ETC 88E1020/88E1020S

Transceiver
Solutions
Gigabit Ethernet Over Copper Performance
by Jason Knickerbocker
June 11, 2003
M A R V E L L
®
W H I T E
P A P E R
Advanced Information
This document provides advanced information about the product described. All specifications described herein are based on design goals only.
Do not use for final design. Visit the Marvell ® web site at www.marvell.com or call 1-866-674-7253 for the latest information on Marvell products.
Disclaimer
No part of this document may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and
recording, for any purpose, without the express written permission of Marvell. Marvell retains the right to make changes to this document at any
time, without notice. Marvell makes no warranty of any kind, expressed or implied, with regard to any information contained in this document,
including, but not limited to, the implied warranties of merchantability or fitness for any particular purpose. Further, Marvell does not warrant the
accuracy or completeness of the information, text, graphics, or other items contained within this document. Marvell makes no commitment either
to update or to keep current the information contained in this document. Marvell products are not designed for use in life-support equipment or
applications that would cause a life-threatening situation if any such products failed. Do not use Marvell products in these types of equipment or
applications. The user should contact Marvell to obtain the latest specifications before finalizing a product design. Marvell assumes no responsibility,
either for use of these products or for any infringements of patents and trademarks, or other rights of third parties resulting from its use. No
license is granted under any patents, patent rights, or trademarks of Marvell. This product may include one or more optional functions. The
user has the choice of implementing any particular optional function. Should the user choose to implement any of these optional functions, it is
possible that the use could be subject to third party intellectual property rights. Marvell recommends that the user investigate whether third party
intellectual property rights are relevant to the intended use of these products and obtain licenses as appropriate under relevant intellectual
property rights.
Marvell comprises Marvell Technology Group Ltd. (MTGL) and its subsidiaries, Marvell International Ltd. (MIL), Marvell Semiconductor, Inc. (MSI),
Marvell Asia Pte Ltd. (MAPL), Marvell Japan K.K. (MJKK), and Marvell Semiconductor Israel Ltd.
Table Of Contents
1.0
Executive Summary
3
2.0
Ethernet Market Dynamics
4
3.0
Tutorial: How Is Gigabit Ethernet Transmitted Over Category 5 Cable?
5
4.0
Tutorial: Theoretical Challenges of Running Gigabit Ethernet
Over Category 5 Cable
5
5.0
Practical (or Real World) Challenges of Running Gigabit Ethernet
9
Over Category 5 Cable
6.0
Impact of Real World Gigabit Networking Considered
12
7.0
The Marvell Alaska® Gigabit Ethernet PHY Transceivers
Reach Well Beyond the 1000BASE-T Standard
13
8.0
The Marvell Alaska® Gigabit Ethernet Physical Layer Family:
Product Overview
17
9.0
References
19
1.0 Executive Summary
Gigabit Ethernet is rapidly becoming the network of choice from the backbone to the
desktop. Like its predecessor, Fast Ethernet, Gigabit Ethernet is enjoying quick market
adoption due to its ability to offer a significant increase in network bandwidth—ten
times more bandwidth—while maintaining backward compatibility to the installed
base of Ethernet networking systems. The IEEE 802.3ab Gigabit Ethernet standard
represents truly extraordinary technology in that the standard defines 1000 Mbps
data transmission over the same copper media defined for Fast Ethernet at 100 Mbps—
Category 5 Unshielded Twisted Pair (UTP) cable. This paper reviews some of the
challenges in running 1000 Mbps data over Category 5 cable and how the IEEE
standard proposed to address them. It is important to note, however, that the actual
practice of running 1000 Mbps of data over existing Category 5 cabling contains many
challenges, where if not properly addressed can potentially cause significant
performance issues, resulting in unreliable networks and costly downtime. At the
heart of these issues—and the solution to a robust and reliable Gigabit Ethernet
system—is the Physical Layer (PHY) device of Ethernet.
In this paper, some of the theoretical (as addressed by the standard) and practical
(real world) challenges of running Gigabit Ethernet over Category 5 cable are explored
and the impact of the PHY transceiver examined. It will be shown that the performance
challenges and pitfalls of running ten times more data over existing Fast Ethernet cable
lines, may be alleviated through the use of robust PHY transceiver technology—devices
that greatly exceed the requirements of the IEEE Gigabit Ethernet standard. Finally, it
will be shown that the technology that achieves these lofty goals is available through
the Marvell Alaska® family of Gigabit Ethernet PHY transceivers.
Sections 2.0 through 4.0 offer the reader an overview of the Gigabit Ethernet market
and the technology defined for its implementation, as detailed in the IEEE 802.3ab
standard. The reader versed in Gigabit Ethernet may choose to start reading at
Section 5.0.
3
Copyright 2003 Marvell
June 2003
2.0 Ethernet Market Dynamics
Over two decades old, Ethernet has become the most widely deployed and well-known
networking technology in the world. Throughout its existence, Ethernet has evolved
from supporting simple workgroup environments to serving as the primary communications
medium for entire corporate campuses. Additionally, through Ethernet’s evolution, the
line speed (or data rate) has surged from 10 to 100 Mbps, and now to 1000 Mbps.
