I was going to upgrade my LR2 to an SKR Pro 1.2 to enable dual endstop (so 5 channels), but I realized that the driver I have installed (TMC2209) have a builtin sensorless homing function.
So that does raise a few questions:
Is sensorless homing accurate not only with regards to “squareness” but also with regards to positioning : we need this function to be able to swap the tool (from big bit to smaller bit), so need accuracy as well
on the Z side, our 0 would be the top of the board, need automated homing here, but wouldn’t sensorles homing damage the bit by forcing it into the wood, or worse, it would actually create a small cut so subsequent homing accuracy would be affected ? If I decide to the the switch route for Z only (ie metal plate at the location of the Z homing, with an electric connection to the drill bit), can I simply cut the pin on the TMC2209 driver so that it uses the endstop physical connection instead, or do I need to modify the firmware as well ?
We don’t use sensor less homing. We actually bend the pin on the driver. We use endstops and that gets us accurate. Someone will come along that can explain it way better than I can.
At best. Sensorless homing needs the motors to skip a minimum of 2 full steps, and up to 8 depending on settings.
After skipping, the motor will stop at a full step, closest to where the actual physical limit is. This means a variability of an average of a half step, or 0.08mm variability. Most people can live with this, I think.
In theory this should be the same place each time, as the physical stop should be the same location, but in practice, when the motor starts to skip, sometimes it jumps back to more than the last full step. This can be off from the previous position by 0.16mm or even 0.32mm, which is more than my personal tolerances, though this should be rare.
Microswitches are very repeatable, and because the homing sequence will stop the motors in microsteps, homing is repeatable to the limits of the switches. This is always within reasonable tolerances.