + AB) = [A1 + B]1:
A AF
200000 19.998100000 19.99610000 19.961000 19.6500 19.2200 18.2100 16.7
50 14.3
AFmax = 20. If A >> AFmax, AF is insensitive to A. AF is down 3 dB frommaximum when A = 50.
Reduces distortion due to A nonlinearity, allows for variations in amplifiergain from device to device. (What Black wanted back in the 1920s forhis telephone long-distance line amplifiers)
Suppose A is 200000 at low frequency (say 1 Hz) but falling with fre-quency like 1/f at high frequencies due to a built-in low-pass filter withfc = 5 Hz. With feedback, the -3 dB bandwidth would be improved,since AF remains high until A has fallen many orders of magnitude.
8
8/7/2019 Freq Resp Etc
5/38
Phase-Shift Oscillator and AB = -1
Phase-Shift Oscillator
Sinusoidal oscillationwhen AB=-1
4
8/7/2019 Freq Resp Etc
6/38
Amplifier Low Frequency Limitations
5
8/7/2019 Freq Resp Etc
7/38
Amplifier Low Frequency Limitations
(continued)
6
8/7/2019 Freq Resp Etc
8/387
8/7/2019 Freq Resp Etc
9/387
8/7/2019 Freq Resp Etc
10/38
Amplifier High Frequency Limits
Model for parallel (shunt) capacitances to ground in amplifiercircuit:
8
8/7/2019 Freq Resp Etc
11/38
8/7/2019 Freq Resp Etc
12/38
Amplifier High Frequency Limits
Where are these shunt capacitances?
Model for parallel (shunt) capacitances to ground in amplifiercircuit:
8
8/7/2019 Freq Resp Etc
13/38
Common Emitter Amplifier HF Limits
At high frequencies, must consider BE and BC diode capacitances
CBC
CBE
9
8/7/2019 Freq Resp Etc
14/38
Shunt Capacitances In Small-Signal AC Models
How to deal with Cc (or Cgd) which connects input and output?
Base-Emitter Diode Capacitance Base-Collector Diode Capacitance
Simple BJT Model at High Frequencies:
Simple HF JFET Model:
10
8/7/2019 Freq Resp Etc
15/38
Use Millers Theorem to Split Cc (or Cgd)
Apply to HF BJT model in a
common emitter amplifierwith gain = -A:
Input circuit (be) and output circuit (ce) are now separated11
8/7/2019 Freq Resp Etc
16/38
Millers Theorem Proof
QED
Given:
Node 2 (write in terms of v2):Node 1(write in terms of v1):
12
8/7/2019 Freq Resp Etc
17/38
Common Emitter Amplifier Input Stage
CE Amplifier
13
8/7/2019 Freq Resp Etc
18/38
Common Emitter Amplifier Output Stage
14
8/7/2019 Freq Resp Etc
19/38
FET HF Model and Analysis
15
8/7/2019 Freq Resp Etc
20/38
Input Circuit Upper Corner Frequency for 9.54
16
8/7/2019 Freq Resp Etc
21/38
Output Circuit Upper Corner Frequency for 9.54
17
8/7/2019 Freq Resp Etc
22/38
MOSFET Amplifier Example
18
8/7/2019 Freq Resp Etc
23/38
Effect of CS on Low Frequency Response
Simple HF JFET Model:
19
8/7/2019 Freq Resp Etc
24/38
Upper Corner Frequency: Input Stage
Simple HF JFET Model:
20
8/7/2019 Freq Resp Etc
25/38
Output Stage Upper Corner Frequency
Simple HF JFET Model:
21
8/7/2019 Freq Resp Etc
26/38
Ways to Improve Amplifier HF Response
Reduce Miller effect
Common Base amplifier (see solution to Problem 9.21)
Differential Amp using non-inverting input with invertinginput grounded
Cascode circuit similar to above
22
8/7/2019 Freq Resp Etc
27/38
Common Base Detector Amplifier
No Miller effect since cc, cc grounded at base; fast if use fast
BJT (small cc, cc etc.)
*small signalAC model
*
23
8/7/2019 Freq Resp Etc
28/38
Can We Understand Amplifier Operation?
This is an amplifier for short pulses of width ~1 ns.
Pulse response is covered in 116B (i.e., beyond the scope ofthis course)but we can understand its operation based on what we have learned sofar in Physics 116A plus basic physics.
We want the output pulse to have a fast risetime (sharp leading edge). If Rs 50 , the input emitter circuit has fc 2 GHz so the BJT delivers a
short current pulse at the collector which follows the input voltage: ic(t)vin(t)/re.
The collector current is integrated: pulse ic(t) dt = Q = Cv to charge thecombined capacitance C = 2 Cc of the input BJT and the first BJT in theDarlington pair, producing the rapidly rising leading edge of v and theoutput pulse.
Integration occurs because the time constant of the BJT collectorcircuit is = RC = 20 k x 2 pF = 40 ns, much longer than the inputpulse width (assume the base current of the Darlington input
transistor is negligible). The pulse height of the output pulse is proportional to the charge of the
input pulse.
The output is thus expected to look like the sketch. The tail can beshortened using a speedup RC network following the emitter followeroutput (not shown).