Put into perspective, this represents a 100-fold increase in data bandwidth in less
than 10 years.
The advantage of Ethernet’s rapid acceleration in available network bandwidth is
underscored by the fact that Ethernet technology also enables seamless and easy
migration to higher data rates—incrementally, as the network requires it. Each
successive higher Ethernet data rate is backward compatible to existing, lower
speed, Ethernet. For example, a Gigabit Ethernet Network Interface Card (NIC) will
automatically configure to operate at 1000 Mbps if it is connected to a Gigabit Ethernet
Switch; however, in the event it is connected to a Fast Ethernet switch, the Gigabit
NIC will automatically configure to 100 Mbps. Thus, using Ethernet technology,
companies do not have to undergo an expensive, time-consuming infrastructure
upgrade. They can migrate portions of the network to higher bandwidth as the
network requires it and the budget allows for it.
Fast Ethernet systems currently dominate the network industry as measured by both
shipment rate as well as installed base. In this new millennium, with the continued
dramatic increases in desktop computing power driven by Intranet-based operations
and the increasing demands for time-sensitive delivery between networked users, the
network industry is currently driving towards another transition to higher bandwidth—
ten times the performance of Fast Ethernet—or 1000 Mbps Gigabit Ethernet.
The rapid conversion of the LAN from 10 Mbps 10BASE-T Ethernet to 100 Mbps
100BASE-TX Fast Ethernet heavily contributed to the mass deployment of Category 5
copper cable, which is now dominant within today's networked buildings. One of the
key contributing factors that is driving the successful high volume deployment of
Gigabit Ethernet is the fact that Gigabit Ethernet was defined to operate over the
existing installed base of Category 5 cable.
Copyright 2003 Marvell
June 2003
4
3.0 Tutorial: How Is Gigabit Ethernet Transmitted Over
Category 5 Cable?
To understand how Gigabit Ethernet is transmitted over Category 5 cable, it is
necessary to first take a brief look at some of the key characteristics of the Fast
Ethernet standard. The Fast Ethernet standard specifies MLT-3 line signaling, which
defines a 3-level transmission scheme running at a clock rate of 125 Mbaud. Fast
Ethernet, which transmits/receives 100 million data bits per second, uses a 125 MHz
clock as the standard also specifies 4B/5B (4 bits to 5 bits) encoding/decoding, adding
25% data overhead (4B/5B encoding/decoding guarantees minimum bit transitions,
which is required for the Fast Ethernet over fiber standard, or 100BASE-FX). Another
key consideration is the fact that the 100BASE-TX standard defines two Category 5
twisted-pairs, one for transmit and one for receive.
Similar to Fast Ethernet, the Gigabit Ethernet standard defines a 125 MHz clock
resulting in many advantages. First, using the same clock rate simplified the design
of dual data rate 100/1000 Mbps transceivers. An additional advantage is that the
frequency spectrum of Fast and Gigabit Ethernet are similar—both with a null at 125
MHz. The 1000BASE-T standard achieves 1000 Mbps of data transmission by using
four Category 5 twisted-pairs (as opposed to two pairs used in Fast Ethernet) as well
as a more sophisticated five-level coding scheme called PAM-5. Another important
difference relative to Fast Ethernet is that 1000BASE-T sends and receives data
simultaneously on each of the four pairs.
Of the five levels provided in PAM-5 coding (-2, -1, 0, + 1, +2), four levels are used
for data transmission and represents 2 bits of information. The fifth level is used for
“Forward Error Correction” and will be discussed later in this paper. The symbol rate of
PAM-5 coding is only half that of 2-level signaling, and thus a 250 Mbps data signal can
be transmitted at a rate of 125 Msymbols/sec. Therefore, 5-level PAM coding provides
better bandwidth utilization than binary signaling, in which each transmitted symbol
represents just one bit—0 or 1.
4.0 Tutorial: Theoretical Challenges of Running Gigabit Ethernet
Over Category 5 Cable
As the Gigabit Ethernet standard specifies a ten-fold increase in data bandwidth
while maintaining the same cabling infrastructure deployed for Fast Ethernet, clearly
it stretches the limits of technology. This section reviews some of the issues in
transmitting 1000 Mbps of data over Category 5 and how the standard addresses them.
5
Copyright 2003 Marvell
June 2003
4.1 Signal-To-Noise Ratio (SNR) Margin
SNR margin is a measure of a communications system’s immunity to noise. SNR margin
is expressed in dB and represents the level of additional noise that the system can
tolerate before violating the required Bit Error Rate (BER). For example, an SNR margin
of 3 dB means that if the noise level were increased by 3 dB, the system would be
subject to excessive errors. Simply stated, SNR margin is a measure of how much
additional noise a system can tolerate or how far the system is from not working properly.
Figure 1 below demonstrates that increasing the number of signal levels while
maintaining the same transmit voltage results in a degradation of the SNR margin.
The reason for this is that as the vertical opening of the eye gets smaller, the system
can tolerate less noise before bit errors begin to occur. For example, increasing the
number of logic levels from 3 (MLT-3 used in Fast Ethernet) to 5 (PAM-5 used in
Gigabit Ethernet), while keeping the overall voltage constant at 2 Volts, reduces the
vertical eye opening of each logic state by 50%. The noise voltage required to cause
a symbol error on a 3-level signal is half (or 6 dB lower) than the voltage required to
cause a symbol error on a binary signal. So a 3-level signal has 6 dB less SNR margin
than a binary signal.
