Text preview for : 5989-2422EN.pdf part of HP 5989-2422EN HP Publikacje 5989-2422EN.pdf
Back to : 5989-2422EN.pdf | Home
Agilent
Investigating Microvia Technology
for 10 Gbps and Higher
Telecommunications Systems
White Paper
Contents
Introduction ...........................................................................................2
Telecom Physical Layer Overview ......................................................3
Signal Integrity and Differential Spacing .........................................9
Four Port Microvia Measurements...................................................14
Microvia Construction........................................................................17
Modeling and Simulation Case Study..............................................22
Summary and Conclusions ...............................................................27
Acknowledgements .............................................................................28
Introduction
Today's telecommunication platforms depend upon high-speed serial data
transmission. Leading edge digital designers push the performance limit of
what is possible to achieve on copper. The proliferation of serial links
beyond 10 Gbps has exposed signal integrity issues not typically encountered
in the standard digital design laboratory. Optimization of signal integrity by
focusing on the physical layer structures within these high-speed channels
can produce astonishing results. The fundamental insight of how signals
propagate can be clearly understood with the proper design tools and
methodologies.
Network switches and routers have recently employed advanced backplane
technology to break the terabit barrier. This accomplishment is in part due
to sophisticated design techniques within the physical layer components. A
major portion of this design cycle is geared toward modeling, simulation,
and measurement validation. Reflections, crosstalk, impedance mismatch,
and loss can be visualized. These complex phenomenon become intuitive
through the use of design tools that allow both time and frequency domain
analysis.
Real estate constraints of high-speed digital systems necessitate the use of
microvia technology to allow more components to be placed on a single
circuit board. With more companies using microvia technology, the process
has been able to advance rapidly from controlled depth drilling to more
advanced laser ablation techniques. Printed circuit board manufactures are
tasked with developing processes for microvias that meet the aspect ratio
requirements of today's multilayer backplanes. Implementing microvias
opens the door for SMT board to backplane connectors and the overall
system performance improvements inherent in these connector systems.
2
Telecom System Physical Layer Overview
Typical 10 Gbps Telecom System
During the late 1990s and early 2000s the focus of network OEMs was the
delivery of high-performance technology-leading communication systems to
meet the demands for ever greater telecom bandwidths. Chassis and back-
plane design were key differentiators for market-leading manufacturers. The
communications industry today is evolving towards modularity in a manner
very similar to the server world transition in the early 90s.
Telecommunication systems such as that shown in Figure 1 typically achieve
high-speed data transport through a switch fabric interface that can be used
as a secondary communications channel in parallel with the base interface.
In most high-speed networking applications, the base interface will be used
to carry communications between the control-plane processors on each line
card. This physical layer copper interface creates many challenges for signal
integrity engineers designing, developing, and testing network elements.
One of the most challenging and interesting areas for high-speed design is in
backplane applications. Performance of routers and switches is fundamentally
limited by the bottleneck created by the backplane components. Therefore,
this is an area rich for technology breakthroughs and innovation.
Components
Router
Modules
Copper
Line cards
Network elements &
systems Trunk fiber
Figure 1. Typical telecommunication system
3
Backplanes are a critical link
Today's standards efforts for a 10 Gbps Ethernet backplane are under
development within the IEEE 802.3ap task force. The goal is to use an
ordinary copper backplane as shown in Figure 2 to send 10 Gbps Ethernet
signals between line cards using no optics. A standard would give systems
designers a head start, allowing them to choose among several physical-layer
chips that would be pre-wired for the standard. Exciting work is being done
with 10-Gbps serial and novel signaling schemes such as binary signaling or
PAM4 that can help achieve this high-speed data transmission. However, the
ultimate limit of serial rates will be most likely dictated by the signal integrity
of the physical layer backplane. Achieving a controlled impedance environment
throughout the complete backplane channel from chip-to-chip will demand
careful and meticulous design methodologies. The backplane connector plays
a critical part of this channel.
Figure 2. Examples of ordinary copper backplanes
4
Backplane connectors are advanced
Designing a surface mount board to backplane connector has numerous
requirements. First, the interface must withstand the mechanical conditions
faced by standard board applications and be very rugged. Secondly, the
connector must be able to transmit data at speeds exceeding 10 Gbps.
