I've had good experience using the following method (probably only
fast enough for a human-turned encoder rather than a motor-driven
one):

First, you have a timer, let's call it TimerA. TimerA should be
decremented once per millisecond inside a timer ISR. If it reaches
zero, the ISR will not decrement it further.

In your main loop, you keep three registers: last_measurement, state,
and last_state.

Let encoder_measurement be the A and B channels together as a two bit numbe=
r.

The rest I will explain in pseudocode:

// Init code
// Note: make sure that TimerA is set as a volatile variable so that
both the ISR and the main loop can access it.

state=3Draw_inputs()
last_state=3Dstate
last_measurement=3Dstate

// Main loop starts here

main:

last_state=3Dstate

encoder_measurement=3Draw_inputs()

disable_interrupts // to prevent accessing TimerA at the same time as the I=
SR

if last_measurement !=3D encoder_measurement then TimerA=3D20
else if TimerA =3D=3D 0 then state=3Dlast_measurement

enable_interrupts

if state !=3D last_state then // Change happened
{
   if last_state.A=3D=3D0 and state.A=3D=3D1 then
   {
      if state.B=3D=3D0 then DoClockwise()
      else DoCounterClockwise()
  } else if last_state.A=3D=3D1 and state.A=3D=3D0 then
  {
      if state.B=3D=3D0 then DoCounterClockwise()
      else DoClockwise()
  } else if last_state.B=3D=3D0 and state.B=3D=3D1 then
  {
      if state.A=3D=3D0 then DoCounterClockwise()
      else DoClockwise()
  } else
  {
      if state.A=3D=3D0 then DoClockwise()
      else DoCounterClockwise()
  }
}

last_measurement=3Dencoder_measurement

goto main


This waits 20 milliseconds from the last change before doing anything.
It then uses simple comparison of previous and current A and B states
to determine what change just happened. I'm sure that the if
statements can be greatly simplified by using logical bit
manipulations but I did it in a very easily readable format.

Sean


On Sat, Sep 5, 2009 at 9:17 PM, Bob Axtell<engineer@cotse.net> wrote:
> Actually, interrupt on change is just about the only way to make quad
> encoding work. Set the PIC up to capture a change on both the A and
> the B pins. When you get into the routine, immediately read repeatedly
> until you get 3-5 SAME results on BOTH A and B. The XOR with old
> state, etc as is needed. I think you will see things clear up.
>
> --Bob
>
> On Sat, Sep 5, 2009 at 5:58 PM, Josh Koffman<joshybear@gmail.com> wrote:
>> On Sat, Sep 5, 2009 at 8:36 PM, Ray<list@microdesigns.biz> wrote:
>>> Josh,
>>> I was doing 4 quadrant decoding of a rotary encoder on a motor back in =
the 70's
>>> The problem is when the encoder rests on an edge instead of off 45 degr=
ees.
>>> On an edge the dithering =A0of +1/-1 can be very high frequency in X4 m=
ode.
>>> that means you need to do interrupt on edge (plus & minus) of both chan=
nels.
>>> I would guess that is the mode you are in.
>>> That give you the most steps per revolution.
>>> X2 is easier and X1 more so if you do not mind less steps per revolutio=
ns.
>>>
>>> 3x means you are not keeping up with changes.
>>>
>>> I could draw you a logic drawing for X1/X2/X4 that you can convert to l=
ogic.
>>> Or a truth table.
>>> Let me know.
>>
>> Hi Ray,
>>
>> I think I sort of understand what you're getting at. I'm not 100% sure
>> how it applies to me though. My data sheet shows that at each detent
>> my B channel should be transitioning. I'm not sure interrupt on change
>> is a possibility in my situation.
>>
>> So...I guess what I'm saying is that I'm confused. Help!
>>
>> By the way, my current thought is to try feeding the port with my
>> encoders on it through my debounce routine to see what happens if I
>> verify all the current states. My only concern is that this might not
>> be fast enough to capture a quickly moving knob (which isn't that
>> quick, these are user controls, not on a motor). I'd still like to
>> know for sure why this is happening though.
>>
>> Thanks!
>>
>> Josh
>> --
>> A common mistake that people make when trying to design something
>> completely foolproof is to underestimate the ingenuity of complete
>> fools.
>> =A0 =A0 =A0 =A0-Douglas Adams
>>
>> --
>> http://www.piclist.com PIC/SX FAQ & list archive
>> View/change your membership options at
>> http://mailman.mit.edu/mailman/listinfo/piclist
>>
>
> --
> http://www.piclist.com PIC/SX FAQ & list archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist
>

-- =

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