Documente Academic
Documente Profesional
Documente Cultură
1. Introduction dio), 900 MHz (Cellular and into general design considerations
Discrete transistors offer low cost Pager), 1.2 and 1.5 GHz (GPS), for building low noise amplifiers.
solutions for commercial applica- 1.9␣ GHz (PCN), 2.1 to 2.7 GHz Two amplifier designs will be pre-
tions in the VHF through (MMDS and ITFS) and the 2.4 GHz sented along with measured
microwave frequency range. (ISM) markets. results. The application note will
Today’s silicon bipolar transistors finish with a discussion of match-
offer state-of-the-art noise figure Generally, silicon bipolar devices ing network losses and their effect
and gain performance with low are easier to work with at the on amplifier noise figure.
power consumption. lower frequencies because of their TouchstoneTM circuit files and
inherently lower impedances. simulated results for both amplifi-
This application note discusses the However, today’s state-of-the-art ers are included in the Appendix.
design techniques and perfor- low current bipolar transistors
mance of the Hewlett-Packard have considerably higher imped-
AT-3 series of silicon bipolar tran- ances making them comparable to
2. Noise Parameter
sistors as used in typical low noise GaAs FETs at these frequencies. Measurements
amplifiers for use in the various Similar design techniques must be A typical test set-up for measuring
commercial markets. used with these devices to assure noise parameters is shown in Fig-
good performance. Appropriate ure␣ 1. The device under test (DUT)
Although specific designs are pre- design techniques will be is normally inserted into a test fix-
sented for 900 and 2400 MHz, the presented. ture that includes 50 ohm input
techniques are applicable to other and output transmission lines
applications in the VHF through S This application note will begin whose effect can be calibrated out
Band frequency range. This would with an overview of noise param- for the particular frequency. As a
include the 450 MHz (Mobile Ra- eters and definitions and then lead minimum, a double stub tuner or
ΓO
DEVICE
UNDER NOISE
NOISE BIAS BIAS
TUNER TEST TUNER ISOLATOR FIGURE
SOURCE TEE TEE
FMIN.Ga METER
RN
5964-3854E 4-4
equivalent must be used at the in- The noise figure of a linear two to be and then solve 4 equations
put to the DUT to present the port is given by equation (1) shown and 4 unknowns. This method has
required Gamma Opt., ΓO, to the in the table below. become a more accurate industry
device for it to achieve its mini- standard.
mum noise figure. Although not In equation (1),
always required, a tuner can be in- NFmin is the device minimum The input tuner must be capable of
serted at the output of the DUT. noise figure when terminated in transforming the customary
Providing a conjugate match at the Yon, 50␣ ohm source impedance to that
output of the DUT while the input Yon is the generator admit- required for the device to achieve
is presented with ΓO provides a tance at which minimum its rated noise figure. As an ex-
means of measuring associated noise figure occurs, ample, for the Hewlett-Packard
gain at minimum noise figure. One Yg is the generator admittance AT-30511 operated at a VCE of 1
particular manufacturer of auto- presented to the input of the volt and IC of 1 mA, ΓO has a mag-
matic noise measuring equipment device, nitude of 0.76 increasing to 0.96 at
uses a tuner on the input and ter- Rn is the noise resistance 500 MHz. These numbers represent
minates the output in 50 Ω and which gives an indication of impedances that can be increas-
then measures the resultant S22. A the sensitivity of noise figure to ingly difficult to match with low
calculation then provides the asso- termination, and loss. Losses of the tuner become
ciated DUT gain. Bias Tees are Gg is the real part of the genera- more questionable as the ΓO in-
used at the input and output of the tor impedance. creases, plus the ability to design
DUT to bias the device. A noise and build a low loss matching net-
source with a low Excess Noise The equation can be transformed work becomes more of a challenge.
