Clocking - 3.1 English - PG300

This section describes the link clock (rx_lnk_clk), video clock (rx_vid_clk), and Video Bridge to AXI4-Stream master interface clock. When the AXI4-Stream interface is selected for DP RX, the output clock of Video to AXI4-Stream bridge that is m_axis_aclk_stream must be selected such that it is greater than or equal to rx_vid_clk.

The rx_lnk_clk is a link clock input to the DisplayPort 1.4 RX Subsystem generated by the Video PHY (GT).

The hdcp_ext_clk input can be driven from external MMCM or BUFGCDIV where it has a frequency requirement of hdcp_ext_clk = rx_lnk_clk/2 MHz.

The following table shows the clock ranges.

The core uses the following clock domains:

lnk_clk

The rxoutclk from the Video PHY is connected to the RX subsystem link clock. Most of the core operates in link clock domain. This domain is based on the lnk_clk_p/n reference clock for the transceivers. The link rate switching is handled by a DRP state machine in the core PHY later. When the lanes are running at 2.7 Gb/s, lnk_clk operates at 135 MHz. When the lanes are running at 1.62 Gb/s, lnk_clk operates at 81 MHz. When the lanes are running at 5.4 Gb/s, lnk_clk operates at 270 MHz. When the lanes are running at 8.1 Gb/s, lnk_clk operates at 405 MHz.

In the DisplayPort Sink core, lnk_clk is derived from the recovered clock from the transceiver. When the cable is disconnected this clock becomes unstable.

Note: lnk_clk = link_rate/20, when GT-Data width is 16-bit.

rx_vid_clk

This is the primary user interface clock. Frequency of this clock is dependent on whether "frame buffer" is being used in the design or not as well as if the interface is AXI or Native Mode. If an inaccurate clock is used, the DisplayPort FIFO can overflow or underflow causing data corruption.

AXI Mode with a Frame Buffer

With a frame buffer, rx_vid_clk can run up to 300MHz and must be at least equal to the (pixel clock/Configured PPC).

This is the AMD implemented solution as it does not require any external clock management.

Native Mode or AXI Mode without a Frame Buffer

For non-frame buffer designs, the DisplayPort RX core requires the generation of a video stream using the M and N values within the Main Stream Attributes to reconstruct an accurate stream clock. The DisplayPort RX core places this information on dedicated signals and provides an update flag to signal a change in these values. The following figure shows how to use the M and N values from the core to generate a clock. For more details, see the VESA DisplayPort Standard (VESA website).

Recommended: The AMD MMCM is not accurate enough to be used to regenerate the necessary clock for non-frame buffer design. You need to use an external PLL that meets the requirements of the DisplayPort standard. For more details, see the VESA DisplayPort Standard (VESA website).

Figure 1. Receiver Clock Generation

s_axi_aclk

This is the processor domain. It has been tested to run as fast as 135 MHz. The AUX clock domain is derived from this domain, but requires no additional constraints. In an AMD UltraScale™ FPGA, s_axi_aclk clock is connected to a free-running clock input. gtwiz_reset_clk_freerun_in is required by the reset controller helper block to reset the transceiver primitives. A new GUI parameter is added for AXI_Frequency, when the DisplayPort IP is targeted to UltraScale FPGA.

m_aud_axis_aclk

This clock is used by the sink audio streaming interface. It must be = 512 × audio sample rate.

m_axis_aclk_stream<n>

This clock is used to output AXI4-Stream data from the core in the AXI4-Stream mode. This can be of constant frequency (greater than rx_vid_clk) and can run at a maximum frequency of 300 MHz.

For more information on clocking, see the Video PHY Controller LogiCORE IP Product Guide (PG230).