Solar Tracker Controller

I am working on a single axis solar tracker but I am having trouble adding hysteresis to my project. I am using a PIC16F688.

Code:

---

' * James Richards *
' * Solar Tracker  *
' * 5/21/10        *

'----------Configuration----------
INTCON = $80
ANSEL  = %00010100
ADCON1 = %00010000
'----------Variables--------------
ADCVAL1 var word
ADCVAL2 var word
RANGE  VAR WORD
'----------Program----------------
MAIN:
   
    ADCON0 = %10001001
   
    ADCON0.1 = 1
    NOTDONE1:
    pauseus 25
    if ADCON0.1 = 1 then notdone1
   
    ADCVAL1.highbyte = ADRESH
    ADCVAL1.lowbyte  = ADRESL
   
        ADCON0 = %10010001
       
        ADCON0.1 = 1
        NOTDONE2:
        pauseus 25
        if ADCON0.1 = 1 then notdone2
       
        ADCVAL2.highbyte = ADRESH
        ADCVAL2.lowbyte  = ADRESL
'---------------------------------
GREATER:
range = ADCVAL2 - adcval1
if (RANGE > 10) then cw

LESSER:
range = ADCVAL1 - adcval2
if (RANGE > 10) then ccw

goto stay

CCW:
high PORTC.2
low PORTC.3
pause 100
goto main

CW:
low PORTC.2
high PORTC.3
pause 100
goto main

STAY:
low PORTC.2
low PORTC.3
goto main
   

   
end

---

You might try pulling a few samples then average them before acting. It will make up for the stray cloud.

Another way to add hysteresis is to require a return direction change to be greater than an advance direction. So, if your last move was to the left caused by a "1x" reading, and a cloud causes a desire to move right, require the reading to be "1.1x", so it's 10% more than the last.

I would also like to add a small adjustable potentiometer to adjust the hysteresis but I will add that in later. I want the tracker to adjust its position every minute but I do not want to use any extra energy overadjusting. I need it to move to a certain position but when it gets there I want it to stay within certain limits instead of doing small oscillations trying to find an exact spot.

You could think telescope.
Every minute so many steps on an encoder.
The sun is pretty predictable :)

Also, perhaps a larger than usual capacitor on the ADC pin would make it slower to react to minor, quick fluctuations.

Regards,

Anand

I sure would place a large capacitor in this case but also an averaging routine.

Melanie had a suggestion on an older post about having 16 samples, sort them, ged rid off the 8 values (4 top, 4 bottom) and average the rest 8. Maybe for this project this is overkill, but i.m.h.o. the best solution.

A noise can always sneak in.

Ioannis