New approach to Rotary Encoder

[fratello]

New code works pretty well ... But, how can I avoid false roller codes ? If I turn up, not always the the code "see" up ; if I turn down, not always the code "see" down ...Sometimes, the commands (Up/Down) are inversed !
I try to add some "pause" between different commands, but the problem remains.

[HenrikOlsson]

No, don't try to slow it down...speed it up. If you have any pauses in there try removing those. If your program is going to do 'other' things as well (which I'm sure it is) then you may need to sample the encoder several times thru out the main loop.

The reason is because if you are at one state and then turn forward two states but you do it fast enough so that the code doesn't detect the "middle" state, then it looks like you turned one state backwards instead of two forward.

So, you want to sample the encoder as fast as you can possibly turn it - and then some.

/Henrik.

[fratello]

I try both variants, presented in code, but without major improvements.

;==============
;reading buttons on row 0 (pseudo-encoder)
    PortA.1 = 0
;    pause 10
@   nop
    newState = PortA & %00011100         ; this is for my hw : 11100 = 28
    PortA.1 = 1
    If newState <> 28 THEN
        If newState <> oldState THEN
              
            Select Case oldState
                Case 12
                If NewState = 24 THEN 
                dir=up         
                endif
                If NewState = 20 THEN 
                dir=dn
                endif
                
                Case 20
                If NewState = 12 THEN 
                dir=up         
                endif      
                If NewState = 24 THEN 
                dir=dn
                endif
                
                Case 24
                If NewState = 20 THEN 
                dir=up
                endif        
                If NewState = 12 THEN 
                dir=dn
                endif
            END SELECT
            oldState = newState
            GOSUB UpOrDn  
            pause 100
    endif           
    ENDIF

[HenrikOlsson]

Which two variants? I only see one in your previous message.
In that one you have a PAUSE 10 at the beginning, which is commented out, and also a @NOP. Then you have a PAUSE 100 near the end which is NOT commented out.

/Henrik.

[fratello]

Sorry...
"@ nop" ... not good
and then
"pause 10" ....idem.

[Ioannis]

If you place two 100nF to ground for debouncing false results will be limited.
Ioannis

[fratello]

I already tried that...without results ! %$@#^*#@^ roller !