Eye Pattern of MLT-3 Signaling
Eye Pattern of PAM-5 Signaling
Time (in ns)
Time (in ns)
Figure 1: MLT-3 and PAM-5 Signal Levels
4.1.1 Forward Error Correction (FEC)
The Gigabit Ethernet standard specifies the use of 5-level PAM-5 coding, although
only 4-levels are required to represent 2-bits of information.The extra level provides
redundancy for use in FEC. FEC provides a second level of coding that helps to recover
the transmitted symbols in the presence of high noise and crosstalk. The error correction
method consists of Trellis coding (4-Dimensional 8-State) in combination with Viterbi
decoding. It is estimated that the FEC error correction logic enhances the system’s SNR
margin by up to 6 dB. Based on this assumption, the extra 6 dB of SNR margin gives
the 5 level encoded signal the noise immunity of a 3 level signal (i.e., similar to the
Fast Ethernet’s MLT-3 three-level encoding).
Copyright 2003 Marvell
June 2003
6
4.2 Total Noise in a Gigabit Ethernet Channel
Each twisted-pair cable in a 1000BASE-T Gigabit Ethernet channel is subject to three
major sources of noise; crosstalk noise, transmit echo noise and ambient noise.
Crosstalk noise consists of two components, near-end crosstalk (NEXT) and far-end
crosstalk (FEXT). Each noise component is discussed in the following sections.
4.2.1 Crosstalk Noise: NEXT & FEXT
Crosstalk is defined as unwanted or undesirable signals caused by the electric or
magnetic fields of one signal affecting another adjacent signal. As there are a bundle
of four adjacent signals in Gigabit Ethernet transmission, crosstalk becomes a complicated
problem to analyze (see figure below). Crosstalk is characterized in reference to the
transmitter. Near-end crosstalk (NEXT) is crosstalk that appears at the output of a wire
pair at the transmitter end of the cable and far-end crosstalk (FEXT) is crosstalk that
appears at the output of a wire pair at the far end of the cable from the transmitter
(see Figure 2).
Insertion Loss
Far Echo
T
250 Mbps
Hybrid
Near Echo
Hybrid
T
R
R
250 Mbps
FEXT_21
NEXT_21
FEXT_31
T
250 Mbps
Hybrid
Hybrid
T
R
R
250 Mbps
NEXT_31
FEXT_41
NEXT_41
T
R
R
T
250 Mbps
T
250 Mbps
Hybrid
Hybrid
250 Mbps
Hybrid
Hybrid
T
R
R
250 Mbps
[ using 4 wire pairs ]
Figure 2: Sources of Noise in a Gigabit Ethernet System
4.2.2 Transmit Echo Noise
As previously stated, Gigabit Ethernet achieves 1000 Mbps operation, in part, by
transmitting and receiving simultaneously over each of the four twisted-pair cables
(250 Mbps equivalent data rate per line). As a direct result of this full duplex transmission,
transmit echo noise is introduced to the system. In simplified terms, the outbound
transmit signal reflects (or echos) off the channel (or cable) and creates an unwanted
signal at the receiver of the same channel. This echo noise signal interferes with the
7
Copyright 2003 Marvell
June 2003
actual receive signal. The IEEE Gigabit Ethernet standard defines echo cancellers which
attempt to separate the transmit echo from the received signal. This concept of echo
also applies to a telephone connection over a single twisted pair. Your phone has builtin echo canceling (much more simplified than Gigabit Ethernet) so you don’t hear your
voice reflected back.
To implement bi-directional full duplex transmission over each of the four Category 5
pairs, the transmitter and receiver are connected to each pair through a directional
coupler circuit, known as a hybrid, which separates the outbound transmit signal from
the inbound receive signal. Hybrid networks with good trans-hybrid loss minimize the
amount of transmitter signal that is coupled into the receiver but cannot remove all
of the transmit echo signal. The residual transmit signal due to the trans-hybrid loss
combined with the cable return loss produce the unwanted echo noise. The magnitude
of the reflection, or echo, is proportional to the return loss of the channel. Return loss,
usually expressed in dB, may be defined as a measure of the dissimilarity between
impedances in metallic transmission lines and loads. The return loss is a measure of
the amount of power reflected due to cabling impedance mismatches—the higher the
return loss, the greater the cable impedance mismatch.
4.2.3 Ambient Noise
Some examples of typical sources of ambient noise include background white noise
and impulse noise generated by power lines and telephone voltages. Ambient noise can
also include interfering wireless signals and alien crosstalk. Due to its random nature,
the ambient noise cannot be cancelled in the receiver and so directly detracts from the
SNR margin of the system. Thus, the best defense against ambient noise is a robust
system with extra SNR margin. The higher the SNR margin, the less prone the system
will be to errors caused by ambient noise.
