Professional Documents
Culture Documents
The history of converting microwave communications, as same time as the first microwave bipolar transistor.
well as other communications technologies, to solid state With the development of semiconductor manufacturing tech-
electronics is a long one. Early advances were first made in nology and the need for lower energy consumption, the metal
receivers, and then in transmitters. Progress in bipolar tran- oxide semiconductor FET (MOSFET) appeared. The
sistor technology and the production of new semiconductor MOSFET, like the JFET, was first developed for applica-
crystals during the 1960’s made possible the development of tions in circuits that demanded high impedance, such as in-
such new microwave diodes as the GUNN and the IMPATT put circuits in analytical instruments. Field effect transis-
(impact avalanche and transit time). For this reason, the tors, particularly the MOSFET, became widely known for
decade might well be called the renaissance of microwave their use as discrete devices in UHF band communications.
semiconductor devices. A series of microwave communica- However, having focused solely on performance for many
tions amplifiers appeared in the first half of the 1970’s which years, nothing in the microwave band appeared on the mar-
used GUNN and IMPATT diodes. They played a leading ket which was superior to the bipolar transistor.
role in the trend toward solid state technology. In the middle Around the time silicon reached its peak as a transistor ma-
of the same decade, a commercially feasible gallium arsenide terial, Schottky barrier type FETs made of gallium arsenide
field effect transistor (GaAs FET) appeared, and the uses of appeared and quickly gained popularity by demonstrating
this device are still increasing. In the latter half of the 1970’s, their high theoretical performance. This new device, known
demands grew for systems more reliable than those using as the Gallium Arsenide Metal Semiconductor FET (GaAs
IMPATT and GUNN diodes. Users were demanding greater MESFET) showed performance far superior to the bipolar
reliability, and those engaged in research and development transistor.
began working toward this goal. The work is still going on This new device provided lower noise and higher gain in
today. established solid state applications. It also provided high
The demands of the industry turned the emphasis away from frequency characteristics previously unavailable from bipo-
creating new devices to developing competition among manu- lar transistors. It is made by using gallium arsenide (group
facturers of semiconductors to produce devices of higher per- III-V); one of the semiconductor com pounds which has been
formance and greater reliability. Commercialization of GaAs researched continually since the latter half of the 1960’s.
field effect transistors led to lower energy consumption and The electron mobility of gallium arsenide is five to seven
smaller microwave components and systems. These are still times that of silicon.
major concerns today. With new developments in infor Gallium arsenide crystal technology was used to produce
mation, communications and applied microwave systems, the GUNN, varactor and Schottky diodes, and proved to be
the GaAs FET has become an indispensable item. far better than silicon in high frequency performance. A
GUNN diode made of silicon would be inconceivable, so the
appearance of the contemporary GaAs FET contributed
1. THE FIELD EFFECT TRANSISTOR (FET) greatly to developing and commercializing the GUNN di-
ode. The reason for this is that even though the GaAs
In 1952, Shockley conceived the structure of the field effect MESFET is a three terminal device, it is simple in structure
type transistor and pointed out that it could be used for am- and its performance depends only on the crystal quality.
plifier devices. Due to manufacturing difficulties, particu- Advances in crystal technology have made the commercial-
larly in production technology, the field effect transistor was ization of FETs possible.
little known until the early 1960’s. The level of technology The GaAs FET is what is generally referred to as a “nor-
at the time made it very difficult for people to understand the mally ON” type device. Its basic difference from the
importance of the FET. But with the development of planar MOSFET is the use of a Schottky barrier at the gate instead
technology, the micro wave semiconductor industry grew with of an oxide layer. This is an extremely important point. In
explosive rapidity. other words, almost no GaAs FET gates are insulated from
There are three major types of FETS. The simplest of the the channels in terms of direct current. Thus, even though
three is the junction FET (JFET). Because of its simplicity GaAs FETs are called “normally ON” type devices, the maxi-
and ease of manufacture, the JFET was the earliest to be mum gate voltage must be zero. It does not use a dielectric
produced commercially. It was put on the market about the like the MOSFET, so if a positive voltage is applied at the
AN82901-1
gate, direct current flows through it. Since the gate is a very Mutual conductance is defined as the ratio of the change in
small piece of metal (0.5,µ—1.Oµ—2.0µ), the gate electrodes direct current to the minor change in voltage between gate
will fuse completely in almost all cases. sources. This is generally described as the square-law char-
Figure 1 shows the properties of a GaAs MESFET. In al- acteristic and is shown in the equation for ID in Figure 2.
most all cases, a linear amplifier circuit biases the GaAs
MESFET. This applies to other circuits as well, but consid- 2
V GS (1)
ering gate bias alone, the range must be from IDSS, i.e., ID = IDSS 1-
VG = 0 V, to IDS = 0 (at pinch off, VG = VP). In this range, the VP
voltage VDS between the drain and the source has little effect
on the current IDS flowing through the channels. By chang- When ID in expression (1) is differentiated with
ing the gate voltage, VG, the drain to source (channel) cur- respect to VGS, the result is
rent can be controlled. Figure 2 shows the transfer charac-
teristics of a GaAs FET with n channels. This FET transfer dID 2IDSS V GS
=- 1- = gm (2)
characteristic is an important basic parameter in circuit de- dVGS VP VP
sign because it sets the bias conditions and operating point. and the mutual conductance for each value of VGS can be
The operating point line in Figure 2 is directly related to the obtained.
mutual conductance, gm.
which is negative, so the FET can be turned on. The value inductance and care must be exercised when using them at
RS is obtained by combining expressions (3) and (4). higher frequencies. Types D and E in Figure 5 require only
one power source. They are compatible with the previously
VGS VP ID described method. An advantage of this method is that if
RS =- =- 1- (5) the source voltage should increase for any reason, that in-
ID ID I DSS crease will be proportionate to the drop in potential caused
by RS, which is connected in series with the source,
Therefore, these are the basic bias principles.
