Hello. I need to capture a logical input signal that is at 0 at rest and that goes to 1, then to 0, with a defined timing. That is 300ms to 1, followed by 200ms to 0. This capture must be done in Hardware. We will then generate an interrupt at the output (pulse of a few ms for example). This is to wake up a microcontroller. Important: the assembly must be of very low consumption (1-2uA max.). No need for great precision for the 300ms and 200ms (10% ok). Thank you!

 

A lot of detail missing, but you might be able to get away with just an edge detector made with a capacitor, resistor, and diode.

 

Why not just send the signal the micro?

(I must be missing something)

 

What other sequence of pulses could possibly be happening on the same line? So that this device has to be able to tell them apart?

 

Why not just send the signal the micro?

(I must be missing something)

That was my first thought, but I don't know what micro he's using or how it behaves if it receives a long pulse to wake it up. He should look into what does happen -- it whatever happens under those conditions is acceptable, then problem solved.

 

OK. First, thanks for your comments. I wanted to be as brief as possible, but I give you here some more details... We have a RF sensor which wakes up a micro-controller in sleep mode too often, because of RF noise. As soon a glitch is caught, the uC is consuming for a while... That involves too much consuming when the decoding is made by the uC several times per days... This RF sensor must activate the uC only when a certain "pattern" is coming from another device (called B). This device B sends several pulses of about 300ms 'ON' - 200ms 'OFF' -... Then, the idea is to wake-up the uC ONLY when receiving the 'pattern' from the device B.

 

Could it be simplified to waking up the micro every time that there is a pulse longer than 300ms?
If power consumption is a problem, then a complicated wake-up circuit isn’t going to help.

 

@Ian0. Yes, it could be a more quick and simple solution if it is too difficult. With a RC cell, a slow goes up set to about 280ms and a quick goes down, followed by a comparator for example.

 

If there are a few odd occasions when the 300ms criterion is satisfied and the 200ms isn‘t the micro will wake up unnecessarily, and it can then go straight back to sleep, but it will probably be more efficient overall than having the extra circuitry to do the 200ms check.

 

A retriggerable oneshot that inhibits the trigger on low of the 200 ms oneshot. and then the 200 ms OS inhibits the trigger of the FF that otherwise gets the pulse to wake the computer, So the sequence is actively inhibiting triggers until after the sequence complete. The only drain is the CMOS IC idle current, not much.