Is shutting down cogs OK when 8 is not enough?

lardom

My project uses seven cogs. I am using one cog to control four servos. I want to use four DC motors instead of four servos to control a robotic arm. By attaching a pot to a motor I can control it like a servo but I can only control one motor per cog. The problem I have with multiple dc motors is controlling overshoot.
One solution would be to experiment with PASM. The second idea is to control two motors with two cogs at a time. I'm tying up both counters to control motor speed so I'd have to reinitialize the counters each time I switched motors. Is this advisable?
My goal is to scale this thing up so I can make it do useful work.

kwinn

Not enough info in your post for a definitive answer, but it would depend on how much processing needs to be done for controlling each motor. I would look at the calculations to see if a cog has enough memory for the code and variables needed for the four motors.

Duane Degn

What sort of resolution and frequency to these motors need?

There are a variety of PASM PWM drivers for use with multiple motors.

I'm pretty sure I have a PWM driver which produces four signals from a single cog (I have one which also reads four quadrature encoders with the same cog).

If you describe your requirements in more detail, I bet we could help you find an appropriate driver.

lardom

Duane Degn wrote: »

What sort of resolution and frequency to these motors need?

There are a variety of PASM PWM drivers for use with multiple motors.

I'm pretty sure I have a PWM driver which produces four signals from a single cog (I have one which also reads four quadrature encoders with the same cog).

If you describe your requirements in more detail, I bet we could help you find an appropriate driver.

The motors don't need much resolution. I'm using Chip's mcp3208 driver and I set it to return 0 to 63. The concept and the code is simple. I rotate a potentiometer and then the motor, which is also connected to a pot, rotates until it matches the value I sent it. It works well. If there's just one motor I can also adjust the pwm duty cycle and avoid problems with the motor overshooting the commanded position.
This is what it looks like:

Repeat
      MyPot := ptest.get_temp1       
      MotorPot := ptest.get_temp2         
      If MyPot > MotorPot                    
        main(3000, 0)
      elseif MyPot <  MotorPot                      
        main(0, 3000)
      elseif MyPot ==  MotorPot 
        main(0, 0)

When I tried testing two motors I could see that this approach didn't work.

kwinn

One cog running a PASM version of that simple program using integer math should easily handle 4 motors, particularly if you take advantage of the high speed MCP3208 object.

If needed it could even be modified to decelerate as the motors approach the desired position.

User Name

lardom wrote: »

One solution would be to experiment with PASM.

Can't endorse that highly enough.

lardom

User Name wrote: »

Can't endorse that highly enough.

I'm just about ready to tackle my first pasm project. The newest thing for me is "flags". I'll start off with small test code.

Mickster

lardom wrote: »

My project uses seven cogs. I am using one cog to control four servos.

My goal is to scale this thing up so I can make it do useful work.

Based on the above statements, if budget will allow it, I would suggest closing the servo-loops with incremental encoders (like real robots). With only one cog available, you might want to consider motion control co-processors via RS485, such as:

http://www.jrkerr.com/icproducts.html

lardom

@Mickster, Thanks, I bookmarked the page.