Recent designs of Surface Mount (SMT) backplane connectors have evolved
from press fit connector technology, including many of the same mechanical
features such as the 1.5 mm x 2.5 mm pin grid. The two main differences in
the connector designs revolve around the use of SMT signal leads and the
`C-shaped' pin-in-paste ground shield pin. The exploded view shown in
Figure 3 shows the details of construction of a 10 Gbps connector.
A well-designed high-speed board to backplane connector integrates the
mechanical, chemical, and electrical properties of the device seamlessly.
Orientation of differential pairs, spacing of contacts, and selection of compo-
nent materials all play key roles in the overall performance. It is a challenge
to find the proper combination of these design criterions without impacting
the signal integrity of the connector. A great deal of time and effort goes into
the design and modeling of these types of connectors before the first piece of
steel is cut.
High precision
molded
components
Differential
signal traces
Surface mount terminals
Double sided shield
Figure 3. Construction of a 10 Gbps connector
5
Via stubs create capacitive loads
It is essential to reduce the amount of via stub to successfully transmit data
at 10 Gbps. Connectors that require interfaces to Plated Thru Holes (PTHs)
are susceptible to capacitive loading that is inherent to the geometry of
these commonly used board attachments. To overcome this challenge, the
most critical lines would need to be routed closest to the bottom surface of
the PCB as shown in Figure 4 or the via barrels back drilled to reduce the
via stub. This may lead to longer design times and more layers to achieve the
desired signal integrity performance.
Many board designers implementing press-fit connectors try to reduce the
resonant behavior of a PTH by routing signals near the bottom layer of the
PCB, or back-drilling critical lines to reduce the via stub. With surface
mount connectors, there is no need for back-drilling since the connector is
mounted on the top surface of the PCB and the signals are attached with
blind or buried vias. This type of connection scheme allows the system
bottleneck to move from the connector to the PCB material.
Top connection Bottom connection
Figure 4. Top and bottom routed lines
6
Connector to board interface
With an SMT termination the reflective behavior associated with a PTH is
reduced or even eliminated since there is very little dangling stub. The
interface to surface mount devices, whether it is resistors, silicon IC's, or
connectors, must be made on the outer surfaces of the PCB. It would be
impossible to route all of the signal lines of a high density, high-speed,
differential connector on the outer surface alone, therefore an alternate
approach must be taken. Additionally, these high-speed lines will need to
interact and connect with inner routing layers to achieve all of the desired
functionality of the system. Different via structures can be used in
conjunction with the backplane connector to interface the connector to the
inner board traces. Graphical examples of both PTH and SMT are shown in
Figure 5.
Plated thru hole interface
Signal trace
SMT pads
Standard PTH
Surface mount interface
Microvia
Signal trace
Figure 5. PTH and SMT technology
7
Microvia with surface mount (SMT) improves signal
Figure 6 shows a typical PTH and microvia structure within a mixed board
stack-up. The same type of SMT lead from the connector is attached to the
PCB but we can use two different techniques to bring the signals from the
connector to the PCB traces. With the illustration on the left, a PTH
connects the SMT pad to the trace near the bottom surface of the PCB. Since
there is no pin inserted into this PTH, the via diameter can be shrunk to a
size that reduces the capacitive effect while still meeting the aspect ratio
requirements of the board vendor. Using this smaller via allows for added
signal performance with respect to a standard PTH, while creating a cost
savings over more expensive via alternatives. Additionally, using a full plated
thru barrel allows signals to be accessed at any layer within the PCB stack,
although accessing signal traces that are close to the surface will introduce
stubbing effects into the signal path.
In the example on the right, a small microvia is used to connect the SMT pad
to an inner board trace. This via can be made even smaller in diameter, in
relation to the PTH, as it is often formed by more precise methods then the
mechanical drilling process used to create PTHs. Selectively stacking
microvias to reach a desired layer allows the board designer to achieve
optimal signal performance by eliminating the electrical stub.
Figure 6. PTH to SMT pad vs. microvia to SMT pad
8
Signal Integrity and Differential Signaling
Backplane data rates are increasing
The proliferation of many new high-speed digital standards depicted in
Figure 7 push the envelope of what's possible on copper. The data must be
transmitted with very few bit errors to maintain system reliability.
