How to make HPWM slower with 16F887?

[Scampy]

Ideally, I'd like to get it to go as slow as 3 or 4 seconds to complete a duty cycle.

Henrik, if I read this correctly, I believe that he wants to fade the LED up from 0% to 100% over a 4 second period, rather than have a 4 second frequency, but have to agree the description is a little hard to follow.

[keithv]

Henrik, if I read this correctly, I believe that he wants to fade the LED up from 0% to 100% over a 4 second period, rather than have a 4 second frequency, but have to agree the description is a little hard to follow.

Right. I want to make the LED slowly and very smoothly get brighter and dimmer. I do not want it to blink on and off. If there is a better way to do this than PWM or HPWM then please tell me.

[HenrikOlsson]

Using a PIC there's no better way than using the (E)CCP module in PWM-mode - no.

Using the HPWM command to "drive" the (E)CCP module in PWM-mode may not be ideal as have been described but it will work as a starting point. If you want to drive the ECCP module manually there's another thread active right now which discusses just that and there's a lot of info burried in the forum.

And, as have been said a couple of times now, you DON'T want to make the PWM output "slower", you want to slow down the rest of the program. It's not the PWM frequency that determines the "dimming speed" it's how often you update the dutycycle.

/Henrik.

[Scampy]

Basically you need a frequency fast enough so that the eye doesn't detect it, and a smooth ramp up / down in duty cycle over a 4 minute period.

The problem is that with just 255 steps between 0% and 100% you will notice the stepping in brightness at low levels, such as when the value is in single figures. For smooth dimming you need 4096 steps. One solution would be to use an PCA9685 chip which has 16 PWM outputs, each with 4096 steps, and is controlled via I2C bus.

[richard]

The problem is that with just 255 steps between 0% and 100% you will notice the stepping in brightness at low levels

most pics have 10 bit pwm and if used the "stepping" effect is negligible , why not to use the 1024 steps


here is a simple demo with the led on portc.5 "stepping" at low levels even when gamma corrected

the led on portc.3 is as smooth as silk

'****************************************************************
'*  Name    : fader.BAS                                      *
'*  Author  : [select VIEW...EDITOR OPTIONS]                    *
'*  Notice  : Copyright (c) 2016 [select VIEW...EDITOR OPTIONS] *
'*          : All Rights Reserved                               *
'*  Date    : 6/30/2016                                         *
'*  Version : 1.0                                               *
'*  Notes   :  pic16f1825                                                 *
'*          :                                                   *
'****************************************************************
 
   
#CONFIG
             __config        _CONFIG1,    _FOSC_INTOSC & _CP_OFF & _WDTE_ON  &  _PWRTE_ON  &  _MCLRE_ON  & _CLKOUTEN_OFF
              __config      _CONFIG2, _PLLEN_ON & _LVP_OFF
#ENDCONFIG
 
 
DEFINE OSC 32 
 
 
 
goto overpwm 
 
 my_pwm:  
;100 step 1.35 gamma curve   
@ dW   1,  1,  1,  1,  1,  1,  1,  1,  1,  1
@ dW   1,  1,  1,  1,  1,  1,  2,  2,  2,  2
@ dW   2,  2,  2,  2,  2,  3,  3,  3,  3,  3
@ dW   3,  4,  4,  4,  4,  5,  5,  5,  6,  6
@ dW   6,  7,  7,  8,  8,  9,  9, 10, 11, 11
@ dW  12, 13, 14, 15, 16, 17, 18, 20, 21, 23
@ dW  24, 26, 28, 30, 33, 35, 38, 41, 44, 47
@ dW  51, 55, 59, 64, 69, 74, 80, 86, 93,101
@ dW 109,118,128,138,150,162,176,190,206,224
@ dW 243,263,286,310,337,366,398,433,471,511
 
 

ASM
;---[Returns the Address of a Label as a Word]------------------------------
GetAddress macro Label, Wout
    CHK?RP Wout
    movlw low Label          ; get low byte
    movwf Wout
;    movlw High Label         ; get high byte  MPLAB 8.53 killed high
    movlw Label >> 8         ; get high byte
    movwf Wout + 1
  endm
endasm
overpwm:
 
   OSCCON=$70
   ANSELA=0
   ANSELC=0
   paddress var word
   pw       var word
   pl       var byte
   pl2       var word
   inc var word
   t2con=5 
   TRISA = %111110
   TRISC = %11010011         ' set PORTC I/O
   led1 var portc.3       ; 10 bit straight pwm
   led2 var portc.5        ; 10 bit 100 step gamma corrected pwm
 
 
@ GetAddress _my_pwm,_paddress 
pl2 =0  
pl=100 
  
main:
inc=1   
while pl
    pl=pl-1
    gosub setpwm
    pause 50
wend
inc=-1
while pl < 99
     pl=pl+1
     gosub setpwm
    pause 50
wend
goto main

 setpwm:
    readcode paddress+pl ,pw
    CCPR1L = PW>>2;
    ccp1con=12|((PW&3)<<4);     led2
    pl2 =pl2 +inc
    if pl2 >1023 then pl2 =0 
    CCPR2L = Pl2>>2;            led1 
    ccp2con=12|((Pl2&3)<<4);
return

[Dave]

Richard, I can't help but notice your table only goes to 511. That's half of the available pwm pulse width.

[richard]

I can't help but notice your table only goes to 511. That's half of the available pwm pulse width.

that's correct dave.
The point was to demonstrate that its possible to eliminate the visible brightness steps that are most evident at the dimmest fading levels. The steps are quite visible at the lowest 20 or so pulse widths wether its gamma corrected at full 10 bit resolution or linear at 8 bit resolution .
100 steps was all that's necessary to prove the point , it can be done

the missing 10 steps

100	481
101	522
102	568
103	617
104	671
105	729
106	793
107	863
108	940
109	1023