Hallo all,

Charles, thank you for drawing the schematics. Only, pin 5 of CD4046 should additionally connected to ground.

PLL circuit I proposed, was designed for frequency range 0.5Hz to 99Hz. It was able to monitor mains frequency of 50.00Hz with a resolution of 0.01Hz. Of course, PLL can always be improved for achieving better settling time or ripple or even both. It was suggested only as a starting basis.
Circuit demonstrates how elegantly PLL circuit can be designed by using this wonderfull CD4046. Only 2 digital chips, 4 resistors and 2 capacitors are necessary to build a PLL circuit which does wonderfully working for the most applications.

Sorry, Peter, but I cannot see 'tons' of additional electronics. By the way, you would wonder, how often PLL circuits are used in electronics, where a pure mcu solution would heavily fail...

As Michael already stated, measuring RPM of a big machine means, that rotation speed is changing only very slowly. And sometimes I wonder that people try to get information from a physical process which is actually not there: A slowly reacting machine cannot have a fast change of rotation speed. The only fast changing data you measure is some jitter, caused by noise related shifting of threshold of sensor.

If you want to regulate a machine in order to have a constant rotation speed, then why are you willing to measure fast changes of rotation speed? What is this information good for? For regulating, or even displaying? What will you see on a fast changing display? So, if you measure time interval of one period only, there's not enough information at all, if you have some jitter.
Then you MUST average over some periods. Physical process does not allow you to define rotation speed of measuring length of only ONE period, if there is some jitter.

This necessary averaging does the PLL loop filter for you, automatically.

Bye,
Kai