|
First
Pass Accuracy with Momentum GX for WiMAX Design
By William Clausen and Mounir Adada,
Agilent Technologies
The IEEE 802.16e standard, often referred
to as mobile WiMAXTM, specifies air interfaces for broadband
wireless access (BWA) systems. It uses roaming and handoff
to enable laptop and mobile phones to operate. The technology
is expected to energize the BWA industry by opening up new
opportunities to deploy systems in applications that were
previously cost-prohibitive.
Multiple-Input Multiple-Output (MIMO) technology - a capacity
enhancing, multi-antenna technology which is a fundamental
component of mobile WiMAX – is today addressing some
of the issues associated with mobility by extending high
throughput and improving network capacity. Despite these
benefits, cost and power consumption will continue to be
a challenge for designers, especially as they attempt to
achieve optimal system performance. Luckily, Agilent Genesys
users now have access to an industry-proven, 3D-Planar electromagnetic
(EM) simulation technology which is specifically designed
to address such challenges. Referred to as Momentum GX,
this solution helps designers optimize and verify the physical
performance of high-frequency WiMAX circuits and interconnects
within the Genesys design flow.
WiMAX: The Basics
To better understand the role of Momentum GX in the design
and verification of WiMAX circuits, it is first crucial
to have a clear understanding of WiMAX, or as it is also
known, the Worldwide Interoperability for Microwave Access
standard. As opposed to a Local Area Network (LAN) or a
Wide Area Network (WAN), WiMAX is a Metro Area Network (MAN)
based on the IEEE 802.16 specification and supports both
licensed and unlicensed bands. It provides interoperable
broadband wireless connectivity to both fixed and portable
users, at a distance of up to 30 miles or 50 kilometers
of service without the need of line-of-sight. Also, it can
achieve data rates as high as 75 megabits. WiMAX has sufficient
bandwidth to service hundreds of businesses and homes with
a single base station at T1 rates.
What makes WiMAX different is its modulation method and
adaptability (see Figure 1). It uses a
technique called Orthogonal Frequency Division Multiplexing
(OFDM) in which data is transmitted on multiple subcarriers
for each symbol. The subcarriers are spaced so that interference
with each other is minimized, thereby creating an orthogonal
set of subcarriers. Transmitting on multiple subcarriers
also enables signals to be transmitted at lower symbol rates.
In other words, the designer can realize the same throughput
(e.g., bits per second), with less susceptibility to interference
from narrow-band interference or other transmitters, as
well as multipath. Note that good multipath reception is
critical for mobile service.
WiMAX also provides an Adaptive Modulation
and Coding (AMC) technology which enables it to overcome
multipath issues by slowing the symbol rate. Essentially,
it allows the WiMAX system to adjust the signal modulation
scheme depending on the signal to noise ratio (SNR) condition
of the radio link. When the radio link is high in quality,
the highest modulation scheme is used, giving the system
more capacity. During a signal fade, the WiMAX system can
shift to a lower modulation scheme to maintain the connection
quality and link stability. Consequently, when the system
is close to a base station, it might be able to send and
receive at 256QAM. As the system moves further away though,
or the SNR decreases, it reverses to QPSK or BTSK.
WiMAX Design Challenges for RF Engineers
Designing WiMAX systems poses a number of challenges for
today’s RF engineer (see Figure 2).
Consider, for example, that WiMAX provides support for either
Time Domain Duplex (TDD) or Frequency Domain Duplex (FDD)
schemes. In some cases, it supports both duplexing schemes
– a fact which can significantly complicate the designer’s
circuitry and structure, and may even require the use of
switching networks. WiMAX also operates at a higher frequency
band, between 2 and 11 GHz. At these frequencies, the designer
faces a host of problems including increased parasitic effects,
coupling of smaller circuits between networks which degrades
signal integrity, and non-ideal behavior of lumped and distributed
components.
While closed-form design solutions exist that allow the
designer to determine element values and their influence
on the circuit, such solutions fail to provide the degree
of accuracy required at the higher frequencies in which
WiMAX systems operate. And, because many of today’s
WiMAX designs are mobile in nature and therefore powered
by batteries, there is a whole slew of other problems and
issues relating to power consumption that the designer must
contend with. Further complicating matters, designers must
also deal with radiation effects, cost constraints and manufacturability.
Effectively addressing these technical issues requires a
top-level RF system design tool such as Spectrsys, available
in the Genesys environment, which can plan, optimize and
specify component parameters prior to manufacture. It also
requires a synthesis platform to help speed the development
of all system components such as filters, mixers and oscillators.
Both linear and non-linear simulation technology is needed
to verify things like noise figure, phase noise, power,
intermods, spurious effects, and so forth. And, to ensure
the product is manufacturable and that the desired yields
can be achieved, designers must utilize yield analysis tools.
More importantly, they need an accurate EM design tool to
help optimize and verify the performance of the design prior
to going to manufacturing.
