I only read a few lines of the comments you linked, but I think I understand the two basic principles he's talking about:
Open-loop vs closed-loop.
Simply put: Open loop means "stepper mode". That means you set the moving direction of the motor and then "clock" through every single step. Means if you want the motor to turn 90°, you put out exactly as many clocks as it takes steps to turn it by 90°. The clock frequency defines the rotating speed of the motor, which makes it "jerk".
It is difficult (but possible!) to generate a smooth rotation in open-loop mode.
Closed loop means that instead of controlling every step yourself, you build a hardware that will automatically commutate the motor coils, effectively converting your stepper motor into a brushless DC motor that can be controlled with a simple DC voltage, or of course PWM controlled DC voltage.
You will need an encoder in order to be informed about the rotation, and you need to react to the motor's rotation instead of "directly programming the amount of rotation in steps".
This makes the control circuit much more complicated - expecially for "stop-and-go moves" which are pain to control with BLDC motors - but in turn makes the motor's movement much smoother because it will always commute in the right moment instead of jerking from step to step. Closed-loop mode is also a lot more energy-efficient and develops the greatest possible torque.
You can force-stop the motor by hand, it will stop. When you release it, it will continue to turn. The machine doesn't care but instead resumes operation. If you force-stop a stepper, then it will lose steps and not move to the intended position when released again.
I've toyed around with a very nice hardware which supported both modes. For your reference: This control board costs pretty much the same as a whole UM1 kit.
In my limited experience (I only had the chance to do a single 100 hours project myself, but I did have the time to try out both modes for it), I'd say that closed-loop mode is probably not useful in a 3D printer because you need very precise control over short amounts of rotation, which is the strength of stepper mode (open loop).
You don't usually need to build up a constant amount of RPM over a certain time, but instead you will rather move "10 steps left, 300 steps right, 50 steps left, 100 steps right".
Closed-loop mode has it's main strength when you define a target RPM and regulate the output voltage so that it keeps on this target.
If the motor rotates at 100 RPM target speed, and you change the motor load (increase friction -> increase necessary torque), then the control circuit automatically regulates power in order to stay at 100 RPM.
An open loop motor on the other hand would just rotate with the set speed. If you increase the load, it will at some point lose steps and become slower. No regulation present.
But that is nothing that would be useful in a 3D printer...
Imho stepper mode is the way to go with 3D printers. You can simulate a closed-loop system with open-loop mode, basically by adjusting your clocking frequency in a way that makes the motor accelerate and decelerate as it would in closed-loop mode. I believe that is pretty much what the Marlin firmware does, and it doesn't seem to do it too badly...
An FPGA would be a very nice thing to have in order to get a flawless multi-motor control. For professional closed-loop applications, the FPGA is pretty much a must-have anyways.
I'm certainly not an expert on motion control systems or motors, but to be honest a lot of what he's describing in that post sounds like the acceleration/deceleration planning in Marlin, and the 'cheap controllers with USB/PWM/ADC' sound like arduinos.
Maybe he's talking about a deeper level of planing and control. Certainly we aren't using servos, or head position sensors on our Ultimakers, and the pseudo-jerk control in Marlin isn't the real thing. But I'm not sure we're so far away from the utopia he describes.
Link to post
Share on other sites