Unfortunately, the signal integrity suffers when the risetime of the data
transition from a one to a zero becomes faster. This faster risetime
emphasizes poor design technique of any physical layer component in the
system, including every stripline, microstrip, cable, and connector. Frequency
dependent effects are now commonplace across most high-speed digital
designs and knowledge of transmission line theory is now a requirement for
leading edge design. To complicate matters further, the majority of these
standards utilize differential circuit topology. A paradigm shift in
measurement technology is under way to achieve the goals of the advanced
differential interconnect.
Figure 7. Proliferation of standards
9
Transmission lines are differential
Using Figure 8 as a guide for understanding current flow, imagine two trans-
mission lines that are driven by single ended signals that are exactly out of
phase (we call this differential driving). As the signal propagates down the
differential pair, there is a voltage pattern between each signal line and the
reference plane below. In addition, there is a signal between the two signal
lines. This is called the difference signal or differential signal.
Differential impedance is simply the impedance the difference signal sees
that is driven between the two signal lines in the differential pair. The
impedance the difference signal sees is the ratio of the signal voltage
(difference voltage) to the current in the line. The difference voltage is twice
the voltage of the edges driven into each line. The current into each line is
related to the impedance of each individual line in the pair. There is an
additional current between the signal lines that is due to the coupling
between the traces themselves. This is in general a small amount, but cannot
be neglected. If there were no coupling between transmission lines, the
impedance of a line, as defined by the ratio of the voltage across the paths
and the current through them, would be dependent on just the line parame-
ters of the one line. However, as soon as coupling is introduced, the voltage
on one line may be dependent on the current in an adjacent line. To include
these effects, the concept of impedance or characteristic impedance must be
expanded to allow for one trace interacting with another. This is handled by
expanding the impedance into an impedance matrix. Matrix math is very
useful when quantifying the performance of differential transmission lines,
as will be evident in the next discussion that describes another type of
matrix called the mixed mode s-parameter matrix.
Figure 8. Differential driving
10
Single ended parameters
To lay a foundation for understanding how to characterize a physical layer
device within a 10 Gbps telecom system, a brief discussion of multiport
measurements is in order. The four port device shown in Figure 9 is an
example of what a real world structure might look like if we had two
adjacent PCB traces that are operating in a single-ended fashion. Let's
assume that these two traces are located within relatively close proximity to
each other on a backplane and some small amount of coupling might be
present. Since these are two separate single-ended lines in this example,
this coupling is an undesirable effect and we call it crosstalk.
The matrix on the left in Figure 9 shows the 16 single-ended s-parameters
that are associated with these two lines. The matrix on the right in Figure 9
shows the 16 single-ended time domain parameters associated with these
two lines. Each parameter on the left can be mapped directly into its
corresponding parameter on the right through an Inverse Fast Fourier
Transform (IFFT). Likewise, the parameters on the right can be mapped to
the parameters on the left by a Fast Fourier Transform (FFT).
If these two traces were routed very close together as a differential pair,
then the coupling would be a desirable effect and it would enable good
common mode rejection that provides EMI benefits.
Return loss or TDR
Insertion loss or TDT
Near end crosstalk (NEXT)
Far end crosstalk (FEXT) Port 2
Port 1
Port 3 Port 4
Four-port single-ended device
Frequency domain parameters Time domain parameters
Figure 9. Effects of undesirable coupling on two single-ended lines
11
Single-ended to differential s-parameter
Once the single-ended s-parameters have been measured, it is desirable to
transform these to balanced s-parameters to characterize differential
devices. This mathematical transformation is possible because a special
condition exists when the device under test is a linear and passive structure.
Linear passive structures include PCB traces, backplanes, cables, connec-
tors, IC packages and other interconnects. Utilizing linear superposition
theory, all of the elements in the single-ended s-parameter matrix on the left
of Figure 10 are processed and mapped into the differential s-parameter
matrix on the right. Much insight into the performance of the differential
device can be achieved through the study of this differential s-parameter
matrix, including EMI susceptibility and EMI emissions.
