Text preview for : an_1288-1.pdf part of HP an 1288-1 HP Publikacje an_1288-1.pdf
Back to : an_1288-1.pdf | Home
Combining Network and Spectrum
Analyses and IBASIC to improve device
characterization and test time
Using the HP 4396B to
Application note 1288-1 analyze linear and non-linear
components - a 900 MHz AGC
amplifier example
Background
Active components require
linear and non-linear analysis
Active components (and now even
some passive components like
crystal filters) require analysis to
characterize linear parameters
(gain/loss, phase and group delay
or S-parameters) as well as
non-linear performance.
Non-linear analysis is typically
related to measuring signal
distortion generated in the device
such as harmonic or inter-
modulation distortion. Therefore,
Combining vector network analysis, for complete characterization,
spectrum analysis, and a built-in
controller in one instrument offers new both a vector network (VNA) and
capabilities for RF testing.Background spectrum analyzer (SA) are
required, for linear and non-linear
evaluation respectively.
For example, to characterize an
amplifier for cellular applications,
we are typically interested in the
following measurements. Note
that 10 out of 12 measurements
are made with either a vector
network analyzer (VNA) or
spectrum analyzer (SA).
Introduction obstacle to comprehensive Testing a silicon bipolar
characterization. If operating data MMIC 900 MHz AGC
Today's RF designs are is needed for different or changing
increasingly driven by time-to- amplifier for cellular
conditions, re-configuring the test
market. At the same time, station required a large applications
advances in digital commitment of resources and In this example, a 900 MHz
communication techniques are long time delays before the data is automatic gain control (AGC)
placing higher performance available. amplifier was characterized using
requirements on components and
Combining instrument a HP 4396B-controlled mini-ATE
subsystems. These key driving
functions for improved testing system.
forces require that comprehensive
as well as time and cost effective By combining vector-network and The amplifier's block diagram is
measurement techniques are used spectrum analyses in the shown in Figure 1. A test board
in design and manufacturing. HP 4396B, and by using the was used (Figure 2) for easy
built-in HP IBASIC programming connection using SMA 50 Ohm
In this note, test approaches using
capability, a powerful new test cables. The test system consisted
the combination of vector
tool is now available for lab or of an HP 4396B Network/
network, spectrum analysis and
manufacturing applications. As Spectrum Analyzer, two program-
IBASIC program control are
the core of a mini-ATE system, the mable power supply voltage
discussed showing ways to get
HP 4396B can control and test levels, two programmable signal
better device characterization
multiple parameters with a single generators for inter-modulation
with faster results, higher
insertion of the device-under-test distortion (IMD) measurements
accuracy and increased test
(DUT). In addition, tests are easily and switch-controller with two RF
flexibility. Additional benefits
changed or customized for special switches. See Figure 3.
include improved quality control
data and ease of product transfer operating conditions or
from design to manufacturing. one-of-a-kind test requirements.
The remainder of this note will
Characterization is important use an amplifier test example to
but difficult illustrate the principles and
effectiveness of this approach.
Accurate characterization if
fundamental to both the designer
and user of high-performance RF
components. Errors in operating
parameter measurements or
operating in untested regions put
the end product at risk. Careful
and complete characterization
pats benefits during the entire
product development cycle by
allowing better decisions in design
and optimum testing during the
manufacturing phase, a large
potential cost savings. But to fully
characterize RF components often
requires numerous test
instruments and a large
investment of time and effort.
Automating testing to gather Figure 1. 900 MHz AGC Amplifier block diagram
statistical information involves
external computers and
programming. In the past, this
amount of work has been a real
2
Fast automated test results
The measurement parameters and
results from the AGC amplifier
test are shown below. The total
time to make these measurements
using the test system described
was 9.2 seconds. These results are
in summary form for easy review
and comparison with other
devices. A printout was
formulated using IBASIC
programming as a simple addition
to the automatic test control
program. (Printing out cursor
values from the various network
Figure 2. Test board used for automated testing and special data.)
To gain more insight into the
measurement techniques used,
these specific amplifier test are
discussed in more detail:
Inter-modulation Distortion (IMD)
Gain vs. AGC control voltage
Performance change with
different power supply voltages
Figure 3. Automated test system for
component characterization.
Amplifier Measurement Results
(Total measurement time = 9.2 seconds)
Parameter Symbol Value
Output power Pout 23.9 dBm
1 dB gain compression P1dB 23.1 dBm
Power control range Pcr 69.2 dB
Small signal gain Gain 34.4 dB
3rd order intercept pt IP3 30.2 dBm
Input return loss IRL -20.3 dB
Control current Icont 2.2 mA
3
Inter-modulation Distortion
(IMD)
Inter-modulation distortion is a
critical measurement because
these distortion products can fall
in adjacent channels in the
cellular radio band. Thus it is
important to characterize them
accurately and then reduce them
in the design of the system.
