Professional Documents
Culture Documents
AV02 3792EN AN - 5559 MGA 633P8 11sep20120
AV02 3792EN AN - 5559 MGA 633P8 11sep20120
2
Capacitors C3-C5 decouple RF from the bias lines. By se- Vdd
lecting C4 so that its reactance is approximately 6 Ω at f0,
it also works with R1 to roll-off the gain below f0. Rbias
C5 C6
When it is required to switch the LNA via a logic-level con-
trol signal applied to pin 1 (Vbias), C3 should be reduced R1 R2
to the lowest capacitance value that will still provide a low C3 C4
reactance at f0 , for example 100 pF at 450 MHz. This low
C3 value will speed up the switching time that is limited L2
by the RBIAS x C3 time constant. Turn-on time is typically L1 P1 P8
R3
several microseconds for small C3 values in the pF range. P2 P7
Input Output
P3 P6
The printed circuit board (PCB), which measures 21.5 mm C1 C2
P4 P5
x 18 mm x 1.4 mm, is made from 10 mil Rogers RO4350 mi-
Centre tab
cro-strip with a co-planar ground and 1.2 mm FR4 mate-
rial glued to the bottom as stiffener. RF connections were MGA-633P8
made through edge-launch SMA-to-micro-strip transi- Figure 3. LNA Circuit Schematic
tions (Emerson Network Power/Johnson Component P/N
142-0701-856). The DC supply was connected via a 2-pin
straight PCB header.
The LNA demonstration board is powered from a single
5V power supply. At the nominal Idd of 55 mA, approxi-
mately 0.5V is dropped across R1, giving a Vds of approxi-
mately 4.75V.
The component values in Table 1 were used for the dem-
onstration board and circuit but may be slightly different
if either a different layout or PCB material is used.
Table 1. MGA-633P8 demonstration board part list
Position Value Description
C1 100 pF Murata GRM15
C2 18 pF Murata GRM15
L1 82 nH Toko LL1005
L2 68nH Toko LL1005
C3, C4 22 pF Murata GRM15
C5, C6 10 nF Murata GRM
R1 100 Ω
R2 4.7 Ω Figure 3b. Demoboard component placement diagram
R3 270 Ω
Rbias 6200 Ω
SMA Connectors 142-0701-856 Johnson
Note: All components are 0402 size unless otherwise noted
3
The demonstration LNA was simulated in Agilent Tech- tool lets the designer provided weighed values to which
nologies ADS2009U1 software in order to obtain the parameters are most important to the end design. The
matching network components initial values. The MGA- passive models were then replaced with components
633P8 was modelled using Touchstone formatted scatter- from the Murata design kit. The design kit can be down-
ing (s-parameter) and noise parameter files supplied by loaded from the Murata web site. The design kit typically
Avago Technologies. To reduce the circuit complexity and contains measured s-parameters to 20GHz for the capaci-
simulation time, passive chip components SRLC and INDQ tors and above 5GHz for the inductors. R1 was used for
were used. The inductor model used typical QUL values at low frequency termination. Typically 50 would suffice but
the nearest frequency (800 MHz) published by the vendor the optimization tool derived 100 ohm value to minimize
[5]. Inductor parasitic capacitances were calculated from impact on noise figure. R3 was used for increased stability
their published typical SRF values but with an extra 0.1 pF margin. A lower value of R3 can be used if higher margin
added to account for pad-to-pad capacitance. The parasit- is required. R3 had little impact on output linearity. R3 had
ic inductance and resistance, in the SRLC capacitor model to be placed on the demo board at an angle since the ge-
was also obtained from the vendor [6]. The 2-pin header neric board does not have a dedicated pad for a shunt re-
and its associated pads were excluded because they were sistor next to L2. A 4.7uF capacitor was added to the eval-
found to have little impact on the simulated results. The uation board in order to adequately decouple the bench
ADS2009U1 optimization tool was used to help calculate power supplies used in testing.
the values of the matching networks. The optimization
Tee7 TL24 TL25 Tee12
Subst="MSub1" Subst="MSub1" Subst="MSub1" Subst="MSub1"
W1 W mm
W1=W W=W mm W=W mm W1 W mm
W1=W
W2=W mm L=4.0 mm L=4.0 mm W2=W mm
W3=W mm W3=W mm
V1
SRLC5 R3
D=10.0 mil SRC1
R=0.1 Ohm R=6000 Ohm Tee9
H=10.0 mil TL22 Vdc=5.0 V
T=0.7 mil L=0.8 nH Subst="MSub1" Subst="MSub1"
Rho=1 0
Rho=1.0 C=10
C 10 nF W=W mm W1=W mm
W=20.0 mil L=1.0 mm {-t} V7 W2=W mm
SRLC13 W3=W mm
Var VAR1
D=10.0 mil Eqn
Ref
Figure 4. Agilent Technologies ADS2006A simulation circuit of a MGA-633P8 based 900 MHz LNA
4
Results and discussion
The general measurement conditions were f0 = 450 MHz, Although this LNA targets 400-500MHz Cellular applica-
Vdd = 5V, and Idd = 55 mA. Key specifications for design tions, the good noise figure and return loss results from
engineers are low noise in conjunction with good return UHF to S-band show that the MGA-633P8 will be useful
loss (RL). This is because diplexers and filters are detuned in wideband/multi-band applications such as Cable/Sat-
by reflective terminations. ellite TV distribution infrastructure, scanners, and military
and multi-service radios.
