Hi Dan, I would use a pair of dividers to get them to the common frequency of 5 MHz, so divide by 2 and 25, then use a SR flip-flop and then feed it to an op-amp doing a PI-loop. Be sure to make it well damped, so a damping factor of 3 or more. The bandwidth of the loop should be relatively high. I've brute-forced it worse than this and it's been good enough for gigabit links beyond what you are making. The needed components can be taken from fairly standard series, and then nothing special about the components and sizes and you will be just fine.
As long as one has sufficiently good jitter from the 125 MHz VCXO you should be able to do it. For systems like these I assume that the last step has neglible jitter suppression, but it is relatively easy to find reference oscillators with sufficiently low jitter. Assuming we talk gigabit-link, then for bit error rate below 1E-12, we need the jitter to be 1/14 of the symbol period, which in this case is 320 ps, so 23 ps RMS, and well that is not so hard to meet on the reference these days. Also, the last synthesis step will improve jitter anyway. So, I do not think you need to do anything fancy to achieve the goal as far as I have understood your requirements. What I sketch above is fairly straight-forward to try out. It takes a small handful of COTS chips, but nothing that really eats board-space these days or will be hard to source. Cheers, Magnus On 2020-01-02 15:26, Dan Kemppainen wrote: > Hi All, > > Just to clarify, the PLL we're looking for only needs to do the 10 -> > 125MHz. The 125 ->MHz 3.125GHz is in a separate device with it's own > PLL. I have some, but not a lot of control over that. The goal here is > to provide a good source for the 125Mhz, for not a lot of board space > and not a lot of BOM cost. > > Again, I apologize as this is a bit vague. Basically we're trying to > provide a "as good as reasonably possible" 125Mhz source from a VCXO, > in a relatively small (1 to 2 sq inches) board space, for not a lot of > cost ($25 to $50 range). > > Anyway, There have been some good suggestions here, and a few off line > also. > > Thanks! > Dan > > > > > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to > http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com > and follow the instructions there. _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com and follow the instructions there.