(this page is supplementary material not on final)
24
Ch k Si l M d l Wi h SPICE
8/7/2019 Freq Resp Etc
29/38
Check Simple Model With SPICE
Uses MPSH10RF amplifier BJT:cc = 1 pfce = 1.5 pf
SPICE BJT model forMPSH10 is available
from Fairchild website:www.fairchildsemi.com
vin = V(5) (red)vout = V(8) (green)
Good agreementwith simple model
time
voltage
0.0 20.0 40.0 60.0 80.0 100.0
ns
-20.0
-0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
mV V(5) V(8)
Fast Pulse Amplifier
********************VEE!4! 0! DC! -12VBB!2! 0! DC! -6RC! 0! 1! 20KQ1! 1! 2! 3! QMPSH10RE1!3! 4! 20KC1! 5! 3! .01uVS! 9! 0! PWL ( 0 0V 1ns 0.05V 2ns 0V )RS! 9! 5! 50Q2! 0! 1! 6! QMPSH10Q3! 0! 6! 7! QMPSH10RE2!7! 4! 500C2! 7! 8! .01uRO! 8! 0! 1KRI! 5! 0! 10K********************.model QMPSH10 NPN(Is=69.28E-18 Xti=3 Eg=1.11 Vaf=100 Bf=308.6 Ne=1.197 Ise=69.28E-18
+ Ikf=22.83m Xtb=1.5 Br=1.11 Nc=2 Isc=0 Ikr=0 Rc=4 Cjc=1.042p Mjc=.2468 Vjc=.75 Fc=.5
+ Cje=1.52p Mje=.3223 Vje=.75 Tr=1.558n Tf=135.8p Itf=.27 Vtf=10 Xtf=30 Rb=10)
********************
.TRAN! .1ns! 100ns
.control
run
plot! V(5) V(8).endcontrol
.END
(this page is supplementary material not on final)
25
http://www.fairchildsemi.com/http://www.fairchildsemi.com/http://www.fairchildsemi.com/8/7/2019 Freq Resp Etc
30/38
Other Possibilities to Avoid Miller Effect
Note the common base circuit lurking in both
High Av here
High Av here
26
8/7/2019 Freq Resp Etc
31/38
BiFET Op-Amp Simplified Diagram
The differential amplifier and common emitter amplifier use the largeThvenin equivalent AC resistance of a current source along with theinput resistance of the following stage to achieve large gain. See text, Sec.10.2.
Note C1 makes use of the Miller Effect to achieve a large effectivecapacitance for a dominant low-pass filter.
BJTs have identical
characteristics
p-channel JFETs
Current sourceand input
resistance of
next stage play
role of RD or
RC for amplifier
S
D
C
E
IREFVOUT
+VCC
-VEE
C1
Three-stageamplifier:
8/7/2019 Freq Resp Etc
32/38
Effect of Dominant Low-Pass Filter
Upper corner frequency is multipliedby (loop gain +1)loop gain
Finally,
The maximum phase shift is 90(wont oscillate for resistive B).
The product of gain and bandwidth is constant.
High frequency performance is compromised.28
8/7/2019 Freq Resp Etc
33/38
Bode Plot: AF and Bandwidth
AF(dB) A(dB) - (AB)(dB)
At low frequenciesin the example above,
29
8/7/2019 Freq Resp Etc
34/38
8/7/2019 Freq Resp Etc
35/38
Bode Plot for A702A Op-Amp
No large C1: Could make amplifier with BW of several MHz.
Considerable gain left when phase shift equals 180 degrees at 12.5 MHz.
Not fool-proof: A unity gain voltage follower would oscillate.
No dominant pole:has 3 low-pass filters
in series.Amplifier phase shift>180with significant gainand can oscillatewith a resistivefeedback network.
31
8/7/2019 Freq Resp Etc
36/38
BJT CE Large Signal Performance
The maximum output voltage swing is set by BJT cutoff and saturation
Start with the BJT curves of IC vs. VCE for various values of IB, locateQ point
Draw straight line through Q point with slope dIC/dVCE for midband ACsignals (AC Load Line) to determine useful range
For AC, vc = RCic so AC load line slope = ic/vc = 1/RC in this case.Output voltage swing follows AC load line.
3
32
8/7/2019 Freq Resp Etc
37/38
CE Amplifier: DC and AC Load Lines
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
ICvs VCE for 2N2222A npn BJT (SPICE simulation)
IC(mA)
VCE (V)
IB = 8 A
IB = 0 A
IB = 4 A
IB = 12 A
IB = 16 A
Q Point
ACLoadLine
DCLoadLine
VCCRC + RE
VCC
Max. symmetrical voltage swing when Q point centered on AC load line
At Q, no input, BJT power dissipation p VCEIC = 4 V2 mA = 8 mW
If the Q point is centered, the average power dissipated by the BJT ismax. with no AC input actually less when producing a signal. (Seesec. 9.4 in text for details)
4
33
8/7/2019 Freq Resp Etc
38/38
Push-Pull Emitter Follower
Base bias chain keeps both BJTs just at cutoff (or slightly on) at Qpoint No BJT power dissipated if no input signal.
AC input causes one or the other BJT to provide the output.
Maximum average BJT power now 0.1VCEQiC(sat) much more efficientuse of BJTs and power useful for driving low impedance loads at highpower
5