Ok, this is something I have been thinking about for a while, and had mentioned in the electronic home switches thread;
Electronic home switches made easy!
Since the electronic Hall type home switches can sense machine position to better than 0.001" I wondered if it was possible to use the home switch to automatically measure the backlash of the machine, which would enable the machine to automatically change its backlash compensation settings to adjust for wear or for thermal expansion.
I have recently been using my small plastics milling machine to make a few small parts that need tight tolerances, and have been bugged by some thermal expansion issues where my leadscrews grow a couple thou in length when they get warm, so features on the part are displaced by 0.002" or so. That was fixed easily enough by changing my homing routine to re-home frequently during the job and this compensates the thermal expansion.
However my backlash also changes a bit when warm. Measuements (I'll post more on the measuring techniques later) show a cold backlash around 0.0025" and 0.002" for my X and Y axes respectively. But when the leadscrews warm up they expand and the backlash drops to half that, even a bit less, both axes come in around 0.001".
Using the home switch to measure backlash?
This works based on one concept; That the electronic home switch which we know has a reliable ON position detection, would also have a reliable OFF position.
The Hall switch has hysteresis, this term means that once the magnet is close and the switch turns ON, the magnet must be moved away a bit before the switch turns OFF. This hysteresis is designed into the switch, and I assumed the turn OFF position was also thermally compensated etc like the turn ON position. Testing looks promising at this point.
How does the switch do it?
Well it is simplicity itself. The magnet is brought close until the swtch turns ON, this point is recorded.
Then the magnet is moved away a couple mm until the switch turns OFF again. This position is recorded.
The distance between the 2 positions I called the "sensor backlash" for lack of a better term. This distance includes BOTH the actual sensor hysteresis AND the machines mechanical backlash (as the machine has reversed direction).