- Joined
- Dec 8, 2013
- Messages
- 2,651
True, but the loading effect is much less for a DC or PMAC and there is much less overhead in the error correction math. CNC machines do not typically utilize induction motors for that reason. I'm sure someone could dredge up an example, but I've never seen a CNC with induction motors on it.
I don't think it's the math anymore: optimal control can be implemented on a $10 DSP. I think it's primarily the lack of full power speed range. In any case there's no good reason to use an induction motor on an expensive CNC machine with such nice PMAC motors available.
On the other end of the spectrum is the series wound or universal motor. This type is never used for anything requiring tight speed control. Though if you're a firm believer in PID, there should be no reason for this, as in theory it should be just as controllable as an induction motor (which is just as controllable as a DC or PMAC motor).
PID control of a series motor is tricky because there is an extra integrator in the loop. I've done it though: I use one for the Z axis on my mill (it's what I had). The DC treadmill motor on my lathe is also a series motor, but the controller is just proportional. A series motor is not my first choice for precise speed control, though.
I've experienced my share of PID loops and I'm not a big fan. I've had to design my own speed control algorithms in the past when PID failed me. For example, in a wire & cable processing plant I used to work in, the take up and payoff winders were particular about the size of reels they used. You put a reel too big in it, and the PID will be unstable; what it considers a "slight" adjustment of RPM results in a wild change in wire speed, because the circumference of the reel is much larger than the reel which was used to tune the PID. If the reel inserted is too small, then the "slight" adjustment is too slight, and the machine does not respond quickly enough. I had to move the PID out of the VFDs and into PLCs, where I could program a more elegant solution, which would look at an ultrasonic rangefinder to determine reel circumference and adjust the proportional constant appropriately.
You fixed a design error.
I've worked with some old timers in the motion control industry and they say that for some applications there is no substitute for an old school setup with a DC servo, analog tachogenerator feedback, and an analog DC drive. The tach output is time-now; does not need to have any math performed for averaging encoder pulses. The drive circuitry is time-now; does not rely on a microcontroller running on a clock cycle. The output is time-now; no adjustment of PWM duty cycle, just instant correction of linear output. I've never encountered an application requiring this level of literally instantaneous response, but their explanation makes sense to me.
PID control is older than digital encoders and microcontrollers. It was developed to improve on the performance of simple proportional controllers such as you describe. And it did. A modern system with a high-resolution encoder and a fast DSP can do better, though.