16 bit PWM using CCP1

[Bruce]

Hi Darrel,

Pretty cool stuff. Have you tested this on other 18F parts? I don't have a 4520 to test
with, but I couldn't get this to work without clearing CCP1CON before moving 9 to it on
a 452 or 4431.

Had to do this;

ASM
MyInt                    
    clrf CCP1CON         ; 18F452 & 18F4431 require this before it changes CCP1 output
    movlw 9
    movwf CCP1CON
    bcf PIR1,TMR1IF      ; clear TMR1 int flag
    retfie FAST          ; use RETFIE FAST to restore WREG, STATUS & BSR
ENDASM

On an 18F452 or 18F4431, the above works fine, but only if I clear CCP1CON first.

This works on a 452 & 4431 with CCP1 & TMR1 ints enabled on any output pin;

ASM
MyInt                    
    btfss PIR1,TMR1IF    ; TMR1 interrupt?
    bra CCP              ; no - service CCP interrupt
    bcf PIR1,TMR1IF      ; yes - clear TMR1 int flag
    bsf LATB,0           ; set RB0 on TMR1 interrupt
    retfie FAST          ; return
CCP
    bcf PIR1,CCP1IF      ; clear CCP1 int flag
    bcf LATB,0           ; clear RB0 on compare interrupt
    retfie FAST          ; use RETFIE FAST to restore WREG, STATUS & BSR
ENDASM

I was just wondering if this might be a fluke with the 4520 part - since none of the other
18F parts I've tested will toggle CCP1 with just the movlw 9, movwf CCP1CON?

[Darrel Taylor]

Hmmm, very interesting.

I had not tried it on any other chips, and with the 4520 you definitely don't have to clear it first.

But I just tried an 18F452, and you are absolutely correct.
It doesn't work unless you clear the CCPCON register first.

Very interesting indeed,
Thanks Bruce.
<br>

[Bruce]

Thanks. I was just curious. I had fought with this one before on a servo driver, and just
couldn't get it to work without both ints enabled, or clearing CCP1CON first.

I'm thinking it may be due to the huge number of problems Mchip had (see erratas') on
ECCP & CCP with this one?

[Darrel Taylor]

I'm thinking it may be due to the huge number of problems Mchip had (see erratas') on ECCP & CCP with this one?

No doubt!

BTW, I like the any Output Pin variation too.
<br>

[Normnet]

Works good!
Thanks Darrel

Would it be correct to say the maximum PWM frequency for a 40 MHz PIC is 152 Hz?
To slow for music.
Know of any > 10 bit PWM PIC's?

Norm

[Darrel Taylor]

Yup, that's what mister-e's calculator says with 40mhz, 1:1, and 16-bit. 152 hz.

The PIC18F2331/2431/4331/4431 series has a bunch of 14-bit PWM modules. But they are completely different from the CCP module.

Bruce has done a few examples, but I don't think any of them were for music.

best regards,

[Normnet]

PIC18F2331/2431/4331/4431:

FOSC 40 MHz
MIPS 10
PWM Resolution 14 bits
PWM Frequency 2.4 kHz

14 bit to slow for CD quality music.
10 bit sounds good though.

A dsPIC would be ideal.

On my first post my code included saving the state of the critical registers.
Would this be required in a longer program?

Norm

[richard]

just tried your "red" code in the isr on a pic18f45k20

you are right

bsf INTCON,6 ;PEIE = 1 Enable peripheral interrupts
bsf PIE1,0 ; TMR1IE= 1 Enable TMR1
these two lines lock it up. can't see why yet

it looked harmless but its not ,its much worse than useless

[EarlyBird2]

The point of coming here is to learn and I am now comparing the architecture of the 18F452 and the 16F1827. They are completely different, not surprise there.

I have found the sfrs are in bank 5 on the 16F1827 but I have no idea how to use this information.

[richard]

CALLS TO ASM always set bank 0

pbp has a macro you can use "chk?rp sfr" it will ckeck that the appropriate bank is set (see status or bank sfr)
the mpasm assembler can also do a " BANKSEL SFR"

BEFORE YOU RETURN T0 PBP YOU MUST SET BANK BACK TO BANK 0

[richard]

bsf INTCON,7 ;GIE = 1
bsf INTCON,6 ;PEIE = 1 Enable peripheral interrupts
bsf PIE1,0 ; TMR1IE= 1 Enable TMR1

intcon bits 6 and 7 should already be set , setting them again should do no harm
pie1.0 is also set , setting it again can't hurt

it just wastes 12 cpu cycles

the problem lies elsewhere, PWM_VAL must be > 30 ,values from 0-29 wont work could that have happened

[EarlyBird2]

intcon bits 6 and 7 should already be set , setting them again should do no harm
pie1.0 is also set , setting it again can't hurt

That is exactly what I thought and could not find anything to indicate otherwise in the datasheet.

the problem lies elsewhere, PWM_VAL must be > 30 ,values from 0-29 wont work could that have happened

No I started PWM_VAL for zero and incremented up in steps of 1 and it worked all the way rolling over and starting from 0. If 0-29 does not work I could not detect it with my voltmeter.

[richard]

N

o I started PWM_VAL for zero and incremented up in steps of 1 and it worked all the way rolling over and starting from 0. If 0-29 does not work I could not detect it with my voltmeter

I'm using a cro those <30ish pwm's cause the output pin to be permanently high when you expect a very small pw , in real life that could be nasty
try a loop range yourself of 0 to say 28

[EarlyBird2]

N
I'm using a cro those <30ish pwm's cause the output pin to be permanently high when you expect a very small pw , in real life that could be nasty
try a loop range yourself of 0 to say 28

Just tried it no problem 0 gave 0 volts 9 gave 3 mV.

[richard]

not on a cro

using a main loop like this
MAIN:

PWM_VAL = PWM_VAL+1
PAUSE 500
if pwm_val > 29 then pwm_val=0
toggle portd.1
Goto MAIN

[EarlyBird2]

not on a cro

What is a cro?

[EarlyBird2]

N
I'm using a cro those <30ish pwm's cause the output pin to be permanently high when you expect a very small pw , in real life that could be nasty
try a loop range yourself of 0 to say 28

I will take your word for it as I have no instrument that will replicate your results, and as I have found in the past just because I can not see it does not mean it is not there.