This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LMH1983: Phase noise demand for 12G SDI

Part Number: LMH1983
Other Parts Discussed in Thread: LMK03328

Dear colleague,

Our customer has a concern about LMH1983 phase noise.

They viewed the below post:

https://e2e.ti.com/support/interface/f/138/t/528316?tisearch=e2e-sitesearch&keymatch=12G%252525252525252520SDI

https://e2e.ti.com/support/interface/f/138/t/744331?tisearch=e2e-sitesearch&keymatch=12G%252520SDI

These two posts mentioned that the phase noise of the 148.5M and 148.35M clocks output by LMH1983 cannot meet the requirements of XILINX Kintex ultrascale FPGA.

If the 148.5M and 148.35M clocks output by LMH1983 are directly used as the reference clock input of Kintex ultrascale SERDES, the jitter of the 12G SDI signal output by the FPGA may not reach the SMPTE standard. Their current design is also like this, and the jitter of the tested 12G SDI signal does exceed the standard (Alignment jitter is around 0.5UI).

The post also mentioned that LMK03328 or other low-noise phase-locked loop chips can be used to connect the 27M clock input of LMH1983, and the low jitter 148.5M and 148.35M clock output by LMK03328 can replace the 148.5M and 148.35M clock output by LMH1983 as the reference for FPGA clock input.

So,

1. For the design of 12G SDI, is the jitter of the 148.5M and 148.35M clock output by LMH1983 sufficient?

2. Does they need to add a low-noise phase-locked loop chip (such as LMK03328) to meet the SMPTE standard?

Best Regards,

Rock 

  • Hello Rock,

    1. It doesn't look like LMH1983 148.5M or 148.35M will meet the requirements for 12G SDI, hence why our previous explanations of compatible solutions suggest the use of LMK03328. LMH1983 was originally designed to meet 3G SDI requirements.
    2. Some additional PLL or oscillator which can generate the required frequencies would be required, such as LMK03328. LMK03328 is a convenient solution because it can utilize the 27MHz from LMH1983 to generate exactly-accurate 148.5M and 148.5M/1.001 clocks which are synchronous to HSYNC, but alternative solutions could exist; as an example, one customer used LMK03328 to generate 2x 148.5M and 2x 148.5M/1.001 for TX block, and RX block uses VCXOs: see Intel AN-768.

    Regards,