GS2971 Datasheet, PDF (86/148 Pages) Semtech Corporation – 3Gb/s, HD, SD SDI Receiver, with Integrated Adaptive Cable Equalizer complete with SMPTE Audio and Video Processing

When the IGNORE_PHASE bit in the host interface is set HIGH, it is recommended that
the M value be programmed via the host interface. This can be done by setting the
FORCE_M bit HIGH, and programming the desired value into FORCE_MEQ1001. The
correct value can be obtained by reading the M bit from the Video Core Registers.
If the DDPS is locked to phase data and audio data packets are lost or corrupted, the
Clock Generator will flywheel for up to four audio data packets. If no valid audio data
packet with valid phase data is provided within this time, the Clock Generator will
free-run based on the video format, the PCLK and the M value.
If the IGNORE_PHASE bit in the host interface is HIGH, the clock will free-run based on
the video format, the PCLK and the M value, independent of the NO_PHASEA_DATA bit.
In the 720p/24 video format, the total line length is 4125 pixels, which requires a
resolution of 13 bits for the audio clock phase words in the embedded audio data
packets. SMPTE 299M only specifies a maximum of 12 bits resolution. Proposed changes
to SMPTE 299M suggest using bit 5 of UDW1 (currently reserved and set to zero) in the
audio data packet as the MSB (ck13) for the audio clock phase data, providing 13 bits
resolution.
Some audio encoders may hold the clock phase value at a maximum value when
reached, until reset at the end of the line. This produces a small amount of audio phase
jitter for the period of one sample.
To overcome this issue, the audio de-embedder checks for all cases. On detection of the
maximum value, a comparison is made between previous clock phases and the correct
position interpolated. If the clock phase data value starts to decrease, the de-embedder
checks to see if bit 5 (ck13) of UDW1 in the audio data packet is set. If ck13 is set, the
correct value is used. If ck13 is not set, the correct position is interpolated.
4.19.3.2 Detect 5-Frame Sequence Block
5-frame sequence detection is required for 525-line based video formats only. The audio
de-embedder checks the Audio Frame Number sequence in the audio control packets,
when present. If the audio frame sequence is running (repeated 1 to 5 count), the audio
de-embedder uses this information to determine the 5-frame sequence. If the audio
control packet is not present, or the Audio Frame Number words are set to 200h, the
audio de-embedder detects the 5-frame sequence by counting the number of samples
per frame. Figure 4-43 shows the number of samples per frame over a 5-frame
sequence.
Frame 5
Frame 1
Frame 2
8008 Samples
Frame 3
Frame 4
Frame 5
Frame 1
1602
1602
1601
1602
1601
1602
1602
1602 Repeat
Figure 4-43: Sample Distribution over 5 Video Frames (525-line Systems)
When the audio inputs are asynchronously switched or disrupted, the audio
de-embedder continues to write audio samples into the audio buffer, based on the
GS2971 3Gb/s, HD, SD SDI Receiver
Data Sheet
45898 - 2
September 2012
86 of 148