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.

Clock buffer with crosspoint switch, LVDS and LVCMOS outputs

Other Parts Discussed in Thread: CDCE62005

Hello.

I have a complex clock distribution issue with rather unusual requirements. This is part of a timing instrument.

I have 2 LVDS clocks input (A and B). Clock frequency is 10 MHz.

I want to distribute them to:

  • 1 LVCMOS output for clock A, always on.
  • 1 LVCMOS output for clock B, always on.
  • 2 of any differential outputs, clock A or B, can be disabled.
  • 1 LVCMOS outputs, clock A or B, can be disabled.

The differential outputs could be combinations of A and B: both A, both B, or A and B (B and A is optional). This is the issue with almost all clock buffers, because they only include mux and not cross-point switches.

There is other requirements:

  • The jitter on the differential outputs is critical, it must stay under 5ps.
  • The propagation delay must depend linearly of the temperature, or stay within 200ps for -40 to 80 deg C.
  • Only 3.3V is available, power consumption should be minimized, 100mA max when differential outputs are off. There is another 3.3V supply that is disabled when the differential outputs are not needed.
  • Cost is not an issue.

My current solution is to use two LMK00304, one for each clock, and put a 2:2 cross-point switch after, that is powered down when not needed. Two LVDS to LVCMOS translators, that are also powered down when not needed, give me the last LVCMOS outputs.

This is 5 components total. It look overcomplicated, and I am wondering  if I missed something, like a single chip that will do the job.


Thanks!

  • If you can do SPI programming to configure the clock device, you could use CDCE62005 as a clock distributor.  It has two inputs and 5 outputs supporting independently selectable output formats (LVPECL, LVDS, or LVCMOS).  The 5 output buffers can independently select either input clock as a source.

    Regards,

    Alan

  • Thanks, it seems to suit all my needs!

    It consumes a little more than the limit but nothing dramatic.

    The propagation delay is around 3.1 ns for LVDS, this could be a problem if it changes a with temperature. Do you have any data for propagation delay vs temperature or propagation delay vs vcc? Is it linear?

    Paul.