Sifos Technologies PhyView PVA-3000, Reference Manual

Page: 10 / 134

PVA-3000 Reference Manual
December 2, 2019 Sifos Technologies Page
10
On the Internet level, the majority of traffic is governed by a transport layer, most commonly TCP, that has intelligence
to know when packets are missing because they were dropped in some link along the network path. TCP can then
communicate to the original sender that certain packets need to be re-transmitted, thus giving them all a second chance
to traverse the network path.
The bottom line is that it is difficult for a user or observer to know when an Ethernet link gets into trouble unless the
problem is so severe that the link is essentially dead. Problems in physical layer communication can be concealed by
an array of self-healing behaviors both in the link and in the network levels.
1.2.3. Traditional PHY Conformance Testing
To some degree, IEEE 802.3 specifications have taken ownership for defining the types of tests and test methods that
should be applied to verify conformance to each physical layer standard. This is particularly true of 10BaseT and
1000BaseT. 100BaseTx was largely imported from the FDDI TP-PMD (ANSI X3:263 1995) and did not incorporate
as much in the way of testing details.
Much of the focus on testing in these documents related to measurements of PHY transmitter characteristics. This is
not surprising because those are far easier to measure than are PHY receiver characteristics. Common PHY
transmitter parameters that appear throughout these specifications include:

Signal Amplitude (min & max)

Transition Slew Rate or Time (min & max)

Signal Symmetry in Amplitude and/or Time (min)

Overshoot & Droop (max)

Edge Jitter (max)

Return Loss and/or Differential Impedance
Receiver testing, when described, is usually limited to Bit Error Rate (BER, or Packet Error Rate, PER) checks. Each
802.3 PHY specification does clearly spell out the types of channel impairments that those physical layers must assume
to exist, meaning that in theory, receiver testing and qualification must be done using representative worst case channel
impairments. However, the actual testing methods and measurements are loosely defined. Similarly, the 802.3
specifications give wide design discretion to topics such as receiver error detection and indication criteria.
1.2.4. Time Domain Measurements in PHY Conformance Testing
Originally, the baseband nature of 10BaseT signals meant that time domain analysis using oscilloscopes was the natural
approach to qualifying transmitted signal integrity to the standard. Most of the above listed transmitter parameters (
see
Section 1.2.3 ) were a good fit with time domain analysis. With the advent of digital scopes, pulse masks could readily
be defined and applied to captured waveforms. Both IEEE 802.3 clause 14 (10BaseT) and clause 40 (1000BaseT)
define test procedures that include a detailed pulse mask.
The only inconvenience in testing twisted pair Ethernet signals with an oscilloscope was that typical scope interfaces
are either 50

or 1M

coaxial while each Ethernet twisted pair was a balanced 100

channel. This problem could
readily be resolved with a matching transformer (or balun) that would convert balanced 100

to coaxial 50

, the only
limitation being that the balun now became a part of the device-under-test.
An alternative interface method was to take each conductor of the twisted pair and route them as center (signal)
conductors to a 2-channel oscilloscope whereby the signal analyzed would be the difference between the two channel
input. This way, the 50

inputs could combine to a total 100

termination on the twisted pair, though the pair would
no longer be electrically isolated. A third alternative would be to
terminate the twisted pair in a nominal 100

load, then probe the
load with a high impedance, high bandwidth differential (active)
probe. This approach keeps the twisted pair isolated but adds cost
and possible calibration of the active differential probe.
The good news about interfacing to the original 10BaseT
transmitter was that there was only a single pair to ever be
concerned with on any given PHY interface. This changed when
auto-sensing MDI/MDI-X was introduce whereby a 10/100BaseT
PHY would have to be able to transmit on either of two pairs
meaning in effect that there were two transmitters that needed to
be tested. 1000BaseT then complicated this further since all 4
pairs possess transmitters meaning that more elaborate fixturing
was required with more manual connection interventions to
perform measurements across the 4 pairs(
see Figure 1.2 ). This
Figure 1.2 Ethernet PHY Test Fixtures