Ratio, ENR, such as the Hewlett- into an equivalent equation involv- An early paper by Strid[1] discusses
Packard HP346A, is desired as it ing the source reflection tuner losses as well as losses in
minimizes test error by minimizing coefficient, ΓS, and the reflection matching networks. The problem
the range over which the noise fig- coefficient required for minimum with tuner losses is that the tuner
ure meter must remain linear. The noise figure, ΓO. See equation (2) has a different loss for every tuner
HP346A noise source also has below. setting and this effect is more pro-
minimal change in reflection coef- nounced at higher reflection
ficient between the “on” and “off” Once ΓO has been determined and coefficients. The user must rely on
states. This minimizes the ability of NFmin determined, the Rn can be calibration data supplied by the
the DUT to change its gain with determined by making a 50 Ω manufacturer and this data may
varying input termination. Any noise figure measurement and cal- not be guaranteed much above a
change in DUT gain will increase culating Rn. This procedure only reflection coefficient of 0.6 to 0.7.
the measurement error. An isolator works well if ΓO can be deter-
placed at the input of the noise fig- mined by a single measurement. A If the tuner’s calibration were accu-
ure meter is always desirable but more accurate method would be rately known and relatively
may not be possible at the lower to pick 4 reflection coefficients (4 constant with tuner setting then it
frequencies where size becomes terminating impedances) in the would be a simple matter to adjust
more of an issue. vicinity of where one believes ΓO the tuner and DUT for lowest noise
figure and then subtract out the
Equations
tuner loss to obtain the DUT mini-
mum noise figure, NFmin. With
Rn 2 varying loss in the tuner, it is diffi-
NF = NF min + Yg - Yon (1)
Gg cult to determine if adjusting the
tuner and DUT for minimum noise
figure minimizes the DUT noise
figure or the tuner loss. The alter-
native of presenting 4 known
impedances to the device and solv-
2
4 Rn Γs - Γo ing 4 equations and 4 unknowns is
NF = NFmin + (2) preferred.
Zo 2
( 1 + Γo 2 ) (1 - Γs )
4-5
3. General Design an amplifier from delivering its AT-32033 is one of a series of sili-
Considerations rated performance. Other tech- con bipolar transistors that are
Implementing the input match can niques such as resistive feedback fabricated using an optimized ver-
take on any of a variety of circuit and resistive loading can improve sion of Hewlett-Packard’s 10 GHz ft
topologies depending on the fre- stability but can limit power out- Self-Aligned-Transistor (SAT) pro-
quency and the space allowed for put capability. cess. The die are nitride passivated
implementing the network. Alter- for surface protection. Excellent
natives may include: An often overlooked part of an am- device-to-device uniformity is guar-
plifier is the bias decoupling anteed in fabrication by the use of
• Lumped element network, network that must be invisible to ion-implantation, self aligned tech-
• Microstripline network, or the RF matching networks. Gener- niques, and gold metalization.
• Cavity filter match ally they provide a low loss
method of biasing the devices but The AT-3 series of devices has a
A lumped element network can be in some situations can actually be 3.2␣ micron emitter-to-emitter pitch
either high pass, low pass, or used to provide some resistive and has been fabricated in a variety
bandpass and generally 2 or 3 ele- loading for stability both in-band of geometries for various applica-
ments. Below 2 GHz these and out-of-band. Properly designed tions. The 20 emitter finger
networks will generally be lower bias decoupling networks can also interdigitated geometry yields an
loss than a microstripline circuit be used to provide some form of easy to match device capable of
because of substrate losses. Above band pass or high pass filtering moderate power at low to moder-
2 GHz, the lumped element topol- that could help reduce low fre- ate current. The 10 emitter finger
ogy will be very difficult to quency out-of-band gain. A poorly geometry offers higher gain at low
synthesize with realizable compo- designed amplifier with very high current while the 5 emitter finger
nents. The cavity filter approach is low-frequency gain that may be geometry offers the highest gain at
probably the lowest loss matching unconditionally stable according lowest current consumption. The
network but cost and size generally to the computer simulation may smaller devices at very low current
make it prohibitive for most com- actually oscillate if the output can present very high impedances that
mercial applications. radiate back to the input. The en- can make them more of a challenge
closure that houses the amplifier to design with. The impedances
Losses of actual input matching must be designed to offer enough associated with very low current
circuits have been measured at isolation around the circuit such transistors at 900 MHz are very
nearly 0.5 dB at VHF frequencies that it does not make the amplifier similar to those presented by
when attempting to match the high circuit unstable at any frequency. 500␣ micron MESFETs at 900 MHz.
impedances of MESFETs. Similar
impedances can be encountered The manner in which circuit ele- The 900 MHz AT-32033 amplifier is
when using low current silicon bi- ments are implemented will affect designed for a nominal 1 dB noise
polar transistors. Matching a the overall amplifier performance. figure and 10 dB associated gain at
device for lowest noise perfor- The use of etched circuit elements 2 mA collector current. Although
mance does not necessarily as opposed to surface mount dis- the device is capable of sub 1 dB
guarantee the best input VSWR and crete elements offers a cost benefit noise figures, most applications do
performance tradeoffs need to be but may affect losses. Surface not require much below 1.5 dB.
made. A solution is the use of in- mount components offer small size Starting out with a device that has
ductance in the emitter leads to but parasitics and device Q must such a low NFmin allows the de-
create negative feedback which be understood if their effect on cir- signer to make tradeoffs between
can bring ΓO and S11* closer in cuit performance is to be properly noise figure, gain, stability, etc.
value[2,3,4]. The amount of induc- analyzed.
tance must be carefully weighed The schematic diagram of the
against its effect on other circuit 4. 900 MHz Silicon 900␣ MHz amplifier is shown in Fig-
parameters such as gain and stabil- Bipolar Amplifier ure 2. The input noise match
ity. An improperly chosen amount The 900 MHz amplifier uses an consisting of a low pass network in
of inductance can cause out-of- AT-32033 which is in the industry the form of C2 and Z1 provides a
band oscillations that can prohibit standard SOT-23 package. The low Q broad band match. A small
wound inductor could replace
4-6
transmission line Z1. A value in the Z2 C4 Zo OUTPUT
range of 15 to 20 nH would be a INPUT Zo C1 C2 Z1
good substitute. In the actual cir- Q1 C5 C6
cuit it was found that the input
shunt capacitor was not required. R1 R5
Adding a shunt capacitor at this
point will allow the designer to Z3 Z4
make tradeoffs between noise fig-
C3
ure and input VSWR.
C7
R2 R3
The output match consists of a
3␣ element low pass network. The R4
3␣ element network allowed a Vcc
4-7
gain and power output. Maximum
power output rarely occurs when
any device’s output port is conju-
gately matched. How much
improvement can be achieved by
power matching?
Figure 3. 2X artwork for 900 MHz Amplifier using 0.062 inch thick FR-4.
2.0
1.5
NOISE FIGURE, dB
1.0
0.5
0
700 750 800 850 900 950 1000
FREQUENCY, MHz
12
10
8
GAIN, dB
Figure 4. Component Placement for 900 MHz Amplifier using 0.062 inch thick 2
FR-4.
0
750 800 850 900 950 1000
FREQUENCY, MHz
WARNING: DO NOT USE PHOTOCOPIES OR FAX COPIES OF THIS
ARTWORK TO FABRICATE PRINTED CIRCUITS. Figure 6. AT-32033 Amplifier Gain.
4-8
ing the input circuit higher. A spec- 5. 2400 MHz Silicon
trum analyzer can be useful here to Bipolar Amplifier
determine that harmonics are not The 2400 MHz amplifier is designed
high enough in level to distort the around the Hewlett-Packard
power meter measurement. In AT-31011. The 10 emitter finger
small steps increase the input geometry plus the SOT-143 pack-
power and then retune the output age with the two emitter leads
tuner for maximum fundamental offers improved performance at
power. After retuning the output frequencies above 2 GHz. At a
for a power match, it was found rated current of 1␣ mA, the
Figure 7. 900 MHz Amplifier showing that the P1dB increased to nearly
the Placement of Wound Inductors in AT-31011 provides a device noise
place of Microstripline Networks. 2␣ dBm with a reduction in gain of figure of 1.7 dB at 2400 MHz with
1␣ dB over the small signal conju- an associated gain of 10 dB.
