Special Report
QORVO GAN15 TECHNOLOGY
The Qorvo GaN 15 (QGaN15) process is shorter gate length
technology capable of supporting applications through 40
GHz, as compared to QGaN25 (for applications to 18 GHz)
and QGaN50 (for applications to 10 GHz).
2
The process uses
AlGaN/GaN epitaxial layers grown on high thermal conductivi
-
ty silicon-carbide (SiC) substrate. Performance parameters of the
QGaN manufacturing technology include maximum currents
of 1.15 A/mm and maximum frequency of oscillation (f
max
) as
high as 160 GHz,
3
with nominal operating voltage of 28 V.
Models for six of the many transistor devices available
from Qorvo are now included in the Modelithics Qorvo
GaN Library.
4
These are the TGF2933, TGF2934, TGF2935,
TGF2936, TGF2941, and TGF2942 (Fig. 1). Two model versions
were developed for each of the six discrete die transistors, a
small-signal/noise model and a nonlinear model. The small-
signal/noise model provides for highest linear and noise simula
-
tion accuracy in comparison with the small-signal simulations
of the nonlinear models. However, for these die, the nonlinear
models also provide good predictions of small-signal S-param
-
eters and noise parameters if designers happen to prefer to use
the same model for both linear and nonlinear simulations.
SMALL-SIGNAL/NOISE GAN HEMT MODEL VERSIONS
The small-signal transistor models are optimized for broad-
band S-parameters and noise at specific operating bias condi-
tions. These models offer slightly enhanced model-to-measured
accuracy for the small signal simulation, which is of primary
importance in low-noise amplifier (LNA)/receiver applications.
In such applications, the use of a GaN device may be preferred
due to tolerance to large-signal inputs (therefore eliminating
the need for limiter circuitry) and/or linearity requirements.
Precision broadband measurements from two test bench
configurations were used to validate the small-signal model
version. A multi-bias S-parameter test bench was used that
consisted of a vector network analyzer (VNA) along with bias
tees, bias voltage source, RF wafer probe station, and a PC.
Measurement calibration software was used that enabled effi
-
cient bias sweeps along with multi-line TRL calibration with
custom microstrip standards.
The noise parameter testing bench consisted of a Keysight
PNA-X VNA, noise source, biasing equipment, Maury tuner,
noise receiver module, switch box, and power distribution hub
(Fig. 2). This system, which is capable of noise parameter char
-
acterization to 50 GHz, was used for noise modeling of the
QGaN15 transistors to 36 GHz. Figure 3 shows the model-to-
measurement accuracy of the TGF2942 small-signal model.
GPIB
Switch Driver
Power Supply
VNA
Noise Source
NSM
Bias
Tune
MT984AL01
DUT
DUT
MT7553B03
VNA
Bias
2. Shown is a block diagram of the Maury Microwave-/Keysight
Technologies-enabled noise parameter setup.
S-Parameters Model vs. Measured: Bias 2, VDS = 12V, IDS = 40mA, 25C
3A
S11
S21 and Max Gain (dB)
40
30
20
10
0
-10
Max Gain (dB)
S21 (dB)
0 10 20 30 40 50 60
Frequency (GHz)
S12 (dB)
0
-10
-20
-30
-40
-50
0 5 10 15 20 25 30 35 40
Frequency (GHz)
S22
Noise Model vs. Measured: Bias 2, VDS = 12V, IDS = 40mA, 25C
NFmin (dB)
4
3
2
1
0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Frequency (GHz) Frequency (GHz)
Rn (ohms)
140
120
100
80
60
40
20
0
Gamma Opt
3b
3. This figure illustrates the 0.2-to-40 GHz S-parameter (a) and 2-to-36
GHz noise model (b) performance for the TGF2942 small-signal model
(HMT-QOR-TGF2942-SS-001). The parameters are as follows: Vds = 12
V, Ids = 40 mA, 25°C, 50-Ω Smith charts. The solid red lines represent
model data, while the blue circles represent measured data.
32
NOVEMBER 2017 MICROWAVES & RF