Working on calibrating my M4. I built my frame following the following the recommended vertical frame. I am limited in height because my shop is located in my basement so i only had room for a 7x10 frame. I had the idea to mount my frame with hinges to a 2x4 so i could lift the frame over head to store sheet goods in the empty space behind the frame. Unfortunately i had to clip the bottom left corner in order to pass by the main sewer stack. At first i was running into fitment issues when calibrating with my original settings. Slowly working thru everything and trying different frame dimensions until i could complete a calibration process. Ended up getting decent fitment over 0.8 but after jogging around i noticed everything is skewed a bit. I’m assuming with the math trying to square up the work surface with points it was given. Not sure why the math doesn’t use the top two points as constants and then use trig to determine the lower two points based on the first pull right measurement. My next step is going to try and run the calibration process again with the new findings and see if it will slowly solve for the unique situation I’m putting it in.
I can understand that in a perfect world the 4 corners would be close to square as possible but then again hypothetically if i were to mount it to my shop floor i would hope to look for points that might be less susceptible to damage rather than in perfect alignment.
Another thought process would be to force the algorithm to solve for three points leaving out the lower left, then solve for the opposite three points leaving out lower right and then plug those numbers back into the machine.
Working on calibrating my M4. I built my frame following the following the
recommended vertical frame. I am limited in height because my shop is located
in my basement so i only had room for a 7x10 frame. I had the idea to mount
my frame with hinges to a 2x4 so i could lift the frame over head to store
sheet goods in the empty space behind the frame. Unfortunately i had to clip
the bottom left corner in order to pass by the main sewer stack. At first i
was running into fitment issues when calibrating with my original settings.
Slowly working thru everything and trying different frame dimensions until i
could complete a calibration process. Ended up getting decent fitment over
0.8 but after jogging around i noticed everything is skewed a bit. I’m
assuming with the math trying to square up the work surface with points it was
given. Not sure why the math doesn’t use the top two points as constants and
then use trig to determine the lower two points based on the first pull right
measurement. My next step is going to try and run the calibration process
again with the new findings and see if it will slowly solve for the unique
situation I’m putting it in.
The math uses the bottom two points as the line, not the top two. It could have
been done using any two adjacent points as the reference for straight, but bar
happened to pick the bottom two.
also note that shortening the frame from 8’ to 7’ does narrow your effective
vertical work area to about 40" high before you run into places where the arms
will hit the rail supports.
I can understand that in a perfect world the 4 corners would be close to
square as possible but then again hypothetically if i were to mount it to my
shop floor i would hope to look for points that might be less susceptible to
damage rather than in perfect alignment.
it handles cases where they are not close to a good rectangle, but the initial
calibration guess needs to start somewhere.
Another thought process would be to force the algorithm to solve for three
points leaving out the lower left, then solve for the opposite three points
leaving out lower right and then plug those numbers back into the machine.
That has been discussed. I also have an onshape doc that lets you put in some
measurements and have it calculate where the anchors are in the coordinate
system defined by the lower two anchors
This gives me confidence that when i start cutting things won’t be a complete rhombus. Is there a way i can edit the yaml for to use the top two points or is this something that is held deep in the code for configuring square. I guess i just thought with the machine being vertical those would be the most accurate. As to running into the arms i have noticed that it will bump into the supports but just rotates the machine on its access. Would you suggest just lowering my grid size to eliminate any potentially false readings that could change the math for calculating the corners? Thank you for your feedback
This gives me confidence that when i start cutting things won’t be a complete
rhombus. Is there a way i can edit the yaml for to use the top two points or
is this something that is held deep in the code for configuring square.
it’s in the calibration routine, but after you calibrate, you could rotate the
coordinate system any way you want, you just need to keep the relative distances
(including diagonals) the same and still use the lower left as (0,0)
As to running into the arms i have noticed that it will bump into
the supports but just rotates the machine on its access.
That is correct, until a second arm hits the support going the other direction,
at that point you can’t rotate to solve the problem.
It’s not clear how badly this affects your accuracy (people have been struggling
to get good calibrations, so nobody has done a study of this in real life)
Would you suggest just lowering my grid size to eliminate any potentially
false readings that could change the math for calculating the corners? Thank
you for your feedback
Ran another calibration this morning and lowered my grid to a 9x9 in 1800x800, also raised my tension to 1600 as noted in a few other posts and these are the results i got
It is an unfortunate circumstance i had to clip the lower left corner seeing as that’s the primary corner that the machine works off of. All in all fitness score looks great. Started in the 2.0 range and stayed promising. I guess it would be easier to rotate the workpiece to what the machine sees as horizontal and then take some measurement from there to see how true everything is staying
