Using 2 PWM Modules With Distinct Duty Cycles

chAli · Joined: 14 Oct 2005 Posts: 4

Great forum, it's helped me a tonne with getting up on my feet pretty quick with PIC programming.

I've seen a lot of posts on topics similar to this, but am having trouble finding a solution to the current problem I'm having. I'm a bit of an amateur, so maybe it'll just be a quick code tweak - who knows?

Without further adieu:

I'm using a PIC16F877A (40 pin PDIP) chip and want to use 3 digital inputs to control the duty cycles on the PWM pins (CCP1 & 2). I got it to work pretty well with a single PWM channel on the PIC16F88, but when I use two channels, they seem to stay at the same duty cycle, no matter the input. The duty cycles change on the scope - but it appears that PWM 1 is the same as PWM 2 no matter what I do!

My code is below - I'd appreciate any help someone can give.

Ttelmah · Guest

What you show, ought to basically work. However there are some comments:
First you drive the A0 to A3 pins as outputs, with the comments that you are setting them as inputs. You are not. As soon as you use an output on a pin, it's tris control is set for it to be an output. This should be overriden latter when you use the pins as inputs, but shows perhaps a slight misunderstanding of how the I/O control works.
You then set the tris yourself for the same registers, and have the bottom four bits as '0' in the value sent to tris_A, which configures these pins as outputs again. Same comment applies.
Now you don't say how the pins are actually driven, so the possibility exists that the problem may be the levels are not what you think.
You then do a similar thing with the TRIS bits for portC. When used as a PWM output, the TRIS bit for the pin, must be cleared. You set it instead. Generally the compiler is smart enough, to everride errors like this, but you are pushing your luck by generating so many...
Now as a general comment stepping on, why read the three bits seperately, and with intervals between?. Just declare an 8bit value to hold the inputs (called for example 'pins'), and have something like:

chAli · Joined: 14 Oct 2005 Posts: 4

Thanks for having a look at this for me - I'm going to try this stuff out tonight for sure!

And yeah - it's v3.180, and I wasn't sure which way to set the pins as I/O, so I tried both (apparently backwards!). Thanks for clarifying the issues.

I'll let you know how it goes.

Cheers,
Charlie

Ttelmah · Guest

The 'way' to remember the tris directions, is '1' = 'I', and 0' = 'O'. It is a sensible bit of design by MicroChip to keep it simple. :-)
However since you are not using 'fast_io', all the setting is unnecessary. The compiler automatically generates the tris controls.

Best Wishes

chAli · Joined: 14 Oct 2005 Posts: 4

I've encorporated your first suggestion into my code (Thanks!) - and thanks a lot for pointing out those errors, misunderstandings, and oversights. The informed look at my understanding of the I/O settings was key. In re-reading your post and re-writing my code, I realized that I not only messed up the '1 for Input' and '0 for Ouput' settings, but I had the MSB and LSB order reversed when I converted the binary values to hex for setting the tris.

None the less the revised version of my code is below:

Ttelmah · Guest

I would add a delay in the loop, or the wait for the interrupt flag (not an interrupt routine as such). The value fed to the two registers, is latched into the actual comparison reisters at the end of each cycle of the PWM, and the possibility does perhaps exist of some sort of 'hiccup' if the values are being set/changed all the time.
It looks a lot tidier. :-)

Best Wishes

chAli · Joined: 14 Oct 2005 Posts: 4

HUGE thanks. I just tested the circuit and things worked great. I'm going to move to the interrupt rountine next I think. This was a great excercise in setting the I/O registers, and an ... errr ... hilarious(?) revelation in hex-binary conversion.

Again - thanks so much for your help.