|
A bit of a change to Primus |
|
|
|
Written by Pete Harrison
|
|
Sunday, 09 September 2007 |
|
It seems I made a bit of a mistake in how I thought the dsPIC could drive its PWM module. Consequently, there will have to be some changes.
First, what can't change is the need to use T1CK and T2CK as external inputs to count motor pulses. There are no other choices on the dsPIC30F4011 without using interrupts. Since I anticipate pulse rates of up to 100,000 Hz per wheel, life will be a lot easier not having to service that interrupt rate, even with a fast processor and a very efficient ISR.
Somehow, I had it in my head that the PWM module could use different time bases for each channel. Wrong - and stupid. So, how to produce two independant pulse trains at frequencies from about 500Hz to 125kHz?
Well, the dsPIC has a set of Output compare modules. These are quite capable of the job and have a variety of useful modes. Each can use one of two 16 bit timers for its timebase. At first sight, it would be best to use the OC modules for both motors and have each derived from a separate timer. Unfortunately, the two timers we can pick form are Timer2 and Timer3. But we have already committed Timer2 to the pulse counting job. That means that only one OC module can be used and it must be driven from a timebase using Timer 3.
This all makes things a bit tacky. The pin assignments are now a little more confusing and earlier posts will need to be edited along with theschematics and such. Hopefully this will be done by the time anyone gets to read this post. Apologies if anyone has done any work based on the errors.
|
