Hmm, the capture data is weird. I see pulses with length of 2us to 6us. That's way too short when you consider the chip can output a maximum of 16khz. Or should I interpret these pulses as edges? In that case it will match emulation with 4khz R1 mask.