I think it might be useful to relate my experience with V88.
In short I would recommend letting it run, it is a radical change from previous versions, but the result appears to be better.
I have had to relocate my M4 due to the premises I was using no longer being available, so on reassembly of the frame after modifications I was ready to recalibrate.
Using the default settings, I could not extend the belts enough to reach the anchor points. I reset the Extend distance 2300 (which was a bit long) saved the file and restarted the from the FluidNC menu.
Did the retract extend dance then got an error when I clicked on recalibrate saying can not continue until belts fully extended.
At this point I reset Extend to 2050 and saved, then powered off the M4 and went through the rig to rerun config.
The initial readings were really bad (around 0.01) and I was tempted to stop and try to figure out why, but decided to let it run.
Initially it just did a square, nowhere near the original starting point as previous versions had always started from, then it had a think and pulled into the centre (or close too it). From then on it at each iteration the numbers improved until by the time it finished the fitness level was 0.7, the best I have seen for my setup.
I clicked on save to save the log file and it said it had saved it but I was unable to find the file on my computer. Is it possible to automatically save the log file on completion of a config run, especially if it fails? Previously I have found it in the Download folder of the host computer.
Is there anything that we could do to make the process easier to understand, or do you think it’s just that we’re all used to the old system and now things are different which makes it a little confusing?
This is the place I would always expect it to be. Unfortunately we don’t have any control over where the file ends up because we send the file to the browser and the browser decides where it gets saved
I think we need an advisory that initial results are going to be less than expected, previously it started with high values which went down as it proceeded, now that is reversed i.e. starts bad and gets better.
On the other point:
I was using a different browser from the one I normally use and I suspect it didn’t save the serial file at all. A bit disappointing as I was looking forward to analysing the process.
I’m running a Maslow 4.0 with JST-XH controller board and encoders on the default wooden frame, vertical at about 20º of incline. I don’t get to leave it up—no room in our apartment—so I build it wherever I’m doing some work. Currently, it’s in the scene shop of the high school at which I teach, and I just finished running two calibrations with firmware v0.88: one with a 7x7 grid and a second with a 9x9 grid.
Both ran beautifully, and at the end of the second run, I ended up with a fitness of 1.2216100869810815.
Given my struggles over the course of almost a year (all user error), I’m just gobsmacked. Will be cutting on Saturday, and I just can’t bring myself to say I’m optimistic…