Figure 10. Balanced s-parameters of a differential device
12
Differential s-parameters
Interpreting the large amount of data in the 16-element differential s-parameter
matrix is not trivial, so it is helpful to analyze one quadrant at a time. The
first quadrant in the upper left of Figure 11 is defined as the 4 parameters
describing the differential stimulus and differential response characteristics
of the device under test. This is the actual mode of operation for most high-
speed differential interconnects, so it is typically the most useful quadrant
that is analyzed first. It includes input differential return loss (SDD11), for-
ward differential insertion loss (SDD21), output differential return loss
(SDD22) and reverse differential insertion loss (SDD12). Note the format of
the parameter notation SXYab, where S stands for Scattering Parameter or
S-Parameter, X is the response mode (differential or common), Y is the
stimulus mode (differential or common), a is the output port and b is the
input port. This is typical nomenclature for frequency domain scattering
parameters. The matrix representing the 16 time domain parameters will
have similar notation, except the "S" will be replaced by a "T" (i.e. TDD11).
The fourth quadrant is located in the lower right and describes the perform-
ance characteristics of the common signal propagating through the device
under test. If the device is design properly, there should be minimal mode
conversion and the fourth quadrant data is of little concern. However, if any
mode conversion is present due to design flaws, then the fourth quadrant
will describe how this common signal behaves. The second and third
quadrants are located in the upper right and lower left of Figure 11. These
are also referred to as the mixed mode quadrants. This is because they fully
characterize any mode conversion occurring in the device under test,
whether it is common-to-differential conversion (EMI susceptibility) or
differential-to-common conversion (EMI radiation). Understanding the
magnitude and location of mode conversion is very helpful when trying to
optimize the design of interconnects for gigabit data throughput.
Differential signal Common signal
Figure 11. 16-element differential s-parameter matrix
13
Four Port Microvia Measurements
Measurement set up
The test equipment used in this experiment consists of a 4 port performance
network analyzer (PNA) and 4 channel time domain reflectometer (TDR)
running Physical Layer Test System software. Both instruments are simulta-
neously on the GPIB bus and are used to validate measurements between
each other. See Figure 12 for set up picture.
4-port network analyzer
4-channel TDR scope
Figure 12. Test equipment set up
14
Frequency domain analysis
Now that we have a good understanding about 4-port s-parameters, let's
interpret the actual data in Figure 13. The more intuitive parameter to
review first is typically differential insertion loss or SDD21. This is the
frequency response seen by the differential signal as it propagates through
the device. At lower frequencies (DC to 10 GHz), both vias perform nearly
identical. However, the microvia structure clearly shows less attenuation of
higher frequencies when compared to the standard via. This indicates a
channel structure that allows higher frequencies to pass without significant
degradation. This will inevitably result in an eye diagram that is more open,
as will be shown shortly. The standard via, on the other hand, shows higher
frequencies being attenuated more than the microvia.
The second set of curves is perhaps less intuitive, but equally important to
analyze. The differential return loss (SDD11) indicates the magnitude of
reflections occurring at various frequencies within each structure. Again,
the low frequency response is very similar for both vias. However, the
magnitude of reflections in the standard via is higher than the microvia
from 12 to 20 GHz. Reflections are due to a poorly controlled impedance
environment and the spacing between the nulls is related to the spacing of
the resonant cavity within the structure. In the case of the standard via, this
is related to the length of the via stub.
Differential
return loss of
standard via
Differential
return loss of
microvia
Differential
insertion loss of
microvia
Differential
insertion loss of
standard via
Figure 13. Affects of frequency on standard via and microvia
15
Differential eye diagram analysis
The eye diagrams in Figure 14 are synthesized from the 4-port s-parameters.
This method of creating eye diagrams correlates well with the standard
method of compliance testing with a pattern generator and a sampling scope
with standard masks. As can be seen, the eye diagrams for the microvia are
clearly more open than the standard via, even at 20 Gbps.
Figure 14. Eye diagrams for standard via and microvia
16
Microvia Construction
Laser drilling for microvia forming
With Laser drilling a similar process to mechanical drilling is used except
the holes are formed by the ablation of material by the laser. When accessing
layers beyond layer 1, a number of techniques can be used. Different laser
technologies are often used to ensure that the correct features are formed.
UV-YAG lasers will cut thru metal layers, but will not damage the organic
material of the PCB. CO2 lasers will only cut thru the organic material and
stop when they reach a metal layer. Using these types of lasers allows
precise forming of the via down to the desired layer in the PCB. Again, this
physical forming requires additional real estate.
CO2 lasers operate using wavelengths in the 9 to 11