absorbing some of the power. This gate match. In order to revise the
resistor was placed in the circuit to output match to provide a power The schematic diagram of the
raise in-band stability. Placing a match would require breaking the 2400␣ MHz amplifier is shown in
short across this resistor and re- circuit at the collector port of the Figure 8. The input noise match
measuring the 1 dB compression device and measuring the new consisting of a low pass network in
point showed an improvement of Gamma Load (ΓL) presented by the the form of C2 and Z1 provides a
3.5 dB! Also observed was an in- existing circuit plus the external low Q broad band match. The ca-
crease in collector current when tuner. It is interesting to note that pacitor at C2 can be optimized for
the device is driven toward com- the output return loss which was either a noise or conjugate match.
pression. An increase in current greater than 20 dB at 850 MHz is
causes an increase in the voltage now only 8.5 dB at the power The output match consists of a 2
drop across the 180 Ω resistor match condition. element low pass network while
causing the collector voltage to the interstage network consists of
sag. Minimizing the value of this In addition to measuring P1dB at all two short transmission lines and a
resistor will tend to keep VCE high output matches, the two tone third series capacitor. The artwork and
when the device is driven hard and order intercept point (IP3) was component placement guide are
will also minimize power absorp- also measured. For each test, two shown in Figures 9 and 10. Mini-
tion in the circuit. The drawback tones were introduced at the input mal emitter inductance is used to
could be decreased stability. Some to the amplifier which are sepa- preserve in-band gain without sac-
compromise with respect to output rated by 10␣ MHz. The resultant rificing stability. Resistor R1
loads may have to be instituted if third order products were then provides low frequency stability
additional power output is desired. measured and averaged and IP3 while resistors R5 and R10 en-
was calculated. The results are hance overall stability, including
A P1dB of -2 dBm is still slightly shown in Table 1. in-band performance. Two current
lower than the data sheet specifica- sources (resistor R2 connected to
tion. However, the output is still The results show a consistent 20 to the resistive divider consisting of
conjugately matched and not 21 dB of difference between P1dB resistors R3 and R4 and R7 con-
power matched. In order to pro- and IP3. This is somewhat greater nected to the resistive divider
vide a power match, one must pro- than has been measured on other consisting of R8 and R9) provide
vide an alternative output match. larger geometry small signal the necessary base current to pro-
In order to prove that a power devices but it does appear to be duce the desired 1 mA collector
match will provide greater power repeatable. current in each device.
output, a lab exercise can be set-
up. A double stub tuner is con- Table 1. 900 MHz Amplifier Power Output Summary.
nected in series with the existing
conjugately matched amplifier out- Condition P1dB IP3
put circuitry and the power meter. Conjugate Match -5.5 dBm +16 dBm
The tuner is then adjusted for Conjugate Match w/o resistor - 2.0 dBm +18 dBm
greatest power output while driv- Power Match +2 dBm +23 dBm
4-9
C4 OUTPUT
Z4 Zo
Z2 C3 Z3
INPUT Zo C1 C2 Z1 Q2 C11
R10
Q1 R6
R5
Z8
Z6 Z7
Z5
C9
C7
C5 C8
C10
R7 R8
R1
C6 R2 R3 R9
Vcc
R4
Vcc
C1, C4, C5, C9 –10 pF CHIP CAPACITOR R5, R7 – 47 K OHM CHIP RESISTOR (ADJUST FOR RATED Ic)
C2 – 1.3 pF CHIP CAPACITOR R3, R4, R8, R9, 15 K OHM CHIP RESISTOR
C3–1.5 pF CHIP CAPACITOR R5, 16 OHM CHIP RESISTOR
C6, C7, C8, C10,–1,000 pF CHIP CAPACITOR R6, 1 K OHM CHIP RESISTOR
C11– 2 pF CHIP CAPACITOR (ADJUST FOR MIN OUTPUT VSWR) Zo – 50 OHM MICROSTRIPLINE
Q1, Q2 – HEWLETT-PACKARD AT-31011 SILICON BIPOLAR TRANSISTOR Z1–Z4 – ETCHED MICROSTRIPLINE CIRCIITRY
R1, R10 – 50 OHM CHIP RESISTOR Z5–Z8 – MICROSTRIP BIAS DECOUPLING LINES
4-10
The amplifier has a measured noise 6. Other Applications The power delivered to the load is
figure between 1.9 and 1.95 dB The low current bipolar transistors simply the power that would be
from 2400 to 2500 MHz with a can also be used in frequency con- delivered if the load were conju-
nominal associated gain of 20 dB at verter applications. Although not gately matched to the network.
a total current consumption of optimum, the 900 MHz amplifier With the input to the network be-
2␣ mA for both devices. Measured circuit shown in Figure 4 can be ing 50 ohms, ΓS = 0, equation (3)
output 1 dB gain compression used to demonstrate mixer opera- reduces to equation (4).
point is -4.5 dBm with an associ- tion. The amplifier circuit can be
ated IP3 of +7 dBm. See Figures 11 modified for use as a downcon- The measurement of S21 is nothing
and 12. verter to a 10.7 MHz IF by simply more than a 50 ohm available gain
coupling out the IF by attaching a measurement. It is imperative that
2.3 0.1 to 0.3 µH coil to the output cir- the source and load presented to
cuit. The point to couple to should the device be as near a perfect
2.2 be at the junction of C4 and C6 50␣ ohm impedance as possible as
(reference Figure 2). Ultimately, this is the reference impedance for
NOISE FIGURE, dB
2.1
the IF should also have a dc block- the reflection coefficient. Both the
ing capacitor but it was not numerator and the denominator
2.0
required for this simple test. The use only the magnitude of S21 and
LO is injected into the output port ΓL or S11 so it is only necessary to
1.9
of the amplifier and the amplifier measure accurately the magnitude
1.8 input port is the RF input port. and not phase. With a network
With a nominal +3 dBm LO, the with a very high reflection coeffi-
1.7
2,000 2,100 2,200 2,300 2,400 2,500 2,600 circuit without any optimization cient, S22 becomes very large and
FREQUENCY, MHz
provides a nominal 6 dB conver- S21 very lossy. With a low reflec-
sion gain and less than 12 dB noise tion coefficient, S22 is smaller and
Figure 11. AT-31011 Amplifier Noise figure. Optimization of the bias and loss becomes very low.
Figure.
matching structures will improve
performance. Generally, higher LO Several circuits are analyzed for
25 increases conversion gain but loss and the results are shown in
there is generally a nominal LO Table 2. The first circuit is a simple
24
power that produces the lowest series inductor and blocking
23
noise figure. Bias voltage and cur- capacitor providing a noise match
22
rent can be critical, especially for for a 900 MHz amplifier. Loss cal-
21
culated to be 0.23 dB. The second
GAIN, dB
S 21 2 (1 - Γ S 2 )
Ga = (3)
1 - S 11 ΓS 2 (1 - ΓL 2 )
2 2
Ga = S 21 / (1 - ΓL ) (4)
4-11
Table 2. Losses of Various Matching Networks. Actual measurements of Q suggest
that a Qu of 100 to 200 might be a
Circuit Freq. Circuit S21 S21 S22 S22 Loss
better value for maximum Qu. The
Number (MHz) (dB) (dB) (dB)
equivalent series RS can now be cal-
1 900 L -5.8 0.513 -1.4 0.85 0.23 culated. Based on a Qu of 200 :
2 500 L/C -9.8 0.324 -0.54 0.94 0.45
3 150 Cap coupled -5.8 0.513 -1.4 0.85 0.23 jωl
tank
Qu =
Rs
4 900 AT-32033 -0.5 0.944 -12.7 0.23 0.26
Microstrip jωl
Therefore Rs =
5 2400 AT-31011 -2.2 0.776 -4.75 0.58 0.44 Qu
Microstrip
6 2400 ATF-10236 -1.1 0.881 -7.0 0.45 0.14 Solving yields RS = 0.565 Ω. This is
Microstrip the equivalent series resistance of
the 5 turn coil.