4.3 Attenuation
Attenuation is defined as the reduction of signal strength during transmission across
the cable (it is the opposite of amplification). Attenuation, measured in dB, increases
(worsens) with increased frequency and distance. If the signal attenuates too much,
it becomes unintelligible, and will result in network errors. Gigabit Ethernet (as well
as 10 and 100 Mbps Ethernet) define a maximum cable distance of 100 meters. Thus,
given known attenuation models for Category 5 cable, it may be determined what the
minimum transmit amplitude should be to guarantee a certain minimum distinguishable
voltage level at the downstream receiver—up to 100 meters away. Based on this
logic, the IEEE Gigabit Ethernet standard defines these minimum voltage levels. It is
important to note, however, that as the receiver sensitivity is increased, extra cable
distance margin is achieved (i.e., in the presence of higher attenuation, the receiver
can still distinguish voltage logic levels and accurately recover data).
Copyright 2003 Marvell
June 2003
8
5.0 Practical (or Real World) Challenges of Running Gigabit
Ethernet Over Category 5 Cable
Section 4.0 of this paper reviewed some of the key issues/challenges in running 1000
Mbps of data over existing Category 5 cable, and what the IEEE standard for Gigabit
Ethernet specifies as a solution. It is important to note that in order to generate an
industry standard such as IEEE 802.3ab 1000BASE-T, several assumptions dictating
the content of the standard had to be made. Below, a subset of the assumptions
are highlighted:
• Standards-compliant Category 5 cabling is used as specified in ISO/IEC
11801:1995 and ANSI/EIA/TIA-568-A (1995) and tested for additional
performance parameters specified in Section 40.7 of the IEEE 802.3ab standard.
• Four (4) twisted-pairs of the above-described standards compliant Category 5
cable are available.
• The network supports a BER of less than or equal to 10-10.
• The horizontal cabling subsystem is no greater than 100 meters as specified
in ANSI/EIA/TIA-568-A and ISO/IEC 11801.
• The horizontal cabling subsystem consists of standards-compliant Category 5
patch cords, cables and connecting hardware as specified in ANSI/EIA/
TIA-568-A and ISO/IEC 11801.
• The impedance of the Category 5 cable is 100 ohms +/- 15%.
• The Gigabit systems meet or exceed FCC Class A/CISPR operation.
The environment required by the 1000BASE-T standard for successful transmission of
Gigabit data exists for the majority of the current installed base of network infrastructure.
However, due to the extreme requirements and challenges of the 1000BASE-T standard,
as well as the fact that cabling subsystems have been evolving over time, there are
many circumstances in which a 1000BASE-T installation will fail. In the remainder of
this section, we will explore some of these cases.
5.1 The Quality and Existence of Category 5 Cabling Is Not a Given
Like networking standards, cabling standards are constantly changing and evolving to
meet the changing and growing needs of the industry. Cabling standards are primarily
defined by the ANSI/EIA/TIA-568 specification. The initial ANSI/EIA/TIA standard was
published as recent as 1991 and, at that time, was targeted for 10BASE-T electricals
or Category 3 cable with a maximum frequency of 16 MHz. Prior to this specification,
cabling was not standardized, and as a result, cabling existed that was sub-Category 3.
Additionally, a standard-defined Category 4 grade of cable was created with an operating
frequency up to 20 MHz, and was intended for use in Token Ring networks. In 1995,
Category 5 cable was standardized (100 MHz frequency), with the initial driving force
being a copper version of the fiber-based FDDI networking protocol, and later, Fast
Ethernet. With respect to Gigabit Ethernet performance, it’s important to realize that
9
Copyright 2003 Marvell
June 2003
not all Category 5 cable was/is created equal. Category 5 cable existed prior to the
standard being ratified and as a result could result in poorer performance. Additionally,
the performance of Category 5 cable is, in part, a function of the chemical compounds
used in the manufacture of the cable. It is estimated that from 1994 on, there are 105
different electrical designs of Category 5 cable.
In summary, the majority of the installed base of cabling is Category 5 rated compliant
to the ANSI/EIA/TIA-568-A standard; however, as cabling technology and standards
have been evolving over the past decade or so, there exists sub-Category 5 cable in
today’s installed base. When constructing Gigabit Ethernet networks using system
equipment that merely meets the IEEE 802.3ab 1000BASE-T standard in these less
than ideal cabling installations, there exists the possibility for significant network
performance problems.
5.2 The Length or Attenuation of the Cable Is Not Always to Specification
As stated previously, the ANSI/EIA/TIA-568-A and ISO/IEC 11801 standards specify
that the horizontal cabling subsystem is to be no greater than 100 meters. The
horizontal cabling subsystem is defined as the connection between a workstation
and the local telecommunications closet and consists of cables, patch cords and
connecting hardware (see Figure 3). With proper, certified installation it is very likely
that the 100 meter maximum distance limit will be guaranteed—in fact, this is the case
in the vast majority of new cabling installations. However, as with most things, there
are exceptions. There are numerous cases where the 100 meter maximum distance
limit is exceeded due to uncertified or careless installations, as well as intentional
violations. For example, consider a case where a small number of workstations slightly
exceed the 100 meter distance limit to the wiring closet. As opposed to incurring
the expense of installing an additional intermediary wiring closet, the distance limit
may be stretched.
