> >I am wondering, have anyone tested max speed for a polled program
> >decoding an rotary encoder ? Would an interrupt driven program
> >do better ?
> >
> >PIC16F877 @ 20 MHz. Approx 50 instructions in the active loop
> >Encoder is 100 pulses/rotation, quadrature decoding (400 flanks/rotation)
> >I have reached around 2000-2500 rpm. 3000 rpm gives overspeed error.

> Let's see, 3000 rpm is 60 rps, with 400 counts per revolution, that is
> about 24k counts per second.  With some code tweaks, I think you could

3000 rpm/60 = 50 rps, right ?
50 rps * 400 counts = 20k counts/s

> easily be doing more than 5x better than that as I got at least that many
> counts per second out of a 4MHz part before tweaking my code a bit more and
> using it on a 20MHz part - but I didn't test its performance after
> that.   I think polled mode has a _slight_ edge if you aren't doing much
> else, I do recommend doing interrupts on encoders if accuracy is important,
> just make sure that all your other interrupts are kept to a minimum.

I am not doing anything besides polling the encoder to find the max speed it
will handle so I can have 2-3X safety marginal when designing the mechanics.
Its supposed to handle 4 encoders and rs232/485 later on.

Hmm, 5x faster eh ? Then I wonder... might it be possible that the encoder
itself gives up at <3000 rpm ? Unfortunately I dont have the datasheet for it.

> Here are several examples you can look over to compare with your code:
> http://www.piclist.com/techref/microchip/qenc.htm

I got the code idea from there; first program on that page. I've been on the
piclist so many years I know where to start looking ;-)
I really like the structured layout of that asm, just needed a little rewrite to compile

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics