PIC18F97J60_11 Datasheet, PDF (220/492 Pages) Microchip Technology – 64/80/100-Pin, High-Performance, 1-Mbit Flash Microcontrollers with Ethernet

PIC18F97J60 FAMILY
19.1.5 EMI EMISSIONS CONSIDERATIONS
Most locales have limits on unintentional EMI or EMC
emissions that govern the amount of electromagnetic
energy that may be radiated into the environment
across a range of test frequencies. Ethernet applica-
tions normally do not include intentional radio
frequency emissions sources. They may experience
occasional regulatory failures though, due to the rela-
tive ease at which high-frequency noise may radiate
out of a long attached Ethernet cable. Long cables can
act as unintentional antennas.
The PIC18F97J60 family Ethernet module transmit
engine internally operates by stepping the 25 MHz
base Ethernet clock up to a high frequency via a PLL
embedded in the PHY module. Then, the high
frequency is used to turn on/turn off small current sinks
on the TPOUT+ and TPOUT- pins. This current-mode
drive technique allows the PHY to generate an Ether-
net TX waveform that resembles an analog signal, with
most spectral energy at or below 20 MHz.
However, while low in amplitude, the high frequency
used to synthesize the waveform can, in some applica-
tion circuits, radiate out of the circuit and result in
regulatory emissions compliance failures. Such failures
caused by the Ethernet module will normally be exhib-
ited as excess emissions at 200 MHz and occasionally
400 MHz.
To minimize the chance of failure, the use of the LC
low-pass filter is recommended on the TPOUT+ and
TPOUT- pins, as shown Figure 19-2.
In this circuit, 120 ohm ferrite beads are used along
with 56 pF±5% capacitors to form a low-pass filter with
a -3dB breakpoint that is above 20 MHz, but below
200 MHz. 10Base-T Ethernet signaling requires only
about 20 MHz of spectral bandwidth, so minimal distor-
tion is done to the Ethernet signal by using these filters.
However, noise at 200 MHz or 400 MHz, generated by
the PHY, is reduced by several decibels before having
a chance to radiate out of the application and cause a
regulatory failure. In this circuit, the ferrite beads must
have a saturation current rating of at least 100 mA.
If EMI emissions regulations are stringent in your
locale, additional care should be taken when selecting
the Ethernet magnetics to further minimize unintention-
ally radiating common-mode signals out of the Ethernet
cable. Ethernet magnetics with a high differential to
common-mode rejection ratio should be used.
The differential to common-mode rejection parameter
is normally expressed in magnetics manufacturers’
data sheets, in units of negative decibels across a test
frequency range. In the absence of test data indicating
otherwise, a more negative specification at higher
frequencies is recommended for the PIC18F97J60
family Ethernet module. For example, a device rated
for -40 dB @ 100 MHz is likely preferable to -33 dB @
100 MHz, even if the performance at 30 MHz is similar
or better on the -33 dB @ 100MHz magnetics.
Often, the use of “5-core” magnetics, or magnetics involv-
ing a center tapped inductor or auto-transformer on the
TX path, is also desirable for EMI emissions reasons.
19.1.6
AUTOMATIC RX POLARITY
DETECTION AND CORRECTION
10Base-T Ethernet signaling is performed on the Ether-
net cable as a differentially encoded Manchester data
stream. This signaling is polarized; therefore, it is
required that the RX+ Ethernet signal on the Ethernet
cable reach the TPIN+ pin, and the RX- Ethernet signal
reach the TPIN- pin. Connecting RX+ to TPIN- and RX-
to TPIN+ (by way of Ethernet isolation transformers)
will cause the PIC18F97J60 family Ethernet module to
successfully link with the remote partner. However, all
receive data will be corrupted by the polarity mismatch
and will be internally discarded by the PHY as if it were
noise on the wire.
Higher speed 100Base-TX and 1000Base-T Ethernet
technologies uses different signaling schemes. They
use Multi-Level Transition 3 (MLT3) and Five-Level
Pulse Amplitude Modulation (PAM5) encoding on the
wire, respectively. These encodings are non-polarized.
Therefore, swapping the differential wires will have no
impact on the Ethernet controller's ability to
communicate with the remote node.
A limited number of modern 3rd party 10/100 and
10/100/1000 rated Ethernet devices (switches, routers
and end devices) connect their TX+ and TX- signals to the
incorrect pins on their RJ-45 Ethernet jack. These devices
are not IEEE Standard 802.3 compliant. However,
because 100Base-TX and 1000Base-T communications
continue to work without correct polarization, some 3rd
party vendors mistakenly release their products to
production without catching these polarization errors.
Due to these circumstances, current revisions of the
Ethernet controller in the PIC18F97J60 family of
devices are not compatible with a limited number of 3rd
party Ethernet devices. The PIC18FXXJXX devices will
link up with the partner and the PHY RX activity LED (if
enabled) will blink whenever a packet is transmitted to
the PIC18FXXJXX device. However, no packets will be
successfully received and written in the Ethernet
SRAM buffer when the polarity is incorrect. To eliminate
this problem, and obtain maximum interoperability with
3rd party devices, it is possible to externally add an RX
polarity swapping circuit to PIC18F97J60 family
applications. Figure 19-3 demonstrates the use of bus
switches to facilitate the swapping of the RX signals.
In Figure 19-3, a general purpose output pin is used to
select the polarity of the RX signals. When the select line
is held low, the A ports of the switches will connect with
the B0 ports, leaving the B1 ports disconnected. This will
allow the TPIN+ pin to be connected to Pin 3 of the
RJ-45 jack while TPIN- is connected to Pin 6. These
connections accommodate the IEEE Standard 802.3
specified polarity.
DS39762F-page 220
 2011 Microchip Technology Inc.