noise figure at 500 MHz. Compared 8. Calculating Circuit The loaded Q of the circuit, Ql, can
to circuit number 1, the measured Losses now be calculated or measured
loss has increased 0.2 dB because The calculation of circuit losses is with the device terminating the
of the losses associated with only as accurate as the models of matching network and a 50 Ω
matching to a higher impedance. the individual circuit elements. source termination. The loaded Q
The third circuit is a parallel tuned The inductor in matching circuit #1 can be found by dividing the 3 dB
circuit with a series input capacitor will be analyzed. bandwidth into the nominal center
used to provide the high imped- frequency, fO. Another alternative
ance transformation to a reflection The inductor used is an air wound is to plot Zin of the network termi-
coefficient of 0.85 at 150 MHz. solenoid with 5 turns #26 guage nated with the device. Any point on
Notice that the measured loss is enamel wire with a 0.075" I.D. The the Smith Chart represents an im-
the same as circuit number 1 with inductance is calculated as [7,8] : pedance consisting of both real
a similar reflection coefficient but and reactive components. Dividing
at a different frequency. Circuits 4, n2 • r2 the reactive part into the real part
5, and 6 are micro-stripline designs L (µH) =
9 r + 10 l provides the Q of the network. The
for 900 and 2400 MHz.. Circuit 4 is Ql is approximately 1 for this net-
the input network for the 900 MHz work. Network insertion loss can
where n = number of turns
amplifier using the AT-32033 previ- now be calculated with the follow-
r = radius
ously described. Subtracting the ing equation:
l = length
0.26 dB for the input loss reduces
the noise figure to 1 dB which is Insertion Loss I.L. =
Solving yields L ~ 20 nH
the noise figure as predicted with- -20 log [(Qu - Ql) / Qu]
out circuit losses. The loss of the
The unloaded Q can be calculated as
input match for the 2400 MHz Solving yields an insertion loss of
follows:
AT-31011 amplifier was measured 0.043 dB. If the Qu of the inductor
at 0.44 dB. This is as a result of is only 100 then the loss would cal-
Qu = 2 r A f1/2
using lossy FR-4/G-G10 at 2␣ GHz culate at 0.087 dB, which is
and a higher reflection coefficient. probably more consistent with the
where r = radius
Contrast this result with circuit measured results. Other factors
A = 100 - 130 for 1/r from
number 6 which is a noise match such as radiation loss,
2 to 20
for the ATF-10236 FET etched on microstripline loss and capacitor Q
f = frequency in MHz
ER = 2.2 material[6]. can make up the difference in the
Solving yields Qu = 259. calculated versus measured loss.
4-12
Some packaged and molded induc- Appendix I.
tors have a Qu of only 25 and with AT-32033 900 MHz Low Noise Amplifier Touchstone Circuit File
a higher Ql the loss can approach !SINGLE STAGE DESIGN
0.5 dB! !A.J.WARD 11-28-94
!REVISED 06-15-95
DIM
9. Conclusions FREQ GHZ
At low bias currents, the AT-3 IND NH
series of devices can have a ΓO as CAP PF
high as 0.94 at sub 1 dB noise fig- LNG IN
ures. This allows the designer VAR
more flexibility in making tradeoffs W1=.03 !INPUT LINE WIDTH
