|
Models
Help Designers Specify New High-Power Devices
By Beatrice Branger, Application Engineer; Jean-Marc
Coupat and Sandra Gendraud,
Design Engineers; and Nelsy Monsauret, CAD and Modeling
Engineer; Freescale Semiconductor
The multicarrier power amplifiers in wireless
base station transceivers must accommodate signals that
employ highly-complex modulation schemes. To do this effectively,
devices such as the MRF7S18170H/HS, MRF7S19170H/HS, and
the MRF7S21170H/HS from Freescale Semiconductor must deliver
high CW power levels (about 170 W) in bands between 1.8
and 2.2 GHz while maintaining high efficiency and linearity.
Dissipating heat at such high RF power levels is also a
major consideration. To help designers integrate these devices
in their RF subsystems, Freescale offers validated models
that can be downloaded from the company’s Web site.
These models can reduce design time and make it easier to
select the best operating conditions for a specific application
and accurately predict device behavior. The models have
demonstrated their ability to create first-pass design success
at the circuit board level.

The 170-W Devices and Models
The RF power transistors were developed in Freescale’s
seventh-generation (HV7) LDMOS technology, which when compared
to its sixth-generation predecessor delivers about 1.5 dB
higher gain and 2% higher efficiency. The devices (Figure
1) operate from 1805 to 1880 MHz, 1930 to 1990
MHz, and 2110 to 2170 MHz respectively, with output power
greater than 170 W CW. The devices operate at +28 VDC, have
internal input and output matching, and are housed in low-thermal-resistance,
metal-ceramic packages.
Freescale has developed non-linear electro-thermal models
of the three devices, each one validated against small-signal
and large-signal measurements. They include the effects
of the active die, package, and linear matching networks.
Although the simulations described in this article were
completed using Agilent Technologies’ Advanced Design
System (ADS) software, models can be used with APLAC Analog
Design Tool and AWR Microwave Office. They are available
at www.freescale.com/rf/models.

Model Verification
Under small-signal conditions, modeled and measured devices
can be compared using the same quiescent current and drain
voltage. Performance of the MRF7S19170H has been evaluated
from 1.8 to 2.4 GHz with the TRL calibration method used
to calculate the S-parameters in the package reference planes.(1)
The small-signal parameters of the MRF7S19170H model agreed
well with those of actual measurements over a large bandwidth.
Under the quiescent current specified in the product datasheets,
the modeled and measured products were loaded at the input
with optimum impedances for best input return loss. Pulsed
load-pull measurements were then performed at the 1-dB gain
compression point. The impedances for best power-added efficiency
and best output power at 2.11 GHz are shown in Figure
2 at 1-dB gain compression. In Figure 2a,
the curves are power contour and Zopt for optimum output
power, and in Figure 2b they are efficiency
contour and Zopt for maximum power-added efficiency.

Good correlation was obtained for the optimum impedance
locations at 2.11 GHz. Device output was verified at 1-dB
gain compression and the results are shown in Table
1. Simulated and measured RF performance is compared
in Figures 3a and 3b at
1.88 GHz versus back-off from the P1dB output (to obtain
optimum efficiency) and shows good agreement between the
two. Similar results have been obtained by loading the model
and the device with the optimum output power impedances.
Performance Evaluation
Freescale has implemented a new peak-to-average ratio (PAR)
test based on customer saturated power requirements. From
the average output power (Pout_avg) under W-CDMA test conditions
and the PAR at the output of the device, the saturated output
power (Psat) can be calculated:
Psat (dBm) = PAR (dB)+Pout_avg (dBm)
This formula can characterize a device operating in compression
to reflect its behavior when modulated signal peaks appear
during data transmission. The optimization is performed
with the goal of obtaining manageable impedances at the
gate and drain lead reference plane, so the device is easy
to integrate into a customer application.
Load-pull measurements were made with a single-tone CW signal
at the highest frequency in the band (2170 MHz) and were
performed on one die that was internally matched to select
the transistor periphery and to more effectively size the
device. Matching was then optimized for each frequency band.
To evaluate the behavior of the devices in conditions approximating
those of their intended applications, the test-signal employed
in production is a single-carrier W-CDMA signal that is
compliant with the 3GPP specification. This complex signal
has a source PAR equal to 7.5 dB at 0.01% probability on
a Complementary Cumulative Distribution Function (CCDF)
curve.
The main characteristics of each device (such as gain and
input return loss), are measured with RF power meters, and
Adjacent Channel Power (ACP) is measured with a spectrum
analyzer that also measures the PAR at the device output.
A peak-power meter can also be used to measure PAR. The
test bench is calibrated to determine input and output power,
input return loss, gain, efficiency, PAR, and ACP in the
transistor reference planes.

The production test fixture for the MRF7S19170H is tuned
with chip capacitors across the RF transmission line in
order to present the best impedances to the transistor and
have 50 ohms at the connector planes. The test fixtures
for each device have been tuned to achieve the best combination
of PAR, efficiency, gain, ACP, input return loss, and flatness
over the frequency range.
Conclusion
In order to ensure that Freescale’s new HV7 high-power
RF transistors can perform in actual service conditions,
the company subjected them not only to the conditions specified
for W-CDMA by 3GPP, but to more demanding tests as well.
The models created to predict this performance have been
verified many times by several organizations and have shown
themselves to be in very good agreement with actual device
measurements. These models are available to all designers
wishing to employ the MRF7S18170H/HS, MRF7S19170H/HS, and
the MRF7S21170H/HS.
Reference
1. “Impedance measurement for high-power RF transistor
using TRL method”, Jean-Jacques Bouny, Microwave Engineering
Europe, April 1999, p. 118.
Freescale
Semiconductor
www.freescale.com
TXTLINX.COM
72
Email this article
to a friend!
|