Cars already have "digital" pulses where the frequency of the pulses is proportional to engine rotation speed. Cars with ECUs have more than older cars.
For example, my mid 90's Toyota has a crank angle sensor (one pulse per crank revolution), 4 electric fuel injectors (one pulse every other rev), spark coil pulse (two pulses per rev), speedo pick-up from drive line (? pulses proportional to vehicle speed), and others I haven't thought about. The car has an analog meter tach, but I have not looked at what drives it.
If I wanted a digital tach, I would use a PIC to measure the period between one of the pulses mentioned above. I would run the PIC off ~1Mhz resonator or crystal. I would use the Timer hardware in the PIC to count 1MHz clocks between successive engine pulses, effectively capturing the period each revolution of the engine. To convert period to revs/min, it takes one multiplication and one division and then a binary two's complement to decimal conversion in the PIC software before outputing the RPM value to a suitable display.
This method is the one used in high-end digital tachs,
like this one I have in my airplane. The advantage of using period measurement vs counting rev per unit time measurement is that revs per min takes a full minute to update. If you shorten the counting period, you give up numerical resolution. The period measurement method gives full four or five digit resolution at a very fast update rate.
btw-older engines with point-ignition systems use the waveform across the points in the mechanical distributor as a pulse train to operate the analog tachometer. This waveform has a lot of ringing and swings from ~+150V to ~-250V. This is clipped and filtered to produce a "logic level signal" for processing inside the tach...