RS X ∆ID
(2) VD POSITIVE VD
(a) VG
VD
ID
(1) VS POSITIVE VS
(2) VS NEGATIVE VS
(c) VG VS
I DSS
RS = b
a
ID
VD ONLY POSITIVE VD
b
RS a RS
ID
(d) VD
VP V GS VG ONLY NEGATIVE VG
I D X RS
(e) VG
Figure 4. Self (Auto) Bias Method Figure 5. GaAs FET Bias Circuits
Figure 5 shows the five general bias types: A, B, C, D, and Here, ∆ID is the increment in drain current caused by the
E. Type A is the previously described dual power source increment in source voltage.
bias method appropriate for use in the higher frequencies. Drain current increase will be automatically suppressed, due
When directly connecting the source to the ground terminal, to the proportionate negative bias on the gate. Generally, if
source inductance can be made relatively small. By using a single source type (self biasing type) is selected, D is used;
this method, higher gain can be obtained and a lower noise if the only available source is a negative one, then E should
factor anticipated in the higher frequencies. be selected.
All other bias methods insert a bypass capacitor into the Figure 5 shows the order for adding bias when a dual power
source. Even if the high frequency performance of the by- source is used. This is to prevent, as much as possible, a
pass capacitors can be guaranteed, there will always be a
loss (tan δ) resulting from the material’s dielectric proper-
ties. Even with chip capacitors, there is always some
AN82901-1
large current from flowing through the FET. The GaAs FET at the center of the bias line and wanders off in some direc-
being discussed here generally has a high mutual conduc- tion, the wave form will be distorted. Also, the amplification
tance, giving it excellent frequency characteristics. If the operation will be degraded, and DC will flow through the
gate voltage is near zero, gm is at a maximum and oscillation gate. In a similar manner, the maximum variation in voltage
may occur. For this reason, a bias scheme should be adopted between the gate and the source when there is no signal is
that first applies a negative bias to the gate and then turns on
the drain with a positive bias. There are slight differences ∆VGSQ = VGSQ(max) - VGSQ(min)
between a bias circuit for a small signal GaAs FET and one
for a power GaAs FET, but the above methods can be con- IQ is a function of both the temperature, T, and VGSQ
sidered for both.
Moving one step further, let’s examine a practical bias cir- IQ = f (T, VGSQ )
cuit. It is important to:
1. Obtain the necessary operating voltage and operating If, with a variation in temperature of δT, IQ changes by only
current, and δTQ .
2. Ensure a high level of stability.
δIQ = ∂IQ . δT+ ∂IQ . δVGSQ
∂T ∂VGSQ
RD
∂IQ . δT - gmδIQ R. S
=
∂T
and from this
RS
I DSS(max)
∂IQ .
δIQ (1 + gmRS) = δT
∂T
is obtained.
RS
I DSS(min) If RS = 0, that is, if there is no series resistance, then
I Q(max)
I Q(min)
∂IQ . δT = δI (0)
Q
∂T
V P(min) V P(max) VGS(max)
represents the drift. The bias stability coefficient is defined
VGS(min)
as
Figure 6 shows the bias graph, indicating all possible bias = δ IQ = 1 (6)
points. Bias will be in the range determined by δIQ (0) 1 + gmRS
This means that fluctuation in the entire circuit is reduced by
VGSQ = IQRS means of negative feedback due to RS. This effect is particu-
larly important when using small signal FETs.
Given the bias range, the operating point for the maximum Figure 7 gives an example of a bias circuit. Since IDSS and
amplitude value, P, and the operating point where there is VP often vary between devices, it is important that the bias
no signal, Q can be obtained. The change in IQ from point P circuit can absorb such variation, the optimum operating point
to point Q, or ∆IQ, becomes can always be established, and operation is stable regardless
of ambient temperatures. The circuit in Figure 7 is basically
∆IQ = IQ(max) - IQ(min) a fixed bias circuit and not a self bias circuit.
This shows that with a variation in IQ, the drain voltage will
change only as RD . ∆IQ. If the signal midpoint is not fixed
AN82901-1
Feedback is applied to the circuit by the insertion of RS into If there is no signal, the drain current ID is
the source.
