How to filter PulseIn readings

[vmx10]

I am looking for a way to filter the signal from the pulseIn function. I want to get rid of both spikes on the falling and rising edge.

I tried using two filters

Exponentially-Weighted Moving Average (EWMA) -> failed to remove the second spike (falling edge)
Debounce -> the value is not updated until a constant speed is reached. In my case it can take up to 30 seconds. Therefore, it is not applicable in my situation.

Board: Atmega2560 (Arduino Mega)
uint32_t pulse = pulseIn(13, LOW, 60000);
Signal that I measure -> from 20Hz (50% duty) to 1100Hz(50% duty)

[jremington]

The pulseIn function just returns a number. How does that "spike"?

Or, are you referring to electrical spikes you see on the input using an oscilloscope?

[mancera1979]

Have you tried an RC filter?

[vmx10]

The problem is not in the electronics. The circuit works just fine. I feed the signal from the PWM generator. On the oscilloscope I can see the correct signal. I take the measurement directly on the microcontroller leg. Not even the slightest ripple.

PulseIn, instead of returning a value of '0' immediately,
returns slowly increasing values

Instead of: 3000 > 0
I have: 3000 > 10000 > 40000 > 0

Of course, the values I've given are only illustrative (different each time)

[mancera1979]

Can you post your sketch?

[jremington]

Sorry, that output makes no sense to me.

Please post a wiring diagram, the code, using code tags, and examples of the actual output.

What is the "PWM generator"? What is the actual waveform that it produces?

[Paul_KD7HB]

Seems the OP is confusing the A/D functions with the Pulse Length function.

[jremington]

I suspect a wiring problem -- no ground connection, or a floating input.

[ninja2]

Strong chance it's just noise on the leg (hard to see with a CRO or DMM) or line capacitance . Just add a small capacitor from the leg to GND, or R+C if needed..