![Xilinx Virtex-4 RocketIO User Manual Download Page 128](http://html1.mh-extra.com/html/xilinx/virtex-4-rocketio/virtex-4-rocketio_user-manual_3383739128.webp)
128
Embedded Tri-Mode Ethernet MAC User Guide
UG074 (v2.2) February 22, 2010
Chapter 4:
Physical Interface
R
If the PHY device sources the receiver SGMII stream synchronously from the shared
oscillator (refer to the PHY data sheet), the MGT receives data at exactly the same rate
as that used by the core. That is, the RX elastic buffer neither empties nor fills because
the same frequency clock is on either side.
In this situation, the RX elastic buffer does not underflow or overflow, and the RX
elastic buffer implementation in the MGT is recommended to save logic resources.
•
Case 2
, assume that both clock sources used are
50 ppm. The maximum frequency difference between the two devices is 100 ppm,
translating into a full clock period difference every 10000 clock periods and resulting
in a requirement for 16 FIFO entries above and below the half-full point. This case
provides reliable operation with the MGT RX elastic buffers. However, the designer
must check the PHY data sheet to ensure that the PHY device sources the receiver
SGMII stream synchronously to its reference oscillator.
Using the FPGA Logic Elastic Buffer
illustrates a simplified diagram of a situation where the Ethernet MAC in
SGMII mode is interfaced to an external PHY device with an independent clock. The
MGT’s elastic buffer has been bypassed and the FPGA elastic buffer is used.
Using the SGMII in this configuration eliminates the possibility of buffer error if the clocks
are not tightly controlled enough to use the MGT elastic buffer.
Figure 4-21:
SGMII Implementation Using a Logic Buffer
Ethernet MAC
MGT
TXP/TXN
RXP/RXN
Twi
s
ted
Copper
P
a
ir
S
GMII Link
10BA
S
E-T
100BA
S
E-T
1000BA
S
E-T
PHY
FPGA
100 ppm
125 MHz – 100 ppm
GT11
El
as
tic
B
u
ffer
RX
El
as
tic
B
u
ffer
UG074_3_83_012408
www.BDTIC.com/XILINX