Smajdalf:
I measured voltage when no cap was connected: just 20Mohm from IN+ to GND. I got 0.7V at output, so I concluded there is 0.7V over the resistor. 0.7V/20M=35nA.
Op-amp input leakage current.
I suggest making the test resistors smaller (like maybe 100k) then add gain to the op-amp to read leakage.
With a pair of 40 meg resistors in series, simply humidity, breath or fingerprints can seriously change the resistance value.
A rule of thumb I use is that if R is greater than 470K, I'm doing it wrong.
Smajdalf:
What?
I will send a pulse each 100ms while the button is pressed. I expect 100 pulses will be enough for Roomba to register it, maybe less.
You DO realize that IR remote controlled devices respond to particular CODES which are typically made up of a 40 khz "carrier" (IR led blinking on and off at 40 khz) with variable ON and OFF times to represent binary 0 and 1 (don't you?).
These codes typically also contain checksum data, detector AGC stabilization pulses (at the beginning) and other stuff.
You cannot simply "blast 100 pulses" at anything and expect it to work.
Realize that the blue and yellow vertical "bars" are actually "x" number of 50 % duty cycle IR pulses at 40 khz.
At first I would like to thank Jiggy-Ninja for the energy-saving text. There is hint to disable pull-ups on buttons when not needed. At first I thought it is pointless - just <100uA of current wasted. Nothing to worry about from long term point of view. But this current consumption adds to total consumption and if power source has high (internal) resistance it matters a bit: 0.1 V voltage drop "for nothing" with 1K battery internal resistance is bad.
I have 2 screenshots from my tests. It is from my own "Arduino oscilloscope". My remote is powered from 2.5V 1k "internal resistance" voltage source. Vcc, not GND is connected between remote and Arduino so "voltage drop" is in fact rise of measured voltage. Without cap I got this:
Voltage "drops" from 2.5 V to 3.3V, so only 1.7V over remote when sending signal. Only 0.8 mA got through the LED in this setup. With 220uF cap it was much better:
