Circuit Design Suggestion for variable voltage input to set voltage squarewave output

I have an input signal which can be any of the following (maybe others in the future too):

a true +/- sine wave of +400mV to -400mV
a true +/- sine wave of +6V to -6V
a square wave signal between 0V and anything between 3V and 15V

There are little variations in those as well (like rectified positive humps instead of a true square wave for instance) but those are the main input signals.

I then need to convert this signal to a 0V to VDD (3.3v or 5v) signal with a square wave. Duty cycle is not very important, the frequency is what is important. Phase shifting is OK too. Basically counting edges.

So I am entertaining ideas on how to make a generic input that can take these small signals below a diode drop and these big signals above VDD and magically make them digital lol.

I was thinking something like a zener diode for the higher voltages could be sent into a comparator with the other side of the comparator at VDD and ground. Then for the small signals an amplifier with some gain then through another comparator. Using the zener turn on I could maybe switch a FET to disconnect the small signal input to the comparator.

I wondering if there is an already general solution to these types of problems so I am not reinventing the wheel. All suggestions are welcome, Thanks.

Seems as though you could use an AC amplifier that has very high gain, with the output consisting of the positive waveform only.

The schematic below is all-purpose.



A small amplitude AC signal is unchanged by the capacitor and diodes.

A large amplitude DC pulse train is first turned into AC by sending it through a capacitor. Then the diodes clip it to between 0.6V and -0.6V. (Something will need to be done about the negative portion if the input must not be fed a negative voltage.)

The opamp/comparator amplifies only one polarity of the waveform. You can choose whichever way you want to use the inverting and non-inverting inputs.

Thankyou for the insight. I have sim'd a circuit in LTSpice and it does appear to work. The excess current to make the diode drop is burned through the diodes, so sizing the R appropriately is a must.

I made a 1uF cap and a 1Kohm R. I then fed it through a LTC6252 with the negative set around 300mV and the positive input to the output of the diode structure. This bangs the opamp around rail to rail and I can then use that through an isolation barrier and all is well.

I am going to prototype something in the real world pretty soon.

Any advice on sizing the R and C? It would appear that with a 12v signal I am putting about 11mA into the diodes which isn't too bad.

1k ohm is a common value to use in small signal conditioning, etc.
It's not too big, not too small.

The capacitor should be high enough uF, in order to pass at least some of the lowest anticipated frequency.