Hi Jim,
There are a couple of things you could do. If you lowpass after the
deadband code, then you'll get a smooth transition and it won't bump
quite so much.

Another thing you can do is left-justify your converter results so it is
16 bits with 6 dead LSB bits, add a small amplitude triangle wave to
it(say 6 or 7 bits), lowpass it, subtract half of the max triangle
value, then shift right down to 11 bits, feed that to the deadband code.
I may not have that quite right, but it's like something I did for a
temperature readout once and it worked well. I forget what period I used
for the triangle wave but I think one cycle was about every 32 samples
of the A/D.

Best regards, Bob

On Thu, Nov 26, 2015, at 11:10 PM, Jim Ruxton wrote:
> Thanks Bob,
> This is actually what I have now however it reduces the resolution of my=
=20
> system from 1024 values  to 1024/(deadband+1) . I was hoping to avoid=20
> the drop in resolution. I am building a light dimmer so I want as=20
> gradual a curve as possible when I am moving the POT to control the
> output.
> Jim
> > #define DEADBAND 1
> >
> > if ((oldval + DEADBAND) > newval)
> >      updateVal(newval);
> > if ((oldval - DEADBAND) < newval)
> >      updateVal(newval;
> >
> > Don't worry about rollover/under, as long as all values are unsigned.
> >
> > Still dithering? Increase DEADBAND to 2.
> >
> > Cheers, Bob
> >
> > On Thu, Nov 26, 2015, at 08:28 PM, Jim Ruxton wrote:
> >> Thanks everyone for the suggestions. I don't think in my case it is
> >> really a filtering issue because for example if my POT is set close or
> >> on the edge of an A/D transition I believe I will still see a bit
> >> flipping occasionally in the LSB if I have lots of filtering. Not sure
> >> if I am being clear but I really only want to have the reading change =
my
> >> output if I am moving the POT so thought there would be a clever way t=
o
> >> detect that and only update my output if the input is varying
> >> substantially.  Perhaps to really get the best results I need a
> >> multiturn POT or move to a digital encoder.
> >> Thanks,
> >> Jim
>=20

--=20
http://www.fastmail.com - A fast, anti-spam email service.

--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.