My new house has an Itron C1SR electric meter, which sports an IR LED that flashes once for every watt-hour of power usage. I picked up some IR photodiodes from Mouser this week and used my oscilloscope to verify that the meter's doing what I expected. It is, mostly, but in addition to a 10ms-wide "usage" pulse, there's also an underlying carrier signal containing a high-frequency ~2ms-long burst that repeats. Here's a screen-shot from my 'scope:
Is it possible that there's some additional data embedded in that signal, or is it just noise?
It's more likely a 5kHz signal with a repetition frequency of 250Hz.
Check your power provider if there is some daytime/nighttime price.
Maybe the power meter can select different "price" for day/night using this 5kHz signal.
Interesting. This is sort of a loaded question and going to showcase my ignorance, but what's the best way to use an Arduino to log this signal? The parameters seem to be the length of the 5kHz section, the frequency at which it occurs, the actual frequency of the signal (assuming it's constant for a given occurrence), and perhaps its amplitude? It seems like I could probably put a small capacitor in-line with the signal to remove the DC component, use an op-amp to reduce the in-between sections to zero, look for the rising edge of the 5kHz section, and do some analysis there?
What is the amplitude of your rectangular pulse?
If it's within 0-5V (I cannot see well on picture) you can think of some FFT analysis directly with Arduino Then you can look at any frequency component in 0-5kHz range.
Needs less hardware, but more software.
Perhaps you don't want at first step thorough analysis. Then develop some 5kHz filters and Arduino will do the counting, timing, and so on.
Then you can look at any frequency component in 0-5kHz range.