Scaling ADC Result to a Set Range

[richard]

(MaxDuty - MinDuty) would be 250(2)-100= 150

this would make more sense if your pwm value is a byte
it now becomes

MotorDuty = ADCInVAL */ 150(2) + 100

not knowing whats in the sub ChngMotorHPWM

snippets lead to speculation

[RossWaddell]

I'm using 8-bit resolution for PWM so it should be a byte but I have the variable defined as WORD in case the math takes it over 255:

MotorDuty   VAR WORD        ; Actual duty cycle for motor

ChngMotorHPWM:

    CCP3CON.4 = MotorDuty.0
    CCP3CON.5 = MotorDuty.1
    CCPR3L    = MotorDuty >> 2

    CCP4CON.4 = MotorDuty.0
    CCP4CON.5 = MotorDuty.1
    CCPR4L    = MotorDuty >> 2
    
    RETURN

Do you think the issue would be resolved if I change the datatype of MotorDuty to BYTE? I was also thinking I might switch to 10-bit resolution for both ADC and PWM to get the finest grain control of the motor speed, but that may be overkill.

[HenrikOlsson]

> Do you think the issue would be resolved....

What issue? What exactly is it that doesn't work the way you expect? What do you expect and what does it do?

75% dutycycle (8bit resolution) +/-25% from an 8bit ADC readin:

PWMDuty = 191 + (64 - ADCValue >> 1)

191 is 75% of 255, your initial PWM duty cycle. When ADCvalue (BYTE) is 0 PWMDuty (BYTE) will be 255=100%. When ADCValue is 255 PWMDuty will be 127=50%.

/Henrik.

/Henrik.

[RossWaddell]

The goal is to scale the ADC values of 0-255 to a range of duty cycles appropriate for my project (100-252 with the chosen prescaler). If I manually calculate the results are correct:

• ADCInVal =0 yields 100
• ADCInVal=127 yields 175
• ADCinVal=255 yields 252

But perhaps I don't have the order right in the expression because I don't see those results with the above code.

[amgen]

from excel chart, multiply adc val by .6 and add 100........ but can't mult decimal so multiply by 6 then divide by 10 then add 100

[HenrikOlsson]

Hi,
> but can't mult decimal
Sure you can, sort of, by using the */ or ** operators.

PWMDuty = 100 + (ADCInValue ** 39322)  ' Same as 100 + ADCInValue * 0.600006 but likely faster than 100+ADCInValue*6/10

/Henrik.

[RossWaddell]

Thanks Henrik! I'll try that tonight.