2
vgs (7)
i D = IDSS 1 -
Vp
VR L1
+Vcc
Vout
C1
RF
Vin
VDD
RS
Rs
V
KF=
4(VGSQ PP)(1+gmRF)
IM= V1V2
2(VGSQ VP)(V12+V22) 1/2(1+gmRF)
Figure 7. A Low Cost, Stable Bias Circuit Figure 8. Improving the Distortion Factor
Using a Series Feedback Resistance RF
As stated previously, since the voltage between the gate and
the source is VGS = -IDRS in a self bias circuit, and if ID is Adding a simple sine wave voltage (Vsin wt) to the non-
known, RS can be readily determined. In such a fixed bias signal voltage results in
circuit as the previously mentioned dual source type, the gate
voltage is selected independently of RS and ID. Thus, it is υgs = VGSQ + Vsin wt
understood that the value RS can be higher than that when
using a self-bias circuit, and the stability coefficient S can If the expression is rearranged and substituted into the fol-
therefore be improved. lowing, then the harmonic distortion KF is
= V (8)
In this section, examples and explanations of GaAs FET 4(VGS - VP)
applications will be given. The first issue is the argument as
to whether or not the FET is better than the bipolar transis- V is the maximum amplitude of the signal.
tor because of the FET’s cross modulation characteristic. As expression (8) shows, the distortion factor approaches its
Discussion will then turn to applications of the small signal minimum as VGSQ approaches zero. However, as the input
FET and the large signal or power FET. signal increases and enters the realm of forward gate bias,
quite naturally the distortion factor increases.
1. Distortion Factor Next, we will consider what happens to the cross modula-
The transfer characteristic of a GaAs FET can be approxi- tion (intermodulation) produced when two sine wave sig-
mated using a quadratic equation, as was shown in expres- nals are amplified at the same time. That is
sion (1). If the transfer characteristic could be perfectly rep-
resented by such an equation, then the second harmonic will VGS = VGSQ + VI sin wt + V2 sin wt
increase to its maximum and will actually include the higher
order harmonic components. To determine the value of the Since the output current includes the sum and difference
second harmonic for a basic frequency, substitute the total components of two sine waves, crossmodulation
input voltage and total output current into the equation.
AN82901-1
(intermodulation) distortion, IM, is defined as follows: in terms of noise, gain and output-power saturation char-
acteristics. Other than their primary use in ultra-high fre-
Relative value of quency amplifiers, such as those for electronic countermea-
IM = Relative value of sures (ECM), most small signal FET applications are in low-
noise amplifiers. They are used in both line-of-sight and
cross modulation component over-the-horizon microwave communications, and in earth
fundamental harmonic stations communicating with satellites.
V1V2 A low noise amplifier is designed by minimizing the noise
= (9) measure, M, shown in expression (12).
2(VGSQ VP)(V12+V22) 1/2
Even here, we see that the distortion factor decreases as bias
M+1=1+ NF - 1 (12)
is brought closer to VGSQ = 0. The main cause of the distor-
1 - (1/G)
tion is the curve (non linearity) in the transfer characteristic
line, but there are other causes as well.
Distortion can also be caused by the change in output con- NF is the amplifier noise factor.
ductance, gd, related to the operating point and the drain G is the amplifier gain.
voltage, VDS. To improve distortion, carefully select gm and In single-stage amplifiers, the general procedure for match-
gd, which to a certain degree work to reverse this effect. ing input circuits is to mimimize the noise factor, NF; for
Another point to consider is that since distortion can be pro- matching output circuits, the gain is maximized.
duced internally, KF and IM can both be improved by apply- From observations of the input-output impedances of a GaAs
ing feedback to the circuit. FET, it is noted that there is generally a difference in imped-
As shown in Figure 8, by adding a resistance, RF, in series ance between maximum gain and minimum NF. This dif-
with the source resistance, the distortion factor for the non- ference is particularly apparent at lower frequencies. As the
bypass feedback is calculated, using expressions (10) and (11). frequencies go higher, the difference seems to decrease. The
noise factor, which is a function of device gain, will be low
when the gain is maximized at high frequencies. However,
V of the commercially available GaAs FETs having character-
KF= 4(VGSQ VP)(1+gmRF) (10)
istics which allow a gain of 8 to 10 dB or more, there is still
a difference between the impedance at maximum gain and
V1V2 the impedance at miminum NF. This difference will be seen
IM= 2 2 (11)
2(VGSQ VP)(V1+V2) 1/2(1+gmRF) until the frequency range approaches the X band.
+1.5
sible to have a wide-band amplifier with a low distortion fac-
6
+0.
.0
tor by initially designing the amplifier for high gain and tun-
+2
5
S2 4
ing the gain to its optimum level by using feedback.
.4
12 2 6
+0
S11 GHz
GH +3
3
1
z
2
+4
11
2.0
1.0
0.4
been marketed.
0
8
The greatest advantages for these devices are found in the
higher frequency bands. Compared to the silicon bipolar 7
-0.2
2
-5
transistors and tunnel diodes, GaAs FETs are far better 10
6 -4
3
-3
9
.4
4
-0
2
8
7
5
0
6
-2.
-0.6
3
-1.5
4
-0.8
-1.0
Figure 9 shows S11 and S22, indicating input output imped- For these bandwidths, the NF can be reduced to its lowest
ance for the NE24406 and the signal source impedance when absolute value. If the widest possible bandwidth is the ob-
the noise factor, NF, is minimized. Circuits can be matched jective, then reduction of the NF to its absolute minimum is
using these impedances. As mentioned previously, input not possible throughout the band. In the former amplifier,
circuit matching is accomplished by matching to ZFopt; and the most important consideration is reducing, as much as
output circuit matching is accomplished by matching to S22. possible, the loss in the impedance matching circuit. The
In the design of a multi-stage amplifier, the first and second Tchebycheff filter type impedance matching circuit in Fig-
stages are designed so that (M + 1) will be minimized, and ure 10(a) uses a large number of components and is not well
the third and subsequent stages are made in such a way that suited for obtaining a low loss matching network.