and still achieving 1 to 1.5 dB noise L1\1.11529 !INPUT LINE LENGTH
figures at 900 MHz with silicon. In C1=0.7 !INPUT SHUNT CAPACITOR,
! TRADEOFF NOISE FIGURE AND INPUT VSWR
addition to concern over tuner
W3=.03 !OUTPUT LINE WIDTH
losses and their effect ultimately L3=1 !OUTPUT LINE LENGTH
on the accuracy of the noise pa- LL1=.05 !EMITTER LEAD LENGTH, UP TO .1
rameter measurement, the losses STILL OK ON STABILITY
associated with actual noise CKT
matching structures can approach MSUB ER=4.8 H=.062 T=.0014 RHO=1 RGH=0
0.5 dB unless attention is paid to TAND TAND=.002
component Q. MLIN 1 2 W=.1 L=.05
SLC 2 3 L=.25C=10
MLIN 3 4 W=.1 L=.2
References
SLC 4 5 L=.25 C^C1
1. E. Strid, “Measurement of Losses in Noise-
Matching Networks,” IEEE Transactions VIA 5 0 D1=.03 D2=.03 H=.062 T=.0014
Microwave Theory and Tech., vol. MTT-29, pp MSTEP 4 6 W1=.1 W2^W1
247-252, Mar. 1981 MLIN 6 7 W^W1L^L1
2. L. Besser, “Stability Considerations of Low !IND 6 7 L#0 14.82035 30 !OPTIONAL INDUCTOR
Noise Transistor Amplifiers With Simulta- RES 4 10 R=50
neous Noise and Power Match,” Low Noise MLIN 10 11 W=.1 L=.1
Microwave Transistors and Amplifiers, H. MSTEP 11 12 W1=.1 W2=.03
Fukui, Editor, IEEE Press, 1981, pp. 272-274. MLIN 12 13 W=.03 L=1.5
3. G.D. Vendelin, “Feedback Effects on the SLC 13 14 L=.4C=1000
Noise Performance of GaAs MESFETs,” Low VIA 14 0 D1=.03 D2=.03 H=.062 T=.0014
Noise Microwave Transistors and Amplifiers, MSTEP 7 16 W1^W1 W2=.1
H. Fukui, Editor, IEEE Press, 1981, pp. 294-
MLIN 16 17 W=.1 L=.1
296.
MSTEP 17 18 W1=.1 W2=.02
4. D. Williams, W. Lum, S. Weinreb,
MLIN 18 19 W=.02 L=.01
“L-Band Cryogenically Cooled GaAs FET Am-
plifiers,” Microwave Journal, October 1980, p. DEF2P 1 19 INPUT
73.
5. “Using the ATF-10236 in Low Noise Ampli- S2PA 1 2 3 C:\S_DATA\BJT\T320333A.S2P
fier Applications in the UHF Through 1.7 GHz DEF3P 1 2 3 DEVICE
Frequency Range”, Hewlett-Packard Applica-
tion Note 1076, Newark, Ca.11/94 MLIN 1 2 W=.02 L^LL1
6. “S-Band Low Noise Amplifiers”, Hewlett- VIA 2 0 D1=.030 D2=.030 H=.062 T=.001
Packard Applications Note AN-G004, VIA 2 0 D1=.030 D2=.030 H=.062 T=.001
5091-9311E (10/93)
DEF1P 1 EMITTER
7. R.W. Rhea, “Oscillator Design and Com-
puter Simulation” Prentice Hall, 1990, pp
140-143 MLIN 1 2 W=.02L=.030
8. Reference Data for Radio Engineers,
MSTEP 2 3 W1=.02 W2=.1
6 th ed., Howard W. Sams, 1975, p 6-3 MLIN 3 4 W=.1 L=.1
CAP 3 5 C=1
VIA 5 0 D1=.03 D2=.03 H=.062 T=.0014
MSTEP 4 6 W1=.1 W2^W3
MLIN 6 7 W^W3 L^L3!OUTPUT SERIES
MICROSTRIPLINE
4-13
Appendix I. (continued)
CAP 7 8 C=3.3
VIA 8 0 D1=.03 D2=.03 H=.062 T=.0014
MSTEP 7 9 W1=.03 W2=.1
MLIN 9 10 W=.1 L=.2
SLC 10 11 L=.25 C=100 !OUTPUT BLOCKING CAPACITOR
MLIN 11 12 W=.1 L=.2
!IND 6 7 L=19 !OPTIONAL INDUCTOR
RES 4 15R=180 !OUTPUT BIAS RESISTOR
MLIN 15 16 W=.1 L=.1
MSTEP 16 17 W1=.1 W2=.03
MLIN 17 18 W=.03 L=1.1 !OUTPUT BIAS DECOUPLING LINE
SLC 18 19 L=.4 C=1000
VIA 19 0 D1=.03 D2=.03 H=.062 T=.0014
DEF2P 1 12 OUTPUT
INPUT 1 2
DEVICE 2 3 4
EMITTER 4
OUTPUT 3 5
DEF2P 1 5 AMP
FREQ
!SWEEP .8 .95 .05
SWEEP 1 6 .05
!STEP .9
OUT
AMP DB[S11]
AMP DB[S21]
AMP DB[S12]
AMP DB[S22]
AMP NF
AMP K
AMP B1
Appendix II.
