Hallo Peter,

Also a PLL need still a very clean VCC to work stable. E.g. I faced often problems of jitter also in the case, that the input signal was very stable (derived from a crystal).

You are right, this is a well known problem, but only when dealing with highest precision applications. It's caused by ground bounce. As long as VCO and phase comparator are sitting in the SAME package, there will always be a certain coupling. Unfortunately power supply decoupling will not help. I have seen some 'ultimative' power supply decoupling 'solutions', but they were not able to remove this coupling. But there's a very easy methode to remove it to a much greater degree: Put VCO and phase comparator in different packages!

And for a lower range of 100RPM you need really a very slow reacting PLL. Means, you would see a delay of many seconds from changing the speed until the new value was displayed.

I agree, very low frequenciesin combination with abrupt changes is always problematically for PLL designs. But there is also a rather simple cure: It's possible to insert an additional, faster time constant, only being active for abrupt changes of frequency. You know, that one which uses a resistor and 2 antiparallel diodes, connected across 1M resistor of my example, which Charles kindly made a drawing of.

Also to get the right feeling you should never display a new value shorter than after 200..500msec.

I would have choosen a counting interval of 1 sec.

So you have all time of the world to do some calculations

Ok, you measure the time of one period. I assume that you average signal before displaying? How do you do it?

So I see only advantes on avoiding the PLL, but no disadvantages.

It's your good right. Of course, I disagree to some amount. I think hybrid solutions, outsourcing some signal conditioning tasks can be of great help in some applications. And one should be open minded to such 'old fashioned' solutions. Sometimes they are really advantegous. I think of many audio applications...
But if you absolutely don't need this damned PLL for solving this problem, and your solution is much better, then throw PLL circuit away...

Bye,
Kai