Harmonic distortion is also
important, but these products are
more easily removed by low- pass
filters. See figure 4.
Figure 4. Inter-modulation distortion (IMD) products from Channel B signals f1
and f2.
Two signal sources are used for
IMD, and SA is then used to
measure the third-order, and for
this amplifier, fifth-order
products. An example IMD
measurement is shown in Figure 5
using sources separated by 1 MHz
for ease of identifying IMD
signals. In this case the distortion
products are easily distinguished
from the noise floor. However, for
many measur- ements, the IMD
values may be much lower, and
hard to distin- guished in the noise
floor of the SA. For this case, a
narrower resolution bandwidth Figure 5. Third, fifth, and seventh order IMD products. Third order shown on
(RBW) is often used, but this cursor.
increases the sweep time. In
addition, IMD signals may be List sweep technique improves frequencies in between can be
much closer together than in this SA measurement speed skipped. Figure 6 shows almost
example. When IMD signals are 7x speed improvement using list
separated only by 10 to 50 kHz, With the HP 4396B, low-level IMD sweep in measuring the test
the RBW must be reduced in order signals can be measured very system IMD floor. (The unbalance
to resolve the signals. In conven- quickly using a feature called list of distortion is due to unbalanced
tional spectrum analyzers, narrow sweep. List sweep allows the signal] generators). In the
RBWs can drastically slow down a spectrum to be broken up into up right-hand display, note the lower
measurement. The HP 4396B uses to 15 segments. Each segment can noise floor for the segments
a `stepped-FFT' technique for all have unique start and stop where the IMD products are
RBWs of 3 kHz and below. This frequencies and different RBW located, resulting from a narrow
results in a factor of 10 to 100 settings. The test engineer can RBW selection. List sweep speeds
times faster sweep time compared select a segment with narrow harmonic distortion measure-
to non-FFT assisted spectrum RBW (slower sweeps and lower ments as well, by skipping
analyzers. Throughput improve- noise floor) targeting only the frequencies between the
ments for IMD measurement are a regions containing the distortion harmonics.
major advantage of using the products. The high-level test
combination analyzer for narrow signals are measured with wide
resolution, wide dynamic range RBW for highest speed, and the
measurements.
4
Figure 6. For IMD testing, using list sweep to segment the spectrum to use different RBWs and skipping unneeded
frequencies increases throughput.
IMD as a function of signal
level
IMD is dependent on signal level.
IBASIC can be used to auto-
matically measure IMD products
over a range of power levels.
In this example (Figure 7) the test
system IMD noise floor was tested
as a function of the dual-source
level, and IBASIC displayed mode.
Note that the best system noise
floor is with signal of -10 to
-25 dBm. If made manually, this
measurement would be extremely
time consuming, considering the Figure 7. IBASIC results (lower display) and IMD list sweep spectrum (upper
9 signal levels and multiple trace) for test system IMD floor test as a function of test signal level.
readings.
Gain vs. AGC control voltage
Figure 8 shows the gain over the
900 MHz band as a function of
AGC control voltage. This
measurement is easily automated
using IBASIC and throughput is
increased due to the network
analysis speeds as fast as 350
usec/pt. Instead of manually
changing the control voltage and
measuring the gain, the DC power
supply and make gain
measurements.
Figure 8. Amplifier gain as a function of AGC voltage.
5
Figure 9 shows a simplified Getting quality control In manufacturing, a simple yet
version of the IBASIC program data powerful method to improve
used on the left. On the right, 4 quality is through control charts.
lines of code are added to also When characterizing components, By using a program to measure
vary the input power. This the repeatability of the test system the test station's accuracy and
automatically provides gain vs. must be quantified. In this system, repeatability at regular intervals,
Control voltage vs. Input power the use of RF switches may affect any problems can be seen as they
enabling the user to determine the the test results. A simple IBASIC develop. The built-in floppy disk
optimum and worst-case program can be used to can store an auto-start IBASIC
performance of the amplifier. automatically determine the program that the operator runs
Getting a `three-dimensional' repeatability and accuracy of each day to test the system.
parameter analysis gives much measurements made with and Results can be used to make
more information, and the IBASIC without the switches. 100 sweeps control charts and monitor the
programmed measurement control were measured to test the effect system performance. This simple
greatly simplifies the task. of the switches. The worst case procedure can eliminate many
data spread was 0.049 dB without component test problems. An
Performance changes with the switches and 0.050 dB with example repeatability control
different power supply the switches. chart is shown in Figure 11.
voltages
The typical performance for this
amplifier is given for two power Gain vs. Control Voltage Gain vs. Control Voltage vs. Input Power
supply voltages: Vcc1 = 4.5V and