Work Area
Transition Point Connector
Telecom Closet
Wall Jack
Work Area
Cable
TO
Equipment
Cable
Interconnect Crossconnect
Channel
Figure 3: ANSI/EIA/TIA Horizontal Cabling Subsystem
As discussed, a key variable in determining cable attenuation is the distance of the
cable—attenuation is proportional to cable distance. However, there are additional
real-world variables that will affect/increase the attenuation of the cable. Some cabling
installations will exhibit higher attenuation due to poorer grades of cable or due to
imperfections in cabling installations resulting in kinked or distorted cables. The
additional attenuation due to these imperfections will decrease the budget for cable
length, or conversely, increase the effective cable length (i.e., a 100 meter cable will
have the attenuation characteristics of a 120 meter cable).
Copyright 2003 Marvell
June 2003
10
5.3 Is the Horizontal Cabling Subsystem Completely Category 5 Rated?
As defined in Section 5.2, the horizontal cabling subsystem is defined as the connection
between a workstation and the local telecommunications closet and consists of cables,
patch cords and connecting hardware (see Figure 3). In addition to specifying the
cabling, the ANSI/EIA/TIA-568 standard defines that the remaining components of the
horizontal cabling subsystem—patch cords, connecting hardware, wall outlets, and
punch-down blocks—be Category 5 or better. In real world installations, however,
occasionally incorrect hardware is used. For example, a Category 3 connector may
inadvertently be used. Incorrect or faulty connectors result in impedance mismatches
in the link and contribute significantly to increased echo noise (as discussed in Section
4.2), which could exceed the SNR margin of the system and result in excessive BER.
5.4 Cable Installation Is a Factor
Cable installation is also a possible source of network errors, even when the proper
standards-defined cable is used. Cable installation methods and procedures are also
standardized by the EIA/TIA-568-A specification. Category 5 cable requires careful
and precise installation to maximize efficiency and system performance. Anything that
kinks the cable or disturbs the precise alignment of the pairs inside the cable has the
potential to create future performance problems. Some sources of failure are as follows:
• Excessive force used when “pulling” the cables during installation. The EIA/
TIA-568-A specification limits the pulling tension to 25 pounds maximum.
• Crushed or pinched cable bundles from sharp edges during installation or
from applied weight.
• Crushed cable bundles from support hardware or staple guns.
• Improper termination to connecting hardware. To meet Category 5 cabling
standards, the untwisting that takes place where cabling is terminated to
connecting hardware should not exceed 13 millimeters, or a 1/2 inch.
• Loose, worn or abused wiring can cause the twisted pair to untwist more than
necessary. The greater the distance that wires run parallel to one another, the
greater likelihood that magnetic fields might cause impedance mismatch or
induce current charges.
• Splicing of wires between the telecom closet and outlet locations.
11
Copyright 2003 Marvell
June 2003
6.0 Impact of Real World Gigabit Networking Considered
As highlighted in the above sections, there are several examples of real world
networking environments that exceed the requirements and assumptions of the
1000BASE-T standard. For the successful market adoption and mass deployment
of 1000BASE-T systems, it is critical that Gigabit Ethernet systems not only be
compliant to the 1000BASE-T standard, but that they exceed the requirements of
the standard enabling the deployment of Gigabit systems in the vast majority of
networking environments. Because in the end, when an end-user or IT personnel
purchase and install 1000BASE-T Gigabit Ethernet equipment, it must work properly
and transparently. This after all, is the legacy of Ethernet technology—it is simple
and works with little or no intervention.
Networking systems manufacturers benefit greatly from ensuring that their equipment
performs to specification, even in the event that the equipment is installed in an
environment that may be in violation of the requirements of the 1000BASE-T standard.
Systems manufacturers do not necessarily have control over the end-user’s cabling
plant/infrastructure, and thus by building extra performance margin into their products,
they are able to cover a wider range of installations. When networking systems fail to
function properly upon purchase and installation, the systems manufacturer is exposed
to the following disadvantages:
• Increased cost for customer support and thus decreased product profit margins.
Support includes phone calls into the support line as well as the potential for
field visits.
• Negative impact to the systems manufacturer’s quality and customer
satisfaction level.
• Potential for lost business/revenue. The end-user or IT personnel may decide
to replace the non-working systems with equipment from a competing supplier.
• For the customer, network reliability is a top priority. If the network equipment
is unstable or unreliable in any given installation, there is higher probability of
network down time. For the end-user, network downtime directly translates to
lost productivity/revenue.
Copyright 2003 Marvell
June 2003
12
7.0 The Marvell Alaska® Gigabit Ethernet PHY Transceivers Reach
Well Beyond the 1000BASE-T Standard
As highlighted in this section, the Marvell Alaska® family of Gigabit Ethernet PHY
devices greatly exceed the IEEE 802.3ab 1000BASE-T standard. By utilizing Alaska
technology, Gigabit Ethernet systems maximize the number of installations a given
1000BASE-T network may deploy. Some of the key advantages of the Marvell Alaska
technology are:
• Full duplex Gigabit transmission up to 180 meters of Category 5 cable while
maintaining a BER of 10-10 or better. This represents a 80% margin in cable
distance relative to the 1000BASE-T standard.
• Tri-speed (10, 100, 1000 Mbps) operation up to 100 meters of Category 3 cable.