their gain is maximized. Or, alternately, the amplifier can As Figure 9 shows, both S11 and S22 go from capacitive to
be designed according to the characteristics determined by inductive with increasing frequencies. Consequently, S11
the frequency and the device. and S22 can use the circuit in Figure 10(b) for frequencies
One impedance matching circuit, shown in Figure 10(a), is which exhibit a capacitive response, or the circuit of Figure
the well known Tchebycheff filter type multi-stage imped- 10(c) for the frequencies which exhibit an inductive response.
ance matching circuit. The distinctive feature of this circuit In Figure 10(b), for frequencies slightly higher than the
lies in its applicability to the input/output circuits of wide- amplifier’s center frequency, the GaAs FET input or output
band amplifiers. In many cases, the circuit is used in ampli- has a resonance of either L1 or L2 (with a reactance of 0),
fiers covering the band range 8 GHz to 12 GHz and above. which is then matched to the desired impedance by the λ/4
Although this type of matching network was originally ap- impedance conversion circuit. This is only one of the count-
plied to comparatively narrow bands (such as a 500 MHz less possible impedance matching circuits. However, this is
bandwidth at 4 GHz), the matching circuits shown in Figure a very effective design method for low noise amplifiers re-
10(b) and (c) are now believed to be best suited for bands quiring a band ratio of 10 to 15 percent when the amplifier’s
having approximately a 10 to 15 percent ratio to the center frequency is 4 GHz or greater.
amplifier’s center frequency.
_ l 34 _ l 34 _ l 23 _ l 23 _ l 12 _ l 12
jX 34 jX12
(a)
l2
l1 λ/4 λ/4
C2
λ/4 λ/4
C1
(b) (c)
3. Low Noise Amplifiers in the 4 GHz Band 0.8 mm thickness teflon glass fiber substrate and transistor
Figure 11 is an example of a two-stage low-noise amplifier leads with lumped constant inductance. Figure 12 shows
for the 3.7 to 4.2 GHz band using either the NE21889 or the the schematics. Figure 13 shows the gain and noise factor
NE72089. The matching circuit in Figure 11 is based on the normally obtained using this circuit.
same idea as Figure 10(b). It uses a microstrip with an
10 pF 10 pF 10 pF 28.0 mm
IN OUT
69.6 mm
1000 pF SHIM STOCK TO GROUND
4 PLACES ALL 4 SOLDER PADS
Figure 11. PCB Layout of a 2-Stage Amplifier in the 3.7 GHz-4.2 GHz Band
λ/4
TRL2(Z0)
TRL1 Z0 C3
Z0 TRL3
C2
λ/4
λ/4
C1
λ/4 λ/4 λ/4 SST4
C13 C14 C15
SST1 SST2 SST3
C4 C5 C6 C7 C8 C9 C10 C11 C12
Vd2
Vg1 Vd1 Vg2
Figure 12. Schematics for a 3.7 GHz-4.2 GHz Band 2- Stage Amplifier
AN82901-1
4.625 cm
24
G 4
NOISE FACTOR, NF
THROUGH-HOLES
GAIN,G (GHz)
20
3
16
2
12 NF 1
IN OUT 8
1.1 1.2 1.3 1.4 1.5
FREQUENCY (GHz)
0.85 cm
Figure 16. NE72089 Low-Noise Amplifier
MATERIAL 0.8 mm THICK TEFLON GLASS When this amplifier was designed, a 4 pF blocking chip ca-
pacitor was used as part of the matching circuit, as shown in
Figure 14. PCB Layout for 1.0-1.4 GHz Amplifier Figure 15. The input matching circuit was designed chiefly
AN82901-1
to obtain matching for the noise factor. The data used for 5. Applications to Other Amplifiers
the NE72089 is shown below:
(A) A technique of attaining wide band performance by
inserting a source inductance.
f ΓFopt* Fmin G NF @ 50 Ω L. Neven, et al., have proposed a type of feed back circuit for
1 GHz 0.59<32.5 0.6dB 17.5 dB 1.73 dB relatively low frequency GaAs FET amplifiers. For amplifi-
1.3 GHz 0.65<47.0 0.6 dB 15.0 dB 1.74 dB ers operating in the 1 GHz to 2 GHz range, these circuits
VDS = 3 V, IDS = 10 mA guarantee a low noise factor as well as good input-output
impedance.
*(I' Fopt signifies the reflection coefficient obtained when comparing the
Figure 18 shows what is considered the universal GaAs FET
signal course impedance with 50 Ω when NF is minimized)
equivalent circuit. In this circuit, drain noise current, Idn, is
The bias circuit used to actuate this amplifier is shown in referred to the input by the Vander-Ziel theory, then the gate
Figure 17. The figure shows a bias circuit for a two-stage source resistances, Ri, becomes noiseless and, RS remains
amplifier. For a single stage amplifier, only one half of the noisy. The noise currents at the gate and drain have been
redundant portion of the circuit would be used. This circuit removed from the two-port. The circuit in Figure 18 is bet-
is a constant voltage, constant current type bias circuit, one ter handled by computers, where the noise and signal are
of the most applicable examples of circuits for use in small- compared as a function of the externally added source in-
signal amplifiers. ductance. In this instance, it is easier to consider the equiva-
lent circuit shown in Figure 19.
