Rotary encoder accuracy

[Johnwa]

@jcdammeyer I’m at a bit of a loss. My comparison of the AS5600 to an optical encoder gave me some unexpected results. There is a periodic difference between the optical encoder and the AS5600 and the number of steps the stepper has taken. On the other hand the AS5600 and the number of steps track reasonable well. On the attached chart the x axis is the position of the optical encoder with the other two lines representing the AS5600 and the position based on number of steps. For this test the optical encoder was directly drive. I have a sneaking suspicion that the AS5600 magnetic field is being distorted by the stepper magnetics. The magnet is glued to the back end of the stepper shaft much like it Is for these “closed loop” steppers https://www.aliexpress.com/item/1005003380684185.html

 

[jcdammeyer]

I wonder if the closed loop steppers use an absolute position rather than a calculated # of steps? So they can't really lose position per revolution and use the A/D value of the magnetic value to issue a step number. It can self correct as the unit rotates.

Think of a grey code encoder that has 9 bits or 9 slots per rev which would be 512 distinct positions determined by the optical sensors. At any point you know exactly where it is. Now if noise or something corrupted the signal it wouldn't matter because a bit further along you'd be back to the correct value.

 

[jcdammeyer]

BTW, when I purchased the stepper-on-line motors for the robot arm I went for the non-integrated motor+encoder assembly. The CUI AMT10 series encoders I purchased are marked as capacitive ASIC so much like the magnetic I think. Their built in encoders right into the back of the motor are listed as magnetic.

https://www.omc-stepperonline.com/nema-23-closed-loop-stepper-motor

Even if they did have a variation they also know how many times they've change current direction in the windings so they have a fundamental 200 coarse steps per rev and only really have to convert the signal, via the A/D for the value between those steps. So theoretically can catch up on a proper reading if indeed they have some variation in the signals.

I went for external so I could install an optical one if I wanted or if I changed to other motors or projects I could move the encoder and just scrap the smaller motor.