• Superior echo cancellation circuitry permits full duplex Gigabit transmission up
to 125 meters of Category 5 cable under worst case impedance mismatch scenario.
The test involved 20% impedance mismatch (100 ohm cable to 120 ohm cable).
• Verified operation even with the use of non-Category 5 connectors and cords.
• Superior alien crosstalk noise rejection and better common mode noise rejection.
• Superior transmit wave shaping techniques resulting in FCC Class
B/CISPR qualification.
Some of the advantages of the Alaska Gigabit Ethernet PHY transceiver technology
are discussed in more detail in the following sections.
7.1 Industry-Leading DSP Technology
Simply stated, a Gigabit Ethernet PHY transceiver is only as good as the DSP
technology used in its implementation. Marvell has a long sustained track record of
delivering to production the industry’s most advanced Digital Signal Processing (DSP)
technology. Since 1995, Marvell has introduced five generations of industry-leading
read channel ICs for the storage market. The read channel may be viewed as the PHY
transceiver of the disk drive and is highly DSP centric. Additionally, there are several
similar functional blocks between the read channel and 1000BASE-T physical layers.
By leveraging the advanced DSP technology developed for the read channel, Marvell
was able to quickly come to market with a highly advanced 1000BASE-T PHY device,
greatly exceeding the performance requirements of the IEEE standard as well as
competing solutions.
As discussed in previous sections, the clock rate of the 1000BASE-T PHY transceiver
is 125 MHz. Thus, the DSP engines implementing the PHY device also run at 125 MHz.
As shown in the Figure 4 below, during the development of the initial Alaska Gigabit
PHY product, Marvell was in production with a read channel that operated at
frequencies up to 750 MHz; therefore, the DSP engines were running at 750 MHz
(1.3 GHz read channels are now currently available from Marvell). Thus, from a DSP
technology standpoint, the requirement of 125 MHz DSP for the Gigabit PHY transceiver
13
Copyright 2003 Marvell
June 2003
did not represent a significant technology barrier for Marvell. On the other hand, other
semiconductor suppliers started with DSP technology operating at much lower speeds,
and thus were forced to “stretch” the technology up to 125 MHz. When technology
must be stretched, the result is less efficient designs from both a power and die size
perspective, as well as an impact on the performance capability of the devices.
PHY
System
Read Channel
Disk Drive
Media
Clock Rate
750 MHz
Marvell
Transceiver
Fast and Gigabit
Ethernet
125 MHz
Cable Modem
Transceiver
Cable Modem
25 MHz
Competition
Baseband
Processor
Wireless
Ethernet
22 MHz
Figure 4: The Marvell Industry-Leading DSP Technology
7.2 Alaska PHY Exhibits Superior Cable Distance Performance
The Alaska Gigabit Ethernet PHY transceivers feature the industry’s greatest cable distance
margin, operating up to 180 meters of Category 5 cable in full duplex transmission
mode. The “eye diagram” of the Alaska device after transmission over 140 meters
of standard Category 5 Unshielded Twisted Pair (UTP) cable is shown in Figure 5. The
key issue to highlight is that the “eyes” at 140 meters are wide open. The eyes are
the black openings between the signal traces. Generally, a larger eye opening results
in better BER performance. To determine the proper logic level, the waveform is
sampled in the middle of the data “eye” in between data transitions as this is the
location where the waveform is most settled, and the height of the eye opening is
at its maximum. Other than the Marvell Alaska PHY devices, no other Gigabit PHY
products in the industry operate at 140 meters—the link cannot be established, the
eye does not exist (it is closed).
Figure 5: The Eye Diagram at 140 Meters Category 5 Cable
Using the Marvell Alaska Gigabit PHY Transceiver
Copyright 2003 Marvell
June 2003
14
7.3 Alaska PHY Exhibits Superior Echo Cancellation Performance
As previously discussed, Gigabit Ethernet achieves 1000 Mbps operation, in part, by
transmitting and receiving simultaneously over each of the four twisted-pair cables
(250 Mbps equivalent data rate per line). As a direct result of this full duplex transmission,
transmit echo noise is introduced to the system. The outbound transmit signal reflects
(or echos) off the channel and creates an unwanted signal at the receiver of the same
channel. Although the ANSI/EIA/TIA-568 standard defines that the components of the
horizontal cabling subsystem (cable, patch cords, connecting hardware, wall outlets,
and punch-down blocks) be Category 5 or better, occasionally incorrect hardware is
used. The use of cabling not exhibiting 100 ohm impedance, or the use of incorrect or
faulty connectors result in impedance mismatches in the link and contribute significantly to
increased echo noise. The Marvell Alaska family of Gigabit Ethernet transceivers feature
extremely robust echo cancellation circuitry resulting in added SNR margin and proper
functionality meeting or exceeding the specified BER— even in the presence of these
non-standard cabling environments.
