Setting up 5i25+7i85+8i20

In any case you will need speed feedback, sure how that will work with just hall effect
sensors

andypugh wrote:

PCW wrote:

In any case you will need speed feedback, sure how that will work with just hall effect
sensors

It would probably work quite well if the bldc component had a velocity output. But that is something I forgot to add.

A ddt on the angle data ought to work, except that it wraps every rev.
So probably a little hal component would be needed

May sound silly but can't the signal fromnone hall sensor pin be connected to the encoder component to provide a velocity? It's been a while since I messed with hal

I didn't think quadrature would work since the hall sensors are not at 90 degrees.

By smoothing, do you mean something like the lowpass filter?

I assume Ill want to dump this into the motion controller, " => motion.spindle−speed−in"?

Using the hall sensors with the encoder component for a velocity feedback seems to work well. I would expect to see 16 counts from the 2 hall sensors in x4 mode but I see 12 per revolution so I have the encoder scale at 12. Lowpass gain of .01 seems to give a smooth velocity.

I am having alot of trouble with the PID loop. I can get it decent when the max output is limited to 2(amps), when I raise it to 7 I can't get the speed to come down. One thing I read suggests when tuning the PID, you should set I and D to 0 and raise P until the output oscillates, then I until the oscillation stops. I assume by "oscillate" it means starts making minor corrections when it's at speed. and not swinging between positive and negative values.

I seen PCW mention in another post that I and D should be left at 0 until it's close and only use P and FF0 until then, I have not tried this yet. I'm pretty sure I had to backdown on P because the motor would bang back and forth between hall counts when stopped. Looking at the docs, it says " the deadband should be set slightly more than one-half count". Does this actually mean .5 which is 1/2 of 1 count, or 6 which is 1/2 of counts per rev?

For a spindle where the PID command is velocity and the PID output is torque,
you mainly want P and I

FF0 makes sense if the PID input and output are velocity but that's not the case with a torque mode drive.

Tuning a velocity loop is going to be tough with such poor (hall effect only) feedback, unless the spindle inertia
is quite high relative to the torque

PCW wrote:

For a spindle where the PID command is velocity and the PID output is torque,
you mainly want P and I

FF0 makes sense if the PID input and output are velocity but that's not the case with a torque mode drive.

Tuning a velocity loop is going to be tough with such poor (hall effect only) feedback, unless the spindle inertia
is quite high relative to the torque

Would it be better to use the spindles drilled disk for feedback than the hall sensors on the motor? It's a single photo interrupt sensor so no quadrature, and the disk only has 4 holes but theyre more consistently spaced than the 2/3 hall sensors. It'll be a bit before I get the parts to add the differential encoder to the motor so one of these will have to do for now.