An Effective Solution
The Agilent Genesys solution, with its full suite of tools,
fully supports the requirements of today’s RF engineers
tasked with designing WiMAX systems. In addition, it also
now offers Momentum GX - an industry-proven technology for
simulating complex multi-layer 3D-planar networks anywhere
from DC to light.
Momentum GX works seamlessly with Genesys to compute and
produce the S, Y or Z parameters. It is a 3D EM simulator
that is two dimensional with Z-directed currents for multi-layer
board analysis. Based on a method of moments, it enables
RF and microwave engineers to expand the range of accuracy
for their passive models. A conformal geometry meshing system
ensures that designer gets optimal coverage from both very
small and very large components over varying distances,
such as from matching elements, and takes into account parasitic
effects.
In addition to conformal meshing, Momentum GX offers two
EM solvers: RF mode and a full-wave microwave simulation
mode - both in 32-bit and 64-bit versions. The RF mode is
a quasi-static mode which offers designers a much faster
method of simulating. It does not perform a full-wave analysis,
but in cases where circuit dimensions are less than half
of a wavelength, a full-wave analysis is not necessary.
When the designer does need to account for dispersive effects,
radiation effects, or when the circuit starts to approach
something larger than a half of a wavelength, the full-wave
microwave mode can be utilized.
Momentum GX also offers a Genesys co-simulation technology
which allows the designer to embed components in an EM simulation
and examine the results as a combination. An Adaptive Frequency
Sweep (AFS) function provides the designer broadband frequency
information with much fewer points. It also accounts for
thick metal modeling for accurate coupling between circuits
and traces.
Because of its functionality, Momentum GX is suitable for
designing passive matching structures or filters, and component
modeling. The close-form models typically used for microstrip
elements are not as accurate at higher frequencies. Momentum
GX can also be used for layout verification to help the
designer put components on pads and analyze how those pads
effect the ultimate response, as well as for signal integrity.
In the latter case, Momentum GX enables the designer to
determine how much coupling occurs between different layers
within the board. The solution can also be used for antenna
design – a task which is very important, especially
in WiMAX, and for designing and verifying Low Temperature
Co-Fired Ceramic (LTCC) material designs.
Other uses for Momentum GX include:
• New topologies, where there are no closed-form solutions
or equations to describe a
network, a tool like Momentum GX becomes essential
for modeling and analyzing the
network.
• Manufacturability. Momentum GX can be used to look
at the effect of changes in substrate
permittivity variations and their effects on the
final design.
• Reducing component count. At higher frequencies
very small capacitors and inductors
can be replaced with printed elements, but the models
that describe those elements
are few and usually not very accurate. Momentum GX
generates EM accurate models.
• Improved passive models for standard structures
such as steps, tees and open
grounds. In many multi-layered designs, the designer
winds up with an open in
the ground that either a signal line or power
line can go through.
The 3D-planar EM technology in Momentum GX differs greatly
from the typical grid-based EM solver, as represented in
Figure 3, which relies on the use of grid
dimensions as opposed to those of the component values.
Consider, for example, a 115 mil-wide line and a grid with
20 mil grid cells. In this scenario, the designer would
only be allowed to snap to, or to analyze, a situation where
the line was either 100 or 120 mils. In order to see 115
mil, the designer would have to change the grid dimensions
to 5 mils, thereby allowing the tool to snap on a grid point.
But changing the grid dimensions increases the size of the
matrix, along with the time it takes to solve the problem.
In some cases, the designer may be unable to solve it at
all, in part because the solution space might be aimed at
the multi-gigabyte memory.
Typical grid-based EM solvers also suffer from the “big/small”
problem in which there are large distances between small
elements or different-sized elements. In Figure
3, this problem can be witnessed with the spiral
inductor. Clearly, too coarse of a mesh has been used and
it is not meshed accurately enough. As a result, the simulator
might easily resolve it as a solder blob when in reality
it is actually a spiral inductor. The same issue holds true
for the meander line in Figure 3. Essentially
then, if the designer creates a grid to achieve accuracy
with the smallest element, he may wind up with a very big
solution space. When solving matrices, a general rule of
thumb is that as the solution space increases, so to does
the number of required calculations (e.g., by a factor of
two). Put simply, the “big/small” problem can
be a very daunting challenge for the designer.
Thanks to its unique conformal meshing technique, Momentum
GX is much more accurate for all geometries – even
small ones – essentially overcoming the “big
small” problem. Figure 4 shows the
same network as pictured in Figure 3, only
this time it is meshed in Momentum GX. Note that there is
no accuracy lost in the smallest element. On the contrary,
even with the small mesh that occurs it is clear the spiral
inductor has been uniquely meshed and provides enough information
for accurate results. The same is true for the meander line
and the microstrip components. They are meshed a little
bit larger, allowing the designer to get accurate simulations
for the various geometries. Far less memory is required
with this approach and it also provides a faster solution.