Figure 6 shows a test setup defined in the IEEE standard and the UNH Interoperability
Lab. To create impedance mismatch, 1 meter segments of 120 ohm cables are used at
each end of the wiring subsystem. The 100 ohm impedance Category 5 cable will vary
in length for this test. In this test scenario, a 20% change in impedance causes a 9%
reflection (or echo) of the signal back to the transmitting node. In the above test
setup, systems containing the Marvell Alaska PHY transceivers function properly with
the 100 ohm Category 5 cable segment up to 125 meters in length. Running the same
test using Gigabit Ethernet systems based on non-Marvell devices, result in errors
(down link) beyond 41 meters of 100 ohm Category 5 cable.
120 Ω
Cable
(1 m)
Gig PHY
100 Ω
CAT5
Cable
CAT5
Connector
120 Ω
Cable
(1 m)
CAT5
Connector
Gig PHY
Figure 6: Gigabit Ethernet Impedance Mismatch Test Setup
15
Copyright 2003 Marvell
June 2003
The Alaska Gigabit PHY devices feature extremely robust noise cancellation circuitry,
greatly exceeding the requirements of the standard. The DSP-based NEXT and echo
cancellers implemented in the Alaska PHY transceivers are the largest functional block
of IC (Integrated Circuit), consuming nearly 40% of die real estate. There are 160
echo taps and 20x3 NEXT taps for each of the four channels contained in the Alaska
device. Competing Gigabit PHY transceiver solutions typically implement only as
many as 60 taps.
7.4 Alaska PHY Exhibits Superior EMI Performance
As reviewed in Section 5.0, the 1000BASE-T standard requires only that Class A/CISPR
FCC requirements be met. Using superior pulse shaping techniques, the Alaska family
of devices meets the more stringent Class B/CISPR FCC requirements. Pulse shaping
is used to minimize the transmitted signal energy at frequencies where distortion
and disturbances are significant. Rather than typical wave shaping filters, the Alaska
devices use a complex SIN(X)/X function for superior performance.
7.5 The 2-Pair Problem of Gigabit Ethernet
As reviewed in prior sections, the Gigabit Ethernet standard requires four Category 5
pairs for Gigabit transmission. However, in actual cabling systems, there are cabling
installations that contain only two Category 5 cables. Typically cabling bundles include
four twisted-pair conductors; however, in many cases IT personnel have routed only
two twisted-pair conductors to an end-user. For example, consider a case where a 4pair bundle is split between two users in order to reduce installation costs. This would
be a common practice particularly in environments of employee expansion (expansion
beyond the existing cable infrastructure), or in environments where end-users are
consolidated in a single office/cubicle.
Prior to Gigabit Ethernet technology, splitting a four-pair bundle into two sets of two
twisted-pairs could be viewed as acceptable, as the Ethernet technology at the time
(both 10BASE-T and 10/100BASE-T) required only two twisted-pairs for transmission.
Today, deploying Gigabit Ethernet systems in a two-pair cabling environment would
result in disaster—a link or connection would not occur. In this environment, the
Auto-Negotiation process would be carried out successfully as this scheme involves
transmission on only 2 twisted-pairs. Further, the Auto-Negotiation process would
inform the Gigabit systems at each end of the link that the other is Gigabit ready.
Data transmission would proceed, however, this will result in failure as Gigabit data
transmission requires four pairs. The data link would then go back into AutoNegotiation mode and this process will continue infinitely.
Marvell recognized this issue and as a result designed a feature into the Alaska PHY
devices that will automatically detect this 2-pair environment and downshift operation
to 100 Mbps. Downshifting to 100 Mbps enables the network to properly link. Under
similar conditions, competing devices cannot link and the network would go down.
Copyright 2003 Marvell
June 2003
16
8.0 The Marvell Alaska® Gigabit Ethernet Physical Layer Family:
Product Overview
The Marvell Alaska® family of Gigabit Ethernet over copper transceivers, currently
consisting of sixteen different devices, has offered the networking industry the enabling
technology required to ease and accelerate the deployment of Gigabit Ethernet to the
desktop and throughout the network. Marvell continues to be the undisputed technology
and market leader in the Gigabit Ethernet PHY transceiver marketplace by setting
multiple industry and technology milestones with its single, dual and quad-port
Alaska family of Gigabit Ethernet transceivers.
With its first generation single-port Alaska and Alaska+ PHY transceivers, which are
currently in volume production, Marvell set a new technology standard by offering the
lowest power dissipation, smallest form factor packaging and highest performance
devices. Leveraging this Gigabit Ethernet technology expertise, Marvell again led the
market with the introduction of its Alaska II and Alaska II+ PHY transceivers—the
world's first dual-port Gigabit Ethernet PHY transceivers offering the networking
industry even higher integration and lower system cost.
Targeted at high-density Gigabit Ethernet switches, the Marvell Alaska Quad Gigabit
Ethernet transceiver family represents a significant technology breakthrough for the
Gigabit Ethernet market by offering the world's most complex and highly integrated
mixed-signal DSP-based quad-port communications devices ever developed. The Alaska
Quad family offers the industry's lowest power and highest integration Gigabit Ethernet
over copper PHY transceivers. When used with stacked RJ45s, the Alaska Quad device
enables 48-port Gigabit systems.
All of the Marvell Alaska Gigabit Ethernet transceivers, the Alaska+, Alaska II+ and
Alaska Quad+ devices, provide system manufacturers with the ability to bridge fiberoptic and copper networks with the built-in serializer/deserializer (SERDES) function.
