Hi there all on the forum.

I have an application where I would like to demodulate an IR signal (my own home brew RC5 - AKA Manchester Coded ASK).

While it is easy enough to modulate the data with a simple timer set at 2x the Carrier Frequency. it is more difficult in demodulating it.

In your opinion is this do-able?

For the readers reference
38kHz = 26uSecs

for a 50% duty cycle this gives a 13uSec Logic one then a 13uSec Logic zero.

I have the data coming in on a non interrupt pin on a LPC925, I was going to simply sit in a loop and wait for a state change on the incoming pin, to begin with not the flashest but that is all I have to work with at the moment.

My Code would work like the following:

When I have detected the change after sitting in my loop, I could start a timer (say 2uSec, is this too fast) and have a counter tick over wherby the state must not change until I have a counter greater than 5 (10uSecs) before I register it is a logic one this gives a wee bit of a window incase I start the count late.

Then I wait until I get 2 consequtive lows counts and begin a timer to allow for the logic zero, again looking for a transition at around 10uSec.

When I have the required number of transitions, in my case 50 (380bps seems like a good place to start), then I register this as the first complete 1/2 bit of my manchester code. I then wait 50x 13uSecs for silence, to complete the bit.

The point is all this seems a bit of a headache, has anyone done it before? is it worth it? are there any better approaches (not using the 3pin demod jobbers please) I already have the hardware.


Ay comments are appreciated. Hope that all made sense.

Regards Marshall