F-Tile PMA and FEC Direct PHY Multirate Intel® FPGA IP User Guide

ID 720998
Date 12/19/2022
Public

A newer version of this document is available. Customers should click here to go to the newest version.

Document Table of Contents

5. Block Description

The F-Tile PMA/FEC Direct PHY Multirate Intel FPGA IP enables you to use all the features and components of the F-Tile to get more bandwidth than what is available in the existing single rate F-Tile PMA/FEC Direct PHY Intel FPGA IP.
The following block diagram shows the interconnections of F-Tile PMA/FEC Direct PHY Multirate Intel FPGA IP. It consists of a soft IP (SIP) and hard IP (HIP) module that are used for startup, and profile instances.
Figure 18. Block Diagram

The F-Tile PMA/FEC Direct PHY Multirate Intel FPGA IP includes a soft IP (SIP) that enables a customer to use all the features and components of the F-tile transceiver hard IP (HIP) to get more bandwidth than what is available in the existing single rate F-Tile PMA/FEC Direct PHY Intel FPGA IP.

The F-tile is connected to the FPGA fabric using embedded multi-die interconnect bridge (EMIB) technology. The EMIB Deskew block corrects for possible skew over the EMIB interfaces between the main FPGA die and the F-Tile.

The PMA can be of FGT type. The F-Tile PMA/FEC Direct PHY Multirate Intel FPGA IP gets its clock from the system PLL. Internally generated clocks are used to drive the data. The default base profile represents the maximum resources available in the generated IP. It is labeled Profile #0 in the GUI.

The IP receives its main clock input from the F-Tile Reference and System PLL Clocks Intel FPGA IP. The system PLL frequency must be at least twice the frequency of the rate at which parallel data is driven or received. The default system PLL frequency in the IP is 830.078125 MHz and the reference clock frequency for the TX PLL lock and RX CDR is 156.25 MHz.

The soft IP module (SIP) is the control logic for the F-Tile PMA/FEC Direct PHY Multirate Intel FPGA IP. It includes glue logic that multiplexes signals to the correct PMA lane. For example, a 1-bit tx_reset signal can directly go to the corresponding transmitter. The bit width of tx_reset port is equal to the maximum number of fractures possible. For example, in a 100G-4 Reconfigurable group, if the system starts up as one fracture using all four PMA lanes, then the reset for all lanes should come from tx_reset[0].

The purpose of the custom cadence logic is to allow you to pace tx data valid with tx_cadence signal in case there is a possibility of FIFO overflow. The functionality is same as the single rate F-Tile PMA/FEC Direct PHY Intel FPGA IP core.

The deskew logic includes TX deskew pulse insertion in the F-Tile PMA/FEC Direct PHY Multirate Intel FPGA IP. The logic is used to align the lanes belonging to the same transmitter together at the receiver for PMA direct mode. For FEC direct mode, the logic aligns lanes belonging to the same fracture and subsystem.