8.1 The Marvell Alaska Ultra Family of Single-Channel 1000BASE-T PHY Transceivers
The Marvell Alaska Ultra device is the industry’s first Gigabit Ethernet over copper
transceiver optimized for the mobile computing market. The Alaska Ultra transceiver
delivers the lowest power consumption, smallest form factor package and advanced
power management modes. The device supports advanced power management modes
including support for Wake-on-LAN, reducing power consumption to near-zero. These
low power management modes are critical to client applications such as Gigabitconnected laptop computers, where the use of the Alaska Ultra device can
substantially prolong battery life.
17
Copyright 2003 Marvell
June 2003
The Marvell Alaska Ultra+ device, the 88E1011S, is the industry’s first and only Gigabit
solution to provide complete flexibility in media selection—either copper or fiber media.
The Alaska Ultra+ transceiver offers simultaneous support of both copper and fiber
interfaces, whereby the copper port is connected to a transformer and RJ45 connector,
and the fiber port is connected to a Gigabit fiber-optic module or GBIC device. The
device’s Media Detect feature automatically senses whether the end-user has inserted
a Category 5 copper or fiber-optic cable, performs Auto-Negotiation and provides
seamless plug-and-play configuration.
Part Number
Number of Ports
Package
PHY/MAC Interface
SERDES Media
Interface
Media Detect
88E1011
1
117-pin TFBGA or
128-pin PQFP
GMII, RGMII,
SGMII, TBI, RTBI
No
No
88E1011S
1
117-pin TFBGA or
128-pin PQFP
GMII, RGMII, TBI,
RTBI
Yes
Yes
Table 1: The Marvell Alaska Ultra Device Seletion Table
8.2 The Marvell Alaska Quad Family of 1000BASE-T Gigabit Ethernet PHY Transceivers
The Marvell Alaska Quad Gigabit Ethernet transceiver family represents a significant
technology breakthrough for the Gigabit Ethernet market by offering the world’s most
complex and highly integrated mixed-signal DSP-based quad-port communications
devices ever developed. The Alaska Quad family offers the industry’s lowest power
and highest integration Gigabit Ethernet over copper (10/100/1000BASE-T) PHY
transceivers. When used with stacked RJ45’s, the Alaska Quad enables 48-port Gigabit
systems. The Alaska Quad+ (88E1040S/88E1041S/88E1042S) transceivers provide
flexibility in the selection of copper or fiber-optic interfaces on a per-port basis using
a built-in SERDES function.
Part Number
Number of Ports
Package
Package
Dimension
Package Ball
Pitch
PHY/MAC
Interface
SERDES Media
Interface
88E1040/
88E1040S
4
352-pin
TBGA
35mm x 35mm
1.27mm
GMII, RGMII,
SGMII, TBI, RTBI
88E1040S Only
88E1041/
88E1041S
4
352-pin
TBGA
27mm x 27mm
1.0mm
GMII, RGMII,
SGMII, TBI, RTBI
88E1041S Only
88E1042/
88E1042S
4
256-pin
TBGA
27mm x 27mm
1.27mm
RGMII, SGMII,
RTBI
88E1042S Only
Table 2: The Marvell Alaska Quad Device Seletion Table
Copyright 2003 Marvell
June 2003
18
9.0 References
19
[1]
“Physical Layer Parameters and Specifications for 1000 Mbps Operation Over 4-Pair of
Category 5 Balanced Copper Cabling, Type 1000BASE-T,” IEEE Std. 802.3ab, 1999.
[2]
He, R., et al., “A DSP Based Receiver for Gigabit Ethernet over CAT-5 Cables,”
ISSCC Digest of Technical Papers, February 2001.
[3]
P. Roo, et al., “A CMOS transceiver analog front-end for Gigabit Ethernet over
CAT-5 cables.” Submitted to ISSCC 2001, Subject Area: Wireline communications.
[4]
“Gigabit Ethernet 1000BASE-T Whitepaper.” Gigabit Ethernet Alliance, 1997.
[5]
“Running 1000BASE-T: Gigabit Ethernet Over Copper Whitepaper.”
Gigabit Ethernet Alliance, September 1999.
[6]
“Gigabit Ethernet Over Category 5.” Agilent Corporation.
[7]
“802.3ab, A Tutorial Presentation.”
[8]
“Category 5: How Did We Get Here and Where Do We Go Next?” Anixter Inc, 1997.
[9]
“Do You Really Know Your Cable Configuration?” Danpex Corporation Web Site.
http://www.danpex.com/faqs/cable.htm/.
Copyright 2003 Marvell
June 2003
Marvell Semiconductor, Inc.
700 First Avenue
Sunnyvale, CA 94089
Phone 408.222.2500
www.marvell.com
Copyright © 2003, Marvell. All rights reserved. Marvell, the Marvell logo, Moving Forward Faster, Alaska, and GalNet are
registered trademarks of Marvell. Discovery, Fastwriter, GalTis, Horizon, Libertas, Link Street, NetGX, PHY Advantage, Prestera,
Raise The Technology Bar, UniMAC, Virtual Cable Tester, and Yukon are trademarks of Marvell. All other trademarks are the
property of their respective owners.