At 450 MHz the demonstration board LNA achieved a
noise figure, F, of 0.46 dB, a gain, G, of 22.45 dB and both The measured Rollett stability factor, k, is greater than 1
IRL and ORL better than –15 dB. The wide bandwidth of when evaluated from HF to approximately 20 times the
the input and output match is favorable from the system operating frequency. The Geometrically derived stabil-
standpoint as it prevents detuning of the input/output fil- ity factor, Mu, for both source and load is greater than 1
ters out-of-band frequency response. For both the G and also. This means the LNA will be unconditionally stable
RL parameters there is good agreement between simu- with any termination having a positive real part. For ad-
lated and measured results. ditional margin on the stability factors resistor R3 can be
decreased at the expense of lower IP3.
25 2.5
11.00
22 2 StabFact_K
Sim Gain Mu_load
9.00
Noise Figure, dB
Mu_source
19 Meas Gain 1.5
Gain, dB
Sim NF
Meas NF Stability 7.00
16 1
5.00
13 0.5
3.00
10 0
100 300 500 700 900
MHz 1.00
0 2000 4000 6000 8000 10000
Figure 5. Simulated and measured Gain and Noise Figure vs. frequency MHz
Figure 7. Wideband Rollett and Geometrically Derived source and load
The simulated noise figure is 0.1 dB higher than the mea- stability factors vs. frequency
sured result at f0. This small difference is not unusual and
has been ascribed to a system repeatability issues. An ATN
With a 55 mA quiescent bias, the gain dropped 1 dB from
source-pull system was used for device noise character-
nominal at an RF input drive of 3.0 dBm. This corresponds
ization whereas an Agilent 8970S noise figure meter was
to an output 1-dB compression point (P1dB) of 22.3 dBm.
used for the final measurement. There are also different
sample and day-to-day variations that contribute to re- 25 70
peatability [ 5]. 68
Output Power, dBm and Gain, dB
20
Fundamental 66
0 15 Gain 64
Sim IRL Id
Sim ORL 10 62
Id, mA
-5 Meas IRL 60
Meas ORL 5 58
Return Loss, dB
-10 56
0
54
-15 -5
52
-10 50
-20 -30 -25 -20 -15 -10 -5 0 5
Input Power, dBm
5
The point where the fundamental signal power (Pfund) Appendix 1 – Improved input matching version
and the third order intermodulation distortion theoreti-
cally intersect is called the third order intercept point, For an application that requires better input match than
OIP3, is a measure of amplifier linearity. Equation 6 shows -15 dB, the component values under the “improved IRL”
how OIP3 is calculated: column resulted in IRL = -22 dB (fig. 11). An additional
inductor L3 is required. The noise figure was slightly de-
ΔIM graded (0.62 dB) in this LNA version with a 3-element in-
OIP3 = Pfund + Eq. 6 put match.
2
ΔIM is difference between the fundamental and the inter- Table 3. MGA-633P8 demonstration board part list with im-
modulation product power in dB. Two input tones at 445 proved IRL
MHz and 455 MHz were used for this measurement; how- Position Value Description
ever, other frequency spacing is not expected to change C1 100 pF Murata GRM15
the results significantly. In the linear operating region (Pi <
C2, C3 22 pF Murata GRM15
-10 dBm), OIP3 is above 37 dBm at standard biasing.
L1, L2 68 nH Toko LL1005
40 39
L3 100nH Toko LL1005
20 37 C4 47 pF Murata GRM15
0 35 C5, C6 10 nF Murata GRM
R1 100 Ω
IP3, dBm
-20 33
dBm
R2 4.7 Ω
-40 31
Rbias 6200 Ω
-60 29 SMA Connectors 142-0701-851 Johnson
IMD
Pout Note: All components are 0402 size unless otherwise noted
-80 27
IP3
Vdd
-100 25
-30 -25 -20 -15 -10 -5
Input Power, dBm Rbias
Figure 9. Fundamental, 3rd order intermodulation product, and, OIP3 vs. Pi C5 C6
R1 R2
If either higher or lower linearity is required, the OIP3 can C4
C3
be varied as much as 10 dB by operating at different Idd
values in the 25 mA to 75 mA range. Gain and P1dB are L2
minimally affected (ΔG and ΔP1dB ≤ ~0.5 dB) when Idd is L1
P1 P8
varied. Using the same bias and matching networks the P2 P7
Input Output
Vdd was dropped to 3.3V. The Id decreased to 30mA. The P3 P6
IP3 decreased to 32dBm. Gain, noise figure and return loss L3 C1 C2
P4 P5
remained similar to results presented. Centre tab
MGA-633P8
Summary Figure 10. Improved input RL LNA circuit schematic
An MGA-633P8 LNA for 400-500MHz applications has
30.00 5.00
been designed and prototyped. Measured and simulated dB(S(1,1)) dB(S(1,2))
results show that the Avago MGA-633P8 has excellent NF, 20.00 dB(S(2,2)) dB(S(2,1))
Gain, Isolation, IPRL and OPRL, dB
StabFact1
G, and ORL performance with minimal external compo- 10.00 4.00
0.00
Stability Factor, K
nent count.
-10.00 3.00
Table 2. Typical Performance for MGA-633P8 application
-20.00
demoboard. (Results at 450 MHz)
-30.00 2.00
Parameter Typ. Unit -40.00
Gain 22.5 dB -50.00 1.00
OIP3 37.8 dBm -60.00
For product information and a complete list of distributors, please go to our web site: www.avagotech.com
Avago, Avago Technologies, and the A logo are trademarks of Avago Technologies in the United States and other countries.
Data subject to change. Copyright © 2005-2012 Avago Technologies. All rights reserved.
AV02-3792EN - September 11, 2012