15V 2N3643
DUAL
15V
POWER 330
330
SUPPLY
1
2 A1 R2 R3
15V (3V) 500 (3V) 500 (3V)
V DS1 V DS2
100K 100K
10K 100K 4.7µF
2 A1
1
R1 4.3K 4.3K
500 1
2 A2 V GS1 1
2 A2 V GS2
(3V) 100K 6.8V 6.8V
3.3K 3.3K
.01µF .01µF
2.2K 6.8V
100K 100K
5.6K
LG RG Cgs Rd Ld en
+ -
Ci
gd
Ri
g mV I dn IC NOISELESS
Cds TWO-PORT
Iu
RS
LS
LS '
EXTERNAL SOURCE INDUCTANCE
EXTERNAL SOURCE INDUCTANCE
Figure 18. Universal GaAs FET Equivalent Circuit Figure 19. Cascade Connection of Device's Noise
and S-Parameter Circuits
AN82901-1
Figure 19 shows the cascade connections of the device’s noise designed so that matching is obtained for the input-output
and S-parameter equivalent circuits. The circuit should be impedance given to this equivalent circuit. After the design
goals have been chosen, the noise figure, NF, and impedance
matching networks are calculated. The circuits and charac-
10.6 nH
teristics of an amplifier designed by this method are shown
10 nH
in Figures 20(a) and 20(b).
0.8 pF
0.13 11.4 (B) Ultra-wide band amplifiers using GaAs FETs.
1.5 nH pF nH
The method normally used to achieve wide-band amplifica-
tion involves applying negative feed back to the transistor.
NOISE FACTOR, NF (dB)d GAIN, G (dB)
1.5 n H
Figure 21. Balanced Amplifier Allowing
Wide-Band Amplification
GAIN, G (dB)
15
The use of GaAs FETs with negative feedback is an effective
14 method of ensuring low cost and simplicity for bandwidths
G
13
exceeding 10 octaves.
Figure 22 shows the basic circuit for negative feedback am-
12 plifiers. For the sake of simplicity, the DC bias circuit and
NOISE FACTOR, NF (dB)
FREQUENCY (GHz)
(B)
2
gmZ 0
R2
S 11 = 1 1 -
∆ R 2(1 + gmR1 ) (14)
OUT
IN 2
gmZ 0
S 22 = 1 1 -
R1 ∆ R2 (1 + gmR1 ) (15)
2Z0
S12 = 1
Figure 22. Basic Feedback Amplifier Circuit ∆ R2 (16)
-2gm Z0 2R0
S21 = 1 +
∆ 1 + gm R1 R2 (17)
GATE DRAIN
2
2Z0 gm Z 0
V gs gm V gs ∆ =1+ +
R2 R2 (1 + gm R1)
The admittance matrix can be determined by using the From this, the following expression can be derived:
following expression. g m Z 02
1 + = g m R1 = (18)
R2
Z0
S12 = (20)
R2 + Z 0
I1 R2 I2
Then, using expression (18), R1 is obtained as follows
gm Vg 2
Z0 1
Vg R1 = – gm (21)
V1 V2 R2
R1
Since the transconductance, gm, in a GaAs FET is not usu-
ally large enough, the equation S11 = S22 = 0 will not be
Figure 24. Simplified Feedback Amplifier satisfied. However, the range of the gm values needed to
Equivalent Circuit satisfy the expression can be calculated. If R1= 0 then
1 - (-3.16)
= 0.082 S 100
50 4GHz
8
Figure 26. Te vs. Ta for the NE67383/NE71083
6
7. Power GaAs FET Applications
GAIN, G (dB)
I DS
iD B VG =0
i DB A
i DA P VGB
VGB
t
t1
VDS
VD
t1
VDA
VDB
t
Figure 27. Simplified Load Line for a Class A GaAs FET in Small and Large Signal Operation
(A) Introduction
An important factor to consider when working with Power taken at this time since the amount of gate current has a
GaAs FETs is power GaAs FETs are basically a group of bearing on whether or not the FET will be destroyed.
small signal GaAs FETs connected in parallel so that they
can handle a large power source. Therefore, they have an
extremely high transconductance. Another important and 10
(B) Bias Circuits One thing that is very basic when biasing power GaAs FETs
Table 1 shows the maximum allowable gate current for is the need for increased impedance of the gate bias circuit.
each of NEC’s high power GaAs FETS. Care should be The example in Figure 29 shows a type of feedback circuit
taken at all times when planning bias circuits and input power incorporating a resistor in series with the bias circuit which
to ensure that these values are not exceeded. produces a drop in potential when the current across the gate
begins to increase, and adds to the gate series bias. By in-
NE8001 3.0 mA serting a low pass filter, such as an RFC or capacitor, unnec-
NE8002 6.0 mA essary oscillation or phase shift will be prevented.
NE8004 10.0 mA Another important factor is the order in which bias is ap-
NE8008 15.0 mA plied. It is generally disadvantageous, both thermally and
NE3716 20.0 mA electrically, to self-bias a power GaAs FET. Therefore, a
NE9000 0.5 mA positive and negative dual power source should be used to
NE9001 1.3 mA bias the drain and gate separately. For a GaAs FET, maxi-
NE9002 2.6 mA
mum current will flow when the gate voltage is at zero. With
a high output level FET, IDSS is sometimes greater than sev-
NE9004 5.0 mA
eral amps and since typical gm is extremely high, oscillation
NE9008 8.0 mA
will occur and the FET will be destroyed either thermally or
NE8681 3.0 mA electrically. For this reason, any appreciable flow of high
NE8682 6.0 mA current must be prevented from reaching the drain. The
NE8684 10.0 mA bias method used must be one whereby negative voltage is
NE8688 15.0 mA applied to the gate first, after which positive voltage is ap-
NE8691 1.3 mA plied to the drain. The following is a detailed example of
NE8692 2.6 mA this method.
