IR sensor is TSOP38233-ND. It is a 33 KHz sensor. While HP protocol specifies 32768 Hz, this turned out to be close enough — oscilloscope shows something that looks like a valid frame(s) when I send printing commands from my infrared modded WP 34s.
The printer is CSN-A2 thermal printer with TTL serial interface. It is not exactly compatible with HP's prinitng unit but I hope it is close enough to emulate. I've got my from Adafruit. They sell it as a kit with a 5V 2A power supply and a connector. I was having whole lot of weird unpredictable behavior until I realized that printing in graphic mode (bitmaps) and printing block characters drains too much power for this power supply. Oscilloscope showed that voltage drops as low as 2 volts, which sends printer in reset. To be fair, the kit power supply works ok with normal text printing. Anyway, I've searched my bin of power supplies that need home and found a 9V 4A one which works much better. I still have some weird behavior when trying to print bitmaps, but Adafruit sample programs (QR code, logo, etc) print normally, so, I assume, I just need some more figuring out to do.
The controller for now is RPi1B. I installed lirc and only use hardware driver (lirc_rpi) to get pulse/space timits which I process myself with a perl script. Maybe there is a way to teach lirc to do the parsing, but I have not figured it out right away, so went with a perl script. This is a good thing, since I will not be able to use lirc in the decoding layer of the final product. RPi is ok for prototyping, but it takes 60-90 seconds to boot up and I would prefer faster availablity. So, I will not be able to run linux and, by extension, lirc.