[time-nuts] Sub-ps delay line
davidwhess at gmail.com
Tue Feb 7 17:12:32 EST 2017
I did something similar a couple years ago to make an adjustable 75
nanosecond pretrigger for my sampling oscilloscope so I will just pass
along some things I learned.
Power supply noise will create jitter in single ended logic because of
lack of power supply rejection. Temperature will be a problem with
single ended logic also although I did not care about that. The delay
section certainly needs to be differential which the LTC6957-1 neatly
covers; I was planning to use a comparator or line receiver next time
but the LTC6957-1 looks ideal.
With the above in mind, I would use a separate reference quality
regulator for the supply voltages and any reference levels. I only
needed 10s of picosecond stability and picosecond level jitter but I
suspect at the level of precision you desire, ground loops need to be
avoided even through a solid ground plane.
How were you planning on testing the performance?
I need to give this further thought but if you only need 1ns worth of
control and feed one input of the LTC6957-1 with a 100 MHz sine wave,
wouldn't adjusting the level of the other input work to adjust the
delay at the output?
More information about the time-nuts