When you say auto tachometer do you mean a commercial device or the circuit you show.
Auto tachometers have tended to a bit of a nightmare as the ignition waveform is a complete mess. Most of the better commercial ones have solved the problem for car ignition waveforms but they may well not be satisfactory on a windmill.
The 2917 is a very useful chip and will certainly do your job but you will need to be careful to prevent the input stage multiple triggering. The input stage is fast and is a Schmidt with a small amount of backlash. The input signal needed is very small and any harmonic that crosses zero will trigger the input.
You don't say how you are deriving your input signal and I have no idea of your wiring arrangements so it is difficult to help.
The maximum signal input is claimed to be 48v pp but I wouldn't risk that much. It can be done with a small input transformer fed from a pair of ac lines that gives a few volts ac. You need a decent filter with the capacitor close to pin 1 and ground. You can use something like .22uF and you should have plenty of input volts to use a good big series resistor, try something like 47k. Even then you may get false triggering from high frequency noise from inverters or charge controllers. You need to be very careful about how you supply the ground to your 2917 circuit board. Not so bad if you have an analogue meter but if the tacho output goes to other circuits then there is much more chance of spikes coming in this route.
On one machine with a boost converter in close proximity to the tacho circuit I eventually had to resort to an opto isolator. I connected another diode in inverse parallel with the led and fed it through a series resistor from the ac lines. The photo transistor had a resistor in the collector and the collector was capacitively coupled to the 2917 input pin, not forgetting a return to ground resistor on the input pin.
I always use the 8 pin 2917, the more external pins you have the more likely you are to collect spikes from bad board layout.
If you use a small transformer for the trigger signal and you still have trouble with the suggested filter you could make it a 2 pole filter but it shouldn't be needed. feeding the secondary leads of the transformer a couple of times through a ferrite ring may remove common mode spikes. Grounding the core of the transformer to your 2917 board ground may help and if you have a luxury transformer with Faraday screen between windings, then ground that to the tacho ground.
Don't keep the input circuit of your Fig 14, that is not the way to go with a windmill and to be perfectly honest I doubt if it is the way to go for ignition either, but no doubt it sold a few 2917s for National semiconductor.
Flux