If the physical length of this circuit is small compared
to the wavelength of the operating frequency the designer
can even further increase the speed of the process by using
Momentum GX’s RF mode.
Using Momentum GX
Momentum GX can prove especially useful when it comes to
designing WiMAX circuits. To better understand that usage,
consider the design example of a simple three-pole, Butterworth
microwave filter operating in the 3.4 to 3.6 GHz range.
The Genesys synthesis tool is used for fast and easy filter
design.
To begin, the designer must first step through a process
of tabs and options to select the filter type, shape and
subtype that are going to be used in the design (see Figure
5). The type of filter selected depends on several
factors, including: design size, what the designer can live
with in terms of size and ease of manufacturing. Some filters
require tighter spacing tolerances than others. The filter’s
“cost to produce” is also an important factor
as is its recurring frequency response. Note that all distributed
networks exhibit a recurring frequency response at multiples
of either a quarter-wave or half of the operating frequency.
A cone filter, for example, offers the most compact size
relative to the others and allows the designer to choose
where the recurring band pass occurs. Unfortunately though,
it also costs more because it requires the addition of more
components such as capacitors.
Once the filter is selected - in this case, an edge-coupled
filter – the designer must set the reference impedance
which could be 50 ohms or some other impedance value. If
the filter is embedded in another design, then the designer
may want to select either a higher or a lower reference
impedance. Next, the attenuation at the band edges, the
order of the filter and the frequencies are selected, giving
the designer control over the resonator, Zo, and the ability
to optimize the unloaded queue of the resonator. As an option,
the designer can also specify tapped input and the length
of the input line. Following filter set up, a layout is
selected and generated. The manufacturing process is also
selected (e.g., coplanar, strip line, inverted strip line,
microstrips, lab line, etc.) and the design readied for
analysis via Momentum GX.\
Figure 6 depicts the results of a Momentum
GX simulation, including the Momentum GX generated mesh
for the selected filter. Note that the results show a shift
in the filter response. The red trace represents the response
of the linear network that was synthesized and includes
losses. The green trace represents the Momentum GX simulation
or filter and shows that a shift higher in frequency has
taken place and that the bandwidth is slightly less. In
other words, the filter is slightly undercoupled and its
lengths must therefore be slightly lengthened. This is problem
which can be easily fixed, but would likely not have been
found were it not for Momentum GX’s ability to accurately
indicate filter performance.
In some filter cases, especially those with bandwidth
of better than 10 percent (e.g., 15 percent or so), the
input and output coupling on, for instance, an edge-coupled
filter can become very narrow. It may be too difficult therefore,
to hold this level in manufacturing tolerances. Genesys
features a tapped model which offers a work around to this
issue, but the closed-form tapped models are limited in
terms of accuracy and frequency (see Figure 7).
Momentum GX assists with this problem by re-tuning or repositioning
the taps in order to bring the filter back into compliance.
It then further helps by improving the model that is used
to design the filter.
At this point, the design is almost ready to be released
to manufacturing - pending verification for manufacturability.
This is a very important step – as it would be highly
embarrassing to get a board back only to find out that it
is not what was expected. To minimize this risk, the permittivity
for the board must be swept. Figure 8A
shows the Momentum GX swept simulation results of Er from
4.1 to 4.9 GHz. Note that the center frequency has shifted
from 3.3 to 3.6 GHz. With this information, the designer
can specify the tolerances on the manufacturing process
or in the permittivity of the substrate. Figure 8B shows
a Monte Carlo linear analysis based on the spacing dimensions
and the width of the lines for the filter in question.
There are a number of key lessons which can be gleaned from
this example. These lessons include:
• Closed-form distributed models are limited. The
radiation effects and loss modeling
is limited in terms of frequency and accuracy.
• Adjacent conductor coupling does not account for
thick metal. In other words, there are
really no zero-thickness lines.
• Via models are usually limited to lumped equivalents,
and non-adjacent coupling
does not exist. The models for coupling between lines
are only for single-pair coupled
lines. They don't take into account the third or
fourth resonator over - which does and can
affect the shape.
• Swept analysis is necessary to ensure manufacturability.
Conclusion
WiMAX – and especially mobile WiMAX – is today
an emerging telecommunications market. Its proliferation
will be determined in part by the designer’s ability
to overcome the various design challenges involved in meeting
current market demands. Momentum GX, an addition to the
Genesys suite of products, now offers designers the functionality
and flexibility they need to significantly expand the range
and accuracy of their passive circuits and circuit models.
With its unique meshing technology and ability to analyze
arbitrary shapes, on multiple layers as well as to consider
real-world design geometries when simulating coupling and
parasitic effects, Momentum GX is an indispensable and necessary
tool for ensuring first pass design success.
“WiMAX,” the a trademark of the WiMAX Forum.
AGILENT TECHNOLOGIES
www.agilent.com
TXTLINX.COM 104
Email
this article to a friend!
|
