How to DEFINE OSC, in case of 32khz clock frequency?

[mark_s]

Agree with using sleep and IOC. Years ago I had a similar problem. I contacted Charles at Melabs
he gave me a solution where you set osccon to 500Khz which reduces the current to a couple hundred microamps (vdd3.3v). Then trick PBP by multipling your baud rates and dividing your pause statements. DEFINE OSC 4mhz, set osccon to 500khz. The pic will be running 8 times slower. So if you wanted a PAUSE 1 you would make it PAUSEUS 125. Serial out at 2400 baud would be written at 19200. For shorter pause statements you could use nops as suggested above. This combined with sleep you can acheive very low current consumption on average and still use PBP. Also this chip has clock switching between 31khz and high frequency, which you might want to study.

[CuriousOne]

The main problem is that PAUSEUS is not working properly. It won't give you short enough pulses as needed. I've opened thread on that matter before, with precise measurements given.

[richard]

PAUSEUS is not working properly

NOT REALLY

the smallest possible delay from pauseus @32khz is about 3mS . the smallest asm delay would be 125uS ie NOP

[AvionicsMaster1]

Are you using 12F1840 or the 12LF1840? Ifn I read the datasheet correctly you can achieve some significant power savings with the LF that you may not have to do the hokey pokey with the oscillator frequency. Though the LF is about 10% more cost if it gets you to your goal it may be worth it.

Is this a battery application or an experiment? Just curious why a few nA would be important.