NE8694 5.0 mA (1) Add a negative potential to the gate which is
close to -5V or to the pinch off voltage.
Table 1: Maximum Allowable Gate Currents for the NEC (2) Specify the drain voltage in such a way that
Power GaAs FET Series it moves as rapidly and smoothly as possible from zero volts
to the preset VD. If there is no possibility of a surge voltage
being produced and the drain voltage is completely stable,
add the preset voltage instantaneously.
Figure 30 gives three recommended bias circuits for biasing
power FETS. Figure 30(a) is an excellent bias circuit be-
V R 1000 pF RFC TO GATE cause there are no limitations requiring that either positive
or negative voltage be applied to the circuit first. Drain volt-
age, VD, will reach its preset value according to the time
1 µF 4.7 pF
VR
STABILIZED
V VD V VD V VD
-
R D1 +
C
STABILIZED
V V G STABILIZED VG
V
V VG
(a) (b) (c)
Q3 .1 2W
A
+ VDS
S1A 2A Q2 1K
3 1 100 .0003 47µF
+ 6 4
10µF Q1
9 IC1 5
DUAL
±15V
5V R1 .001
2K 20K 8 7
POWER 6.8K 2 10
SUPPLY .01µF
S1B
330 22 6 7
5 8
R2 R3
10µF 3 500 33
+ IC2
2.4K 2 .0003 + 47µF
VGS
1 9
Q1 - SK3448
Q2 - 2N706 10µF 2.5K
Q3 - 2N3771 +
IC1- MC1469
IC2-LM304
constant determined by C and R. When the power source is center frequency at which the IMN circuit is optimized and
OFF, the electrical charge will pass through diode D1, and tested. For example, -4 is at 4 GHz (3.5-4.5 GHz) and -6 is
return quickly to the power source side, thus the FET can at 6 GHz (5.5-6.5 GHz).
then be safely biased. Figures 32 to 68 relate typical characteristics and illus-
Figure 31 shows a bias circuit recommended for biasing a trate application circuits using an 0.8mm thick teflon glass
multistage (5 stages) power GaAs FET amplifier. circuit board. The units used in these charts are millimeters
(mm). Figure 69 gives AM-PM conversions for NE800,
9. Examples of C-Band Application Circuits Using NEC’s NE371, and NE868 series. This data is particularly effec-
Power GaAs FET Series tive when designing a PCM transmitter. Figure 70 gives the
With the exception of the devices with the 96 package, NEC’s standard characteristics for the third order intermodulation
C-Band GaAs FET series using multiple chips contain in- of NE800898-7H and NE868898-6.
ternal matching network (IMN) circuits within the packages. This brief report explains current GaAs FET technology.
The IMNs are not intended to match the devices to 50Ω but Improvement of device characteristics is a major goal for
to facilitate and reduce the external matching networks. The both applied technology and research in electronics. GaAs
GaAs FET series which are in 98 packages have both input FET reliability has already been confirmed and there is little
and output matching circuits close to VSWR, 1:2.5. The doubt that FETs will eventually be used at frequencies of 40
devices in 95 and 98 packages which have internal match- GHz and higher.
ing circuits are identified by the numbers -4, -5, (-5H), -6,
-7, (-7H), and -8. Each of these numbers indicates the
AN82901-1
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm
IN OUT
28
20.0 20.0 P in = 18 dBm
3.0 4.0
3.0 26 f = 5.5GHz
Pout (dBm)
24
3.0
3.0
17.0
2.1
22
20
12 14 16 18 20 22 24
Pin (dBm)
Figure 32. NE800196, 5 GHz Circuit
4.9 5.0 5.1 5.2 5.3 5.4 5.5
f (GHz)
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm
28
