This week's update significantly changes the calibration process to be (hopefully) more reliable, accurate, and most of all user friendly. The system no-longer relies on an input initial guess for how large the frame is which was leading to a large number of issues. Now the system will automatically compute the frame size itself and calibrate automatically.
In place of entering the size of the frame, the system now just asks how much belt to extend.
This is the length of belt in mm that the machine will extend when the "Extend All" button is pushed. It does not need to be enough to exactly center the machine ont he frame or anything like that. It just needs to be more than enough to put the machine on the frame (although roughly centered is nice).
After that calibration will proceed normally with the machine using an initial 6 measurements points near its starting point to compute a rough approximation of the frame size and then from there it will proceed with the regular calibration process.
The system is also more robust to handling situations where the calibration process fails to find a good solution. The process will now automatically restart. We initially added this feature to make the system more robust to handling inaccurate starting conditions (because it is now working with a less good initial guess for the frame size), but an unintended consequence is that the overall accuracy seems to be improved.
This week’s firmware update also adds something else new. While every machine is mostly the same, every Maslow4 is attached to a different frame. One of the biggest factors which can influence the accuracy of Maslow 4 is flex in the frame. If the anchor points aren’t truly fixed in place then the math isn’t able to precisely determine the machine’s position.
To help get more insight into frame rigidity we’ve added an extra measurement during the calibration process. After taking the first measurement in the center of the sheet we take one more measurement in exactly the same spot, but this time we pull the belts a little bit harder. If there is no flex in the system we would expect both measurements to be identical, but if there is some flexibility when we pull harder the frame will flex and give us a different measurement.
The results will be presented like this:
[MSG:INFO: Measuring Frame Flex]
[MSG:INFO: Flex measurement: TLBR: 1.940 TRBL: 4.236]
It’s a bit experimental at the moment, but we would love feedback on if it is valuable information.
There might be some growing pains with this week's update because it's got a LOT of changes under the hood, but overall it feels like a great direction. As always, if you find any bugs let us know and we'll fix them!