[time-nuts] HP-531xx calibrator nearing completion
csteinmetz at yandex.com
Fri Sep 8 18:15:22 EDT 2017
> Mark Sims wrote:
>> It takes a 10 MHz input, feeds it through a sine-to square converter
>> (using a biased CMOS gate)
> Could you send me a snip of what that looks like?
Depending on your application and its tolerance for jitter/PN, you may
want to use a fast analog comparator to do the initial sine-square
conversion. The difference in jitter/PN compared to a biased CMOS gate
can be 10-20dB over a wide range of tau/BW (even worse if the gate has a
Schmidt trigger input, which has WAY, WAY too much hysteresis for a
sine-square converter). I recommend the TLV3501, LT1719, and LT1720 out
of extensive experience with them. These comparators have just the
right amount of internal hysteresis for squaring a 10MHz sine wave, with
10MHz jitter around 10-15pSrms for jitter bandwidths from 10Hz-5MHz.
Of course, if your multiplier/divider chain has more jitter than a CMOS
squarer, it won't make much difference which squarer you use -- so be
careful with its design and construction.
More information about the time-nuts