Thanks bar & dlang – Here are some numbers –
Motor distance = 2985mm with calibration process – 3000.4mm with a tape (shafts center to center)
Top edge of 4x8 sheet to motors – 379.5mm with calibration process – 387.4mm with a tape
Rotation radius = 270mm
We adjusted the motor distance (in GC, not actually) to 2975, and the rotation radius to 270.1, and got our 10" square to 1/64" under on the height, and 1/32" over on the width.
Are we going at this right?
Does anyone know what effect fudging any of the 3 variables up or down has on the final object?
I guess I need to spend some time with the simulator – Also, we have a long, horizontal hardwood 2x4 between the 2 motor support arms, so I don’t think flex is a factor.
Your motor spacing measurement inaccuracy sounds like what I was experiencing before 1.00. I think this is where I would focus my attention. Maslow, since 1.00, measures my motor spacing within 1 mm of the tape measure.
Just checking, but are you absolutely certain you are running 1.00? 1.00 fixed exactly what you are seeing in measurement error.
I would second this. That measurement seems pretty off to me, I would expect it to more or less exactly match the tape measure reading. It’s possible that maybe you are running 1.0 but the settings didn’t update for some reason? Could you send a screenshot of you Advanced Settings window like this:
The simulator mode in GC lets you fudge the variables and see what the resulting
distortion looks like.
I agree with the poster above, the first thing is to find the problem with the
motor spacing measurement.
Just checking, when you measured it with the tape measure, was that with the
chain still under tension from the automatic measurement? if it’s not, and your
frame is flexing enough for that sort of difference, it could account for a lot
of your problems.
I think the discrepancy in the motor spacing was because we still had the motors mounted on an angle – When we re-mounted them horizontally, the difference dropped to 3mm – Routing a 10" square, the width is right on, but the height is 9-15/16" – Is there a setting we should play with to affect the height? Has anyone considered a vertical calibration, like the side to side one included in the current calibration process? Here is our screenshot –
The calibration isn’t directly adjusting the vertical or horizontal distances,
when the machine dimensions match what maslow thinks they are, everything will
be accurate.
When they don’t match, things are not going to be accurate.
The ‘calibration’ step assumes that you have entered something incorrectly and
attempts to introduce the right errors to cancel out the errors that you have
already entered.
In the case of the linkage kit, there are three variables in play
the distance between the motors
the distance from the motors to the top of the work area
the distance from the chain to the center of the bit.
The calibration step currently assumes that you have the first two correct and
is experimentally narrowing in on the third.
If your machine flexed as the chain was drawn tight between the motors, you will
never be completely accurate (calibration may run and get you to a square cut,
but as you move away from the center, you will see errors)
In terms of the calibration cut, what matters is how close the height and width
are to each other, not the absolute value of either one (which will vary based
on how you measure and the width of the bit used)
We are working on a new method for triangular kinematics calibration that only
assumes that the first value (distance between the motors) is correct and
calculates both of the other two values.
My problem with calibration was frame flex. After I beefed up the motor flex things got a whole lot better. I would center the sled horizontally and move it to the bottom and measure the motor distance and move it to the top and measure the motor distance. If you are getting a difference you won’t get a very good calibration.
Stiffened up the frame – Distance between the motors, (Chain length vs.Tape measure), under 1mm. – Remains the same with tension or without, sled at top of sheet or bottom.
Using PTB linkage – Spaced all 4 mounting points out 1" extra from the center to increase clearance for Z axis motor.
Run the calibration procedure – Sled centers on the sheet nicely – Using 270mm for rotational radius, run the horizontal calibration routine – Very close, maybe 1/4" off from 75" – repeat until right on.
Using idea I got from rjon17469, triangular kinematics spreadsheet, I position the router near the top center of the sheet and drill a hole – Then drop down 900mm and drill another – measured distance between them is 889mm.
I can correct it to 900mm by modifying either the rotational radius or the motor distance, but that obviously messes up the horizontal distance (75").
Unless you cut new arms with these new dimensions, this could be the problem . The geometry requires that a line between opposite mounting points pass through the center point, and that the hole spacing of the arm spanning them be the same as between the mounting points. Maybe I misunderstood?
if you move the mounting points, you need to have longer arms. otherwise the
geometry of the arms no longer works (the two arms connecting to the mounting
points should always be parallel)
so if you move the mounting points out 1" each, you need a 2" longer center arm
Thanks for catching our mistake – We put the mounting points back where they should be, but still have the same problem – Horizontal calibration distance = 75" – But vertical distance, between 2 points supposed to be 900mm apart, is stil 889mm, This is almost 1/2" difference – Seems like it should be possible to get more accurate.
Perhaps someone else could try this and tell me what result you get?
We also did a 10" square, which came out 9.98" wide x 9.93" high – Am I running up against the inherent limits of the Maslow?
The spreadsheet isn’t ready for use in this way yet. Have you tried the horizontal cut method in Step 10 of the calibration procedure? It should get you closer than what you describe. @rjon17469 is working on an improved calibration, but good things take time - in the mean time why not try Step 10 and see what that does?
Yes, we ran the horizontal calibration till it measured exactly 75" – Then moved the router to top center, drilled a hole, then dropped it down 900mm and drilled again – measurement between hole centers was 889mm – If I modify RR so that 889 becomes 900, it messes up the horizontal distance – I cant figure out how to change vertical calibration without changing the horizontal
vertical and horizontal are not separate things to calibrate
doing horizontal calibrations really doesn’t do any good. once you get things
correct, they will be correct in both directions.
when you changes RR so that 889 becomes 900, it would also change the chain
lengths for the other hole.
so you would need to do the same thing again (drill two new holes and see how
far off you are). This is the basic approach taken by the current calibration.
Isn’t there some $ or B command that will report the encoder position? If so, it
should be possible to use that to calculate the chain length (or if you can get
the chain length directly from a $ or B command, that should be even better)
In the spreadsheet I assume the horizontal distance between motors is accurately measured based off the chain tension method used in the calibration routine. The calibration algorithm I’m developing then alters the motor Y coordinate along with the rotation disk radius to achieve the desired results. Adjusting just the rotation disk radius is likely to produce undesirable results.
I should have the new calibration routine PR submitted this weekend and ready for testing, so it’s coming shortly!
Absolutely! We did a complete re-install, wiping the Eprom (whatever that is), and we dropped back to the 1.0 GC and firmware – We were having a problem with 1.02 – the sled dropping off the bottom of the workspace during the horizontal calibration process, that we hadn’t had with 1.0.
Could someone with a correctly calibrated Maslow try drilling a hole near the top of the workspace and then punch the down button for 900mm and drill another, and report the result?
On my machine, they don’t seem to be correct in both directions – I can have the horizontal dimension correct, in which case the vertical is off, or the vertical correct, and the horizontal off.
I look forward to trying the new spreadsheet next week – Thank you all for your patience and suggestions.
A small memory on the control board that keeps it’s values during reboots and power cycles. Handy for storing configuration mostly constants.
Erasable Programmable Read Only Memory (actually it’s EEPROM, Electrically etc.) Yes, erasable and read only are contradictions but it’s part of the ROM, PROM, EPROM, EEPROM (and let’s not forget UVEPROM) historical sequence so the contradictoriness has historical significance.
. The geometry requires that a line between opposite mounting points pass through the center point, and that the hole spacing of the arm spanning them be the same as between the mounting points. Maybe I misunderstood?
- in the mean time why not try Step 10 and see what that does?