AT-32033 900 MHz Low Noise Amplifier Touchstone Output File
4-14
Appendix II. (continued)
4-15
Appendix III.
AT-31011 2400 MHz Low Noise Amplifier Touchstone Circuit File
DIM
FREQ GHZ
IND NH
CAP PF
LNG IN
VAR
W1=.03 !INPUT LINE WIDTH
L1=.15 !INPUT LINE LENGTH
C1=1.3 !INPUT SHUNT CAPACITOR
C2=1.5 !INTERSTAGE BLOCKING CAPACITOR
W2=.02 !INTERSTAGE LINE WIDTH
L2=.178 !INTERSTAGE LINE LENGTH
W3=.03 !OUTPUT LINE WIDTH
L3=.4 !OUTPUT LINE LENGTH
LL1=.02
LL2=.02
CKT
MSUB ER=4.8 H=.062 T=.0014 RHO=1 RGH=0
TAND TAND=.002
MLIN 1 2 W=.1 L=.05
SLC 2 3 L=.25C=10
MLIN 3 4 W=.1 L=.05
MCROS 4 5 6 7 W1=.1 W2=.02W3^W1 W4=.1
MLIN 6 8 W^W1L^L1
MLIN 7 14 W=.1 L=.03
SLC 14 15 L=.25C^C1
VIA 15 0 D1=.03 D2=.03 H=.062 T=.0014
MLIN 5 16 W=.02 L=.68
MSTEP 16 17 W1=.02 W2=.1
MLIN 17 18 W=.1 L=.1
SLC 18 19 L=.25 C=10
VIA 19 0 D1=.03 D2=.03 H=.062 T=.0014
RES 18 20 R=50
SLC 20 21 L=.4 C=1000
VIA 21 0 D1=.03 D2=.03 H=.062 T=.0014
MSTEP 8 9 W1^W1 W2=.1
MLIN 9 10 W=.1 L=.1
MSTEP 10 11 W1=.1 W2=.02
DEF2P 1 11 INPUT
S2PA 1 2 3 C:\S_DATA\BJT\T310113A.S2P
DEF3P 1 2 3 DEV1
4-16
Appendix III. (continued)
S2PA 1 2 3
DEF3P 1 2 3 DEV2
MLIN 1 2 W=.03L=.01
MSTEP 2 3 W1=.03 W2^W3
MLIN 3 4 W^W3L^L3 !OUTPUT SERIES MICROSTRIPLINE
CAP 4 0 C=2
SRL 3 5 R=50 L=1 !OUTPUT BIAS RESISTOR
MLIN 5 6 W=.08 L=.04
MSTEP 6 7 W1=.08 W2=.02
MLIN 7 8 W=.02 L=.68 !OUTPUT BIAS DECOUPLING LINE
SLC 8 9 L=.4C=1000
VIA 9 0 D1=.03 D2=.03 H=.062 T=.0014
SLC 4 10 L=.25C=10 !OUTPUT BLOCKING CAPACITOR
MLIN 10 11 W=.1 L=.1
DEF2P 1 11 OUTPUT
INPUT 1 2
DEV1 2 3 4
Q1EM 4
INTER 35
DEV2 5 6 7
Q2EM 7
4-17
Appendix III. (continued)
OUTPUT 6 8
DEF2P 1 8 AMP
FREQ
!SWEEP .8 .95 .05
SWEEP .1 6 .1
!STEP 2.4
!SWEEP 2.3 2.5 .05
OUT
AMP DB[S11]
AMP DB[S21]
AMP DB[S12]
AMP DB[S22]
AMP NF
AMP K
AMP B1
Appendix IV
AT-31011 2400 MHz Low Noise Amplifier Touchstone Output File
4-18
Appendix IV. (continued)
4-19