Title: University of Manchester: Progress on LNA Programme
1University of ManchesterProgress 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
2OUTLINE
- Introduction
- LNA Elements
- Modelling (pHEMTs and passives)
- Noise measurements
- LNA Results
- MMIC using InP (RF noise)
- MIC using off-the shelves components (AVAGO
NEC transistors) - Noise predictions for next LNA
- Conclusions
3Workflow at University of Manchester
Introduction
LNA building blocks library
Process set-up
LNA circuit design
LNA layout design
LNA Fabrication!
LNA Measurement
LNA testing
4LNA Elements
Modelling passives and pHEMTS
4 x 200 µm (XMBE109-Run1)
Vp -1.3 eV Gm 300 mS/mm Ft 30 GHz Fmax 35
GHz
5LNA Elements
InGaAs/InAlAs pHEMTs Noise Measurements
? VDS1V NF50 1dB (lower for
higher current)
? Lowest NF for lower VDS WHY?
? Gate leakage due to impact ionization!
For better noise, the devices MUST be biased at
low VDS ? good for power dissipation!
6LNA Elements
InGaAs/InAlAs pHEMTs Noise Measurements Independen
t Lab MC2 (spin-off IEMN Lille)
VDS1V 10IDSS
- Extraction of the noise parameters relies on the
equivalent circuit. - NFmin 0.5 dB _at_ 1GHz
XMBE109 4x200 µm device Minimum noise figure
extracted from the F50 method.
7LNA Elements
InGaAs/InAlAs pHEMTs Noise Measurements Independen
t Lab MC2 (spin-off IEMN Lille)
VDS1V 10IDSS
- Measurement independent of the equivalent
circuit! - Expensive requires accurate tuners.
- NFmin 0.05 dB _at_ 1GHz !!
- 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 _at_ 1GHz
XMBE109 4x200 µm device Minimum noise figure
measured using the multi-impedance method (tuner).
8LNA Results
InP MMIC design, fabrication and measurement
Transistor biased at 20 IDSS (VD 1V ID40 mA)
9LNA Results
InP MMIC RF and Noise results
Discrepancies with noise highly likely to be due
to NiCr resistors process
Could also be due to measurement issues (no
decoupling probes for DC feed)
10LNA 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
11LNA Results
MIC design, fabrication and measurement
8 different LNAs designed using NEC and Avago
transistors
Single and double-stage circuits being measured
now!
- Very good noise predictions!
- NF lt 0.6 dB !
12LNA Results
InP MMIC predictions
LNA circuit
Comments
Input bias and impedance match off-chip
L series resistances used for drain biasing
13LNA Results
InP MMIC predictions
NFlt 0.35 dB from 0.3 to 1.6 GHz
14Conclusions
- Super low noise InGaAs/InAlAs pHEMTs technology
demonstrated - NFmin lt 0.2 dB _at_ 1GHz using the 1 µm gate
geometry
- The first full MMIC LNA successfully modelled,
fabricated and tested
MMIC
- Still very good agreement between measurement
and models using the equivalent circuit models
- The measured NF in the 50O 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)
MIC
- Noise predictions demonstrated
Next LNA expected to go below 0.35 dB (25K) at RT
in a 50O system between 0.3 to 1.6 GHz