robotic arm control , limit switches ? or

hello in the process of studying and implementing a robotic arm design , many questions arise , i hope someone will help me clear the image ..

1. a-if using incremental encoders and dc motors for joints , is it necessary to use "2" limit switches on each joint to limit the movement or is software counting enough ?? same question for absolute encoders ??
b-if i have to use limit switches is connecting them directly (using some gates ...etc) to SD pin in IR2110 or similar driver a good practice or should I enter them to microcontroller and choose internally to set PWM to zero ? which is safer and which is better for stopping the motor quickly in case of an accident without blowing up the power stage ??

2. I have decided to use PMDC motors and after reading about the servo control , i found out that the control loop consist of 3 main part , position loop , velocity loop and internal torque loop .
I am using only a PID for position which calculates the required duty cycle (voltage control)and current limiting is done in hardware externally . DO i have to worry about velocity and torque ?? where is torque control needed , why are these three always used with each other ??? AM I MISSING SOMETHING ??

thanks for reading all this staff and for your help in advance , please guide me if i need to post the second question in another forum (micro controllers maybe ?)

If you're using an incremental encoder then you have no way of knowing where the arm is when you first power up, so limit switches are an absolute necessity. On power-up you drive the arm in one direction until you see the limit switch; that's your "zero" point. If you are using an absolute encoder, then the limit switches aren't an ABSOLUTE necessity, but they offer some protection for the mechanism if your software fails.

I'll let somebody smarter than me answer the control-loop questions.

1. I think the limit switches are necessary whatever you use incremental encoder or absolute encoder.

barry said:

If you're using an incremental encoder then you have no way of knowing where the arm is when you first power up, so limit switches are an absolute necessity. On power-up you drive the arm in one direction until you see the limit switch; that's your "zero" point. If you are using an absolute encoder, then the limit switches aren't an ABSOLUTE necessity, but they offer some protection for the mechanism if your software fails.

I'll let somebody smarter than me answer the control-loop questions.

Click to expand...

thank u , yes it makes sense . have u ever seen an industrial robotic arm ? i am interested in knowing how they fit those sensors on each joint (for for 6dof about 12 sensors!!) , donot they complicate the mechanical structure ?? and what type do u think is appropriate : magnetic vs photo vs proximity... for indoor vs outdoor .