Transcript Slide 1

University of Manchester:
Progress on LNA Programme
B. Boudjelida, A. Sobih, A. Bouloukou, S. Arshad,
S. Boulay, J. Sly and M. Missous
School of Electrical and Electronic Engineering
University of Manchester
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
1
OUTLINE
•
•
•
•
Introduction
LNA Elements
• Modelling (pHEMTs and passives)
• Noise measurements
pHEMTs
LNA Results
• MMIC using InP (RF + noise)
• MIC using off-the shelves components
(AVAGO + NEC transistors)
• Noise predictions for next LNA
Capacitors
~ 5 cm
~ 1 mm
Conclusions
Resistors
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
Inductors
B. BOUDJELIDA
2
Introduction
Workflow at University of Manchester
Noise measurements
Material assessment
Material growth
DC & RF measurements
Process set-up and fabrication
Parameter extraction &
device modelling
LNA building blocks library
Process
set-up
LNA circuit design
LNA layout design
LNA Fabrication!
LNA Measurement
LNA testing
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
3
LNA Elements
350
300
250
200
150
100
50
0
0.0
S-Parameters (dB)
Id (mA/mm)
Modelling: passives and pHEMTS
Measured
Model
0.5
1.0
1.5
2.0
30
20
10
0
-10
-20
-30
-40
-50
S12
0
VDS (V)
4 x 200 µm (XMBE109-Run1)
Vp = -1.3 eV
Gm = 300 mS/mm
Ft ~ 30 GHz
Fmax ~ 35 GHz
Measured
Linear Model
Non-Linear Model
S21
5
10
15
20
25
Frequency (GHz)
Good agreement between linear, non-linear
and measured data.
For all passives, good “scalable” models successfully obtained as a
function of physical parameters
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
4
LNA Elements
InGaAs/InAlAs pHEMTs Noise Measurements
 VDS=1V : NF50 ~ 1dB (lower for
higher current)
 Lowest NF for lower VDS : WHY?
XMBE109 – 4x200 μm device: Noise figure in a 50Ω system at
different bias points (Freq=1GHz).
 Gate leakage due to impact ionization!
For better noise, the devices MUST be biased at low VDS  good for power
dissipation!
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
5
LNA Elements
InGaAs/InAlAs pHEMTs Noise Measurements
Independent Lab: MC2 (spin-off IEMN Lille)
VDS=1V; 10%IDSS
1.40
1.40
1.00
1.00
• Extraction of the noise parameters
relies on the equivalent circuit.
NFMIN (dB)
NFmin (dB)
1.20
1.20
0.80
0.80
0.60
0.60
•NFmin ~ 0.5 dB @ 1GHz
0.40
0.40
0.20
0.20
0.00
0.00
3.00E+08
0.3
0.8
8.00E+08
1.3
1.30E+09
1.8
1.80E+09
2.3
2.30E+09
2.8
2.80E+09
Freq(Hz)
Frequency (GHz)
XMBE109 – 4x200 μm device: Minimum noise figure extracted
from the “F50” method.
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
6
LNA Elements
InGaAs/InAlAs pHEMTs Noise Measurements
Independent Lab: MC2 (spin-off IEMN Lille)
VDS=1V; 10%IDSS
• Measurement independent of the
equivalent circuit!
0.16
0.16
• Expensive…
tuners.
0.12
0.12
0.10
0.10
NFMIN (dB)
NFmin (dB)
0.14
0.14
requires
accurate
• NFmin ~ 0.05 dB @ 1GHz !!
0.08
0.08
0.06
0.06
0.04
0.04
0.02
0.02
0.00
0.00
3.00E+08
0.3
0.8
8.00E+08
1.3
1.30E+09
1.8
1.80E+09
2.3
2.30E+09
2.8
2.80E+09
Freq(Hz)
Frequency (GHz)
XMBE109 – 4x200 μm device: Minimum noise figure
measured using the multi-impedance method (tuner).
4th SKADS Workshop, Lisbon, 2-3 October 2008
• This method is believed to give
more accurate results BUT the “true”
NFmin is likely to lie between the 2
measurement methods.
 NFmin ~ 0.2 dB @ 1GHz
UMan LNA Programme
B. BOUDJELIDA
7
LNA Results
InP MMIC design, fabrication and measurement
Transistor biased at 20% IDSS
(VD = 1V ; ID~40 mA)
GSG - 100μm pitch probes
Fabricated
LNA
LNA layout
LNA circuit
Comments:
No input inductor, use of large resistor, parameters optimized for best performance
Ld series resistance + Rb are used for biasing the drain
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
8
LNA Results
Simulated (ECM)
Measured
-20
S(1,2) (dB)
S(1,1) (dB)
0
-40
-60
0.2
NFmeas1
NFsim
NFmeas
NFmin
3
0.6
1.0
1.4
2.2
2.6
3.0
Frequency (GHz)
Simulated (ECM)
m2
Measured
freq=1.425GHz
NFmeas=1.289
2
m2
m1
1
0
0.5
1.8
S(2,2) (dB)
20
m1
freq=1.425GHz
NFsim=0.845
1.0
1.5
2.0
2.5
3.0
Stability
S(2,1) (dB)
InP MMIC RF and Noise results
0
-4
-6
-8
-10
0.5
2.2
2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.5
Frequency (GHz)
VD
VG
Simulated (ECM)
Measured
-2
1.0
1.5
2.0
2.5
3.0
Frequency (GHz)
Simulated (ECM)
Measured
Load
Source
1.0
1.5
2.0
2.5
3.0
Frequency (GHz)
IGGsim
IDDsim
IDDmeas
IGGmeas
7.2E-1V
-4.1E-10A
-3.1E-2A
2.3E-6
Discrepancies
with-1.0E0V
noise highly
likely to be
due to NiCr 3.0E-2
resistors process
Could also be due to measurement issues (no decoupling probes for DC feed)
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
9
LNA Results
MIC design, fabrication and measurement
Goals:
Demonstrate the validity of the model predictions
“Easy-to-assemble” using commercial off the shelves components
Could be used for demonstrators such as 2PAD
NEC transistors, Double-stage circuit, optimised for 0.4-2 GHz operation
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
10
LNA Results
MIC design, fabrication and measurement
0
S(2,1)
0
S(x,x) dB
S(x,y) dB
50
-50
S(1,2)
-100
0.2
0.7
1.2
1.7
Frequency (GHz)
2.2 2.5
-5
S(1,1)
-10
-15
-20
0.2
S(2,2)
0.7
1.2
1.7
2.2 2.5
Frequency (GHz)
8 different LNAs designed using NEC and
Avago transistors
Single and double-stage circuits being
measured now!
Very good noise predictions!
NF < 0.6 dB !
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
11
LNA Results
InP MMIC predictions
LNA circuit
Comments:
Input bias and impedance match off-chip
L series resistances used for drain biasing
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
12
LNA Results
InP MMIC predictions
0.7
m1
m1
0.6
0
freq=1.400GHz
m1
NF=0.344
0
0.5
freq= 1.400GHz
m1
NF
dB(S(2,1))=25.959
0.4
20
m2
0.3
40
0.2
60
m2
1.400GHz
80
0.1
NFmin freq=
NFmin=0.279
00
0.0
0.2 0.4 0.6
1.0 0.8
1.2 1.0
1.4 1.2
1.61.4
1.81.62.01.8 2.0
0.2 0.8
0.4 0.6
0
dB(S(2,2))
dB(S(1,1))
Noise Figure (dB)
0
-5
m2
freq= 1.400GHz
dB(S(1,1))=-9.411
-10
m3
-15
m3
freq= 1.390GHz
dB(S(2,2))=-14.739
-20
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
freq,
GHz
Frequency
(GHz)
freq, GHz
70
NF (K)
S-Parameters (dB)
m1
40 m5
.6
60
20 freq=1.400GHz
m1
NF=0.344
S21
.5
50
0
freq=1.400GHz
m5
.4
40
dB(S(2,1))=25.959
-20
m4
m6
.3
30
-40
.2
20
-60
m4
m6
S12
freq= 1.400GHz 10
.1
freq=1.4GHz
-80
NFmin=0.279
NFK=23.9
.0
0
-100
0.2 0.4 0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
6
5
MuS
MuL
.7
4
3
2
1
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
freq,Frequency
GHz
(GHz)
freq
VD1
VG1
1.006 V
-1.080 V
freq, GHz
ID1.i
NF< 0.35 dB from 0.3 to 1.6 GHz
0.0000 Hz
m2
-18.13 m A
IG1.i
-509.4 nA
freq
4th SKADS
October 2008
freq Workshop,
VD2 Lisbon, 2-3
VG2
ID2.i
Pdiss
0Hz
UMan
IG2.i LNA Programme
110
B. BOUDJELIDA
13
Conclusions
• Super low noise InGaAs/InAlAs pHEMTs technology demonstrated
NFmin < 0.2 dB @ 1GHz using the 1 µm gate geometry
• The first full MMIC LNA successfully modelled, fabricated and tested
MMIC
MIC
• Still very good agreement between measurement and models using the
equivalent circuit models
• The measured NF in the 50Ω system is also higher than what predicted by
the simulations  due to Resistors (under investigation, 2nd MMIC run
under way)
• The first fabricated MICs yield measured NF as low as 0.6 dB (~42K)
• Noise predictions demonstrated
Next LNA expected to go below 0.35 dB (25K) at RT in a 50Ω system
between 0.3 to 1.6 GHz
4th SKADS Workshop, Lisbon, 2-3 October 2008
UMan LNA Programme
B. BOUDJELIDA
14