lordpil's blog


it might not be very linear with a low load
try using like 1K with maybe .01uF after it
or 100R, or try doing a lpf with a small series resistance
(like, i would just fuck with it to see what it does with other loads)

Posted by renesis at 12:23 | permalink | 0 comments


bullshit
not in a short average and not if youre syncing with the wave freq
yeah but ive never agreed with nyquist for real work work
you sample twice per cycle, you can hit zero crossings every time
2 samples is pretty hopeless
you can spline shit like that

Posted by renesis at 12:14 | permalink | 0 comments


i would try for at least 10
really the most you can manage

Posted by renesis at 12:09 | permalink | 0 comments


k
pretty sure the hard part is figuring out what to clock the adc so you get enough samples but still get an easy to divide number

Posted by renesis at 12:04 | permalink | 0 comments


hes taking 16 so he can maybe fuck it up like 5%
16 cycles
i dunno i dont see why you cant sample 16 periods worth
and average
the adding and dividing prob takes like 10 proc cycles
why not
take 256 samples and ignore the low byte
per 16, do 1024 and ignore low byte then right shift twice, whatever works

Posted by renesis at 11:22 | permalink | 0 comments


_psi_: use a sampling period of 1/50Hz * 16
if you get the period right, your first and last waves should take care of any weird averaging issues

Posted by renesis at 11:04 | permalink | 0 comments


as apposed to a partial period, i guess
and im pretty sure after a second, the error associated with the incomplete first cycle is either way lower than the precision of the DMM
or its discarded as the end of a rolling average

Posted by renesis at 10:41 | permalink | 0 comments


Top | Add to Technorati Favorites


© 2007 lordpil.   XHTML 1.0! CSS! Site design by GNAA  Blog Engine by pbx | MULTI2 | ian hanschen | lolwat