P in = 18 dBm
IN OUT
26
20.0 20.0
Pout (dBm)
f = 6.5 GHz
4.0 4.0
9.0 2.0 24
22
4.0
4.0
17.0
2.1
20
12 14 16 18 20 22 24
Pin (dBm)
IN OUT 28
f = 6.5GHz
Pout (dBm)
20.0 20.0
3.0 4.0 26
24
5.0
2.0
17.0
2.1
22
2.0
4 16 18 20 22 24 26
Pin (dBm)
P in=25 dBm
32
IN f=4.2GHz
OUT
Pout (dBm)
30
20.0 20.0
4.0 2.0 4.0
28
26
5.0
3.0
17.0
2.1
24
3.0
16 18 20 22 24 26 28
Pin (dBm)
Pout (dBm)
5.0 3.0
30
f = 5.5 GHz
28
4.0
4.0
2.1
17.0
26
4.0
4.0
24
16 18 20 22 24 26 28
Pin (dBm)
Figure 36. NE800495-5 Circuit
4.4 4.5 4.6 4.7 4.8 4.9 5.0
f (GHz)
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm
34
IN OUT
P in = 26 dBm
20.0 20.0 32
3.0 5.0
Pout (dBm)
30
f = 6.5 GHz
3.0
4.0
2.1 28
17.0
26
24
16 18 20 22 24 26 28
Pin (dBm)
IN OUT
32
20.0 20.0
P in = 26 dBm
2.0 5.0 5.0 3.0 2.0
Pout (dBm)
2.0 11.0 30
f = 7.2 GHz
28
4.0
3.0
3.0
3.0
17.0
2.1
26
3.0
24
16 18 20 22 24 26 28
Pin (dBm)
Figure 38. NE800495-7 Circuit 6.6 6.7 6.8 6.9 7.0 7.1 7.2
f (GHz)
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm 34
IN OUT
20.0 20.0 32
30
f = 8.4GHz
28
4.0
3.0
17.0
2.1
26
24
16 18 20 22 24 26 28
Pin (dBm)
Pout (dBm)
3.0 3.0
10.0 14.0
34
f = 4.2GHz
5.0
4.0
17.0
2.1
32
4.0
30
22 24 26 28 30 32 34
Pin (dBm)
Figure 40. NE800898-4 Circuit
3.7 3.8 3.9 4.0 4.1 4.2 4.3
f (GHz)
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm 38
P in = 29 dBm
IN OUT 36
20.0 20.0 f = 5 GHz
Pout (dBm)
3.0 6.0 4.0 6.0 3.0
11.3 34
1.0
3.0
3.0
32
17.0
2.1
30
22 24 26 28 30 32 34
Pin (dBm)
34
2.0
3.0
17.0
f = 5.5 GHz
2.1
2.1
32
4.0
1.0
30
22 24 26 28 30 32 34
2.0 1.0
Pin (dBm)
Figure 42. NE800898-5H Circuit 5.0 5.1 5.2 5.3 5.4 5.5 5.6
f (GHz)
34
1.0
2.0
f = 6.4 GHz
17.0
2.1
32
30
22 24 26 28 30 32 34
Pin (dBm)
Figure 43. NE800898-6 Circuit
5.9 6.0 6.1 6.2 6.3 6.4 6.5
f (GHz)
AN82901-1
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm 38
IN OUT
P in = 29.5 dBm
36
20.0 20.0
6.0 9.0 7.0
Pout (dBm)
34
f = 7.2 GHz
32
3.0
2.0
17.0
2.1
30
28
22 24 26 28 30 32 34
Figure 44. NE800898-7 Circuit Pin (dBm)
20.0 20.0
P IN = 30 dBm
3.0
2.0
2.0 36
2.0
2.0
2.0
5.0 5.0
Pout (dBm)
1.5
34
3.0
f = 7.4 GHz
2.1
17.0
32
30
22 24 26 28 30 32 34
Pin (dBm)
Figure 45. NE800898-7H Circuit
7.2 7.3 7.4 7.5 7.6 7.7 7.8
f (GHz)
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm 38
IN OUT
20.0 20.0 36
3.0 4.0 3.0
10.0 Pin = 30.5dBm
Pout (dBm)
34
f = 8.4GHz
32
3.0
2.0
17.0
2.1
30
28
22 24 26 28 30 32 34
Figure 46. NE800898-8 Circuit Pin (dBm)
36
5.0
4.0
f = 4.2 GHz
17.0
2.1
34
4.0
5.0
32
24 26 28 30 32 34 36
Pin (dBm)
Figure 47. NE371698-4 Circuit
3.7 3.8 3.9 4.0 4.1 4.2 4.3
f (GHz)
AN82901-1
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm 40
IN OUT P in = 32 dBm
20.0 20.0
38
7.0 10.0 3.0 3.0 4.0
f = 5.5GHz
Pout (dBm)
36
34
3.0
3.0
17.0
2.1
32
3.0
2.0
30
24 26 28 30 32 34 36
Pin (dBm)
20.0 20.0 38
Pout (dBm)
36
4.0
4.0
34
3.0
17.0
2.1
32
30
24 26 28 30 32 34 36
Pin (dBm)
Figure 49. NE371698-6 Circuit
5.9 6.0 6.1 6.2 6.3 6.4 6.5
f (GHz)
36
34
5.0
4.0
17.0
2.1
32
30
24 26 28 30 32 34 36
Pin (dBm)
Figure 50. NE371698-7 Circuit 6.6 6.7 6.8 6.9 7.0 7.1 7.2
f (GHz)
36 f = 7.7 GHz
34
4.0
2.0
17.0
2.1
32
30
24 26 28 30 32 34 36
Pin (dBm)
Figure 51. NE371698-7H Circuit 7.1 7.2 7.3 7.4 7.5 7.6 7.7
f (GHz)
AN82901-1
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm 30
IN OUT
Pout (dBm)
30
3.5 3 4 3.8
14 6
20
f = 4 GHz
21 10.5 5
9 2
10
0 10 20
Pin (dBm)
