Below are a pair of simultaneous measurements, conducted to evaluate agreement between a PICTIC II and an HP 5370B counter at two different interpolator gain settings.
The START inputs of both counters were driven by an HP 5065A Rb standard divided down to 1 Hz (for the 5370B) and 10 Hz (for the PICTIC II) with a TADD-2 divider. The STOP inputs were driven by the 10 MHz output from a stock Trimble Thunderbolt GPS clock via a Mini-Circuits splitter.
The 10 MHz
signal at the PICTIC II was squared for CMOS compatibility with a differential pair of 2N3906s, modified
slightly to optimize its output range for 0-5V.
Above, the blue and green traces were taken over several hours with the PICTIC II configured for IGain=400 (@@Cg190). This results in a theoretical resolution of 250 ps. The magenta and red pair were taken in a separate run with the PICTIC II configured for IGain=800 (@@Cg320), resulting in a theoretical resolution of 125 ps.
At least with the stock 10 MHz clock, there appears to be little difference between these settings. Going by the ADEV figure at t=1s, the PICTIC's observed resolution is about 680 ps, maybe a bit lower at IGain=800, while the HP 5370B's observed resolution is close to its specified limit at about 22 ps.
Considering that most counters are lucky to deliver resolution in the microsecond range, this represents excellent performance on the part of the $50 PICTIC II. It will be interesting to see how well the board can perform when its onboard clock is upgraded to a faster, more stable source.