Figure 52. NE868196, 4 GHz Circuit 3.5 4 4.5
f (GHz)
Pout (dBm)
6.5
20
f = 5 GHz
15
10
2
10
0 10 20
Pin (dBm)
4.5 5 5.5
Figure 53. NE868196, 5 GHz Circuit f (GHz)
8.3 5.2
f = 6 GHz
20
6.8
6.5
2
15
10
0 10 20
Pin (dBm)
5.5 6 6.5
20
f = 7.2 GHz
VDS = 9 V
5
6
15
ID = 125 mA
10
0 10 20
Pin (dBm)
6.5 7 7.5
Figure 55. NE868196 Test Circuit (6.5 GHz and 7.5 GHz) f (GHz)
AN82901-1
IN OUT
30
Pout (dBm)
8.5 6.5
11.5
2 9 2 30
Pin = 22 dBm
4
21
9.5
15
2
6
f = 4 GHz
20
10 20 30
Figure 56. NE868296, 4 GHz Circuit Pin (dBm)
3.5 4 4.5
f (GHz)
IN OUT
Pout (dBm)
30
3.5 3
20 4 5 6 8.5 5
30
5.5
21
9.5
10
2
Pin = 22 dBm
f = 5 GHz
20
10 20 30
Pin (dBm)
Figure 57. NE868296, 5 GHz Circuit 4.5 5 5.5
f (GHz)
IN OUT
20
2.5
Pout (dBm)
6.5
Pin = 22 dBm
2 2 7.5 2 30
4
15
5
6
f = 6.5 GHz
4
20
10 20 30
Pin (dBm)
Figure 58. NE868296, 6 GHz Circuit
5.8 6 6.2 6.4 6.6
f (GHz)
IN OUT
Pout (dBm)
20
3 4.5 3 2.7 3
8.7 30
4
Pin = 22 dBm
15
6.5
6.7
2
f = 7.2 GHz
20
10 20 30
Pin (dBm)
6.5 7 7.5
Figure 59. NE868296 Test Circuit (6.5 GHz and 7.5 GHz) f (GHz)
AN82901-1
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm 40
VDS = 9.0 V
IN OUT
Pout (dBm)
28.5 28.5 Pin = 25 dBm
2.5 2 2 1.5
3 18.0
30
f = 4.0 GHz
4.5
10.5
9.5
7.5
20
10 20 30
Pin (dBm)
3.5 4.0 4.5
Figure 60. NE868495-4 Circuit
f (GHz)
5
21.5
3.7
7.7
2.1
8.5
2.1
Figure 61. NE868495-5 Circuit
9.5 14
3.5
Pout (dBm)
3
2.5 2 Pin = 26 dBm
30
7.0
6.5
6
f = 6.5 GHz
20
10 20 30
Pin (dBm)
6.0 6.5 7.0
20.0 11.0
9.0 9.0 f = 7.0 GHz
Pout (dBm)
6.7
2.1
VDS = 9.0 V
20
10 20 30
Pin (dBm)
6.5 7.0 7.5
Figure 63. NE868495-7 Circuit f (GHz)
AN82901-1
IN
MATERIAL -TEFLON GLASS
OUT
THICKNESS: 0.8mm
20
6.2 11
2.4
2 1.5 2.6 2 2
4.5
3.5
7
2.1
15
5
3.3
2.1
Figure 64. NE868495-8 Circuit
2.1
5
21.5
2.1
4
4.3
Pin = 29 dBm
30
5.5
f = 6.0 GHz
4.0
20
10 20 30
Pin (dBm)
5.5 6.0 6.5
f (GHz)
Figure 66. NE868898-6 Circuit
MATERIAL -TEFLON GLASS
THICKNESS: 0.8mm
IN OUT
40 VDS = 9.0 V
25 25
3.5
12 2 14.5
Pout (dBm)
P in= 29 dBm
30
3.6
8.8
f = 7.2 GHz
20
10 20 30
Pin (dBm)
6.5 7.0 7.5
Pout (dBm)
20
2 3
1.5 Pin = 30 dBm
30
20 dBm
3.7
2.1
2.1 25
15
4.8
VDS = 9 V
20
10 15 20 30
Pin (dBm)
7.5 8.0 8.5
f (GHz)
6 6
NE800898-4 NE371698-4
4 4
PHASE (deg.)
PHASE (deg.)
2 2
0 0
-2 -2
20 22 24 26 28 30 22 24 26 28 30 32
6 6 6
NE868196
(NE868199) NE868495 NE868898
4 4 4
PHASE (deg.)
PHASE (deg.)
PHASE (deg.)
2 2 2
0 0 0
-2 -2 -2
-4 -4 -4
0 5 10 15 20 25 10 15 20 25 30 35 10 15 20 25 30 35
Figure 69. AM-PM Conversion for the NE800, NE371 and NE868 Series
AN82901-1
50 10 40
Pout
45 0 30
f = 7.7 GHz, ∆f = 10 MHz
VDS = 9 V
IDS = 1.0 A
40 -10 20
Pout IM3
IM3 (dB)
30 -30 0
IDS = 1.2A,Pin =30 dBm
IDS = 0.8A,Pin =30 dBm
IM3
25 -40 -10
NE800898-7H NE868898-6
20 -50 -20
15 -60 -30
15 20 25 30 35 10 20 30 40
Pin (dBm) Pin (dBm)
Figure 70. Third Order Intermodulation Distortions for the NE800898-7H and NE868898-6
References
[1] Masuo Fukuda, et. al., [5] L. Nevin, et. al., “L-Band GaAs FET
”Low Noise GaAs FET Ampliliers,” MW77-21. Amp.,” Microwave Journal, Apr. 1979.
[2] Charles A Liechti, et. al., MTT-22, No. 5, May 1974 [6] Eric Ulrich, “Use Negative Feedback to Slash
Wideband VSWR,” Microwaves, Oct. 1978.
[3] Herman Sfatz, et. al., ED-21, No. 9, Sept. 1974.
[7] K. Honjo, et. al., “Broadband Internal Matching of
[4] P. Bura, Electronics Letters, Vol. 10, No. 10, p. Microwave Power GaAs MES FETs.,” IE3, Trans,
181-,16th, May 1974. Mit. Jan. 1979.