Have to enter shorter rod length than actual

Alexander Mundy

I had to enter a shorter rod length (279.504) than actual (288.00 to 288.05) to get XY dimensional accuracy. While it is printing spot on now (A 150mm hexagon printed with 3 sides perpendicular to the towers measures 150mm all 3 ways and a 150mm single wall circle is 150mm any angle I measure it), what math am I missing that would cause me to enter a rod length 8.5mm shorter than what it is?

The rods are mag ball rods that I got a few years ago from Ultibots and were pre-assembled. Smoothie gave me good XY dimensions with the average labeled dimensions (288.02) and best I can tell from a 300mm caliper they are accurate ball center to ball center.

Steps/mm are accurate (160 for my 0.9 steppers with 20 tooth G2 pulleys and 1/16 microstepping) and Z height is spot on so that rules that out.

Delta radius is currently set to 165.362 from 13 point auto delta calibration and it doesn't vary much from that figure with auto delta calibration (originally set to 165.25 which was my best manual measuring estimate).

Extrusion is spot on with 100mm filament taken in when extruding 100mm and a printed 0.5mm single wall is 0.0197" measured with a micrometer.

I'm using 1.17e (2017-02-10) firmware on a Duet WiFi.

dc42

Here are some possibilities:

1. Your belt steps/mm is 3% lower than you have configured. Inaccurate belt steps/mm is apparently quite common unless you have the belts really tight. You can check the steps/mm by printing a tall cuboid and checking the height.

2. You have some other geometrical error that is distorting prints. One possibility is unequal bearing spacings at top and bottom of a pair of rods.

Try this test. Print two parallel straight lines, passing either side of bed centre and half the bed radius away from it, each from one edge of the bed to the other. Check whether they come out parallel. Then reset the rod length in firmware to 288mm, recalibrate, and repeat the test.

If the second test gives the better parallel lines, you can use the M579 command to adjust the XY scaling. But it shouldn't be needed if everything is accurate.

Alexander Mundy

I'll give the parallel line test a shot tomorrow.

Thanks!

Alexander Mundy

Made a big tic tac toe grid and parallel test was bad.
Entered real rod lengths and parallel test was spot on but too small.
Manually probed center of bed then with a 3" micrometer calibration bar and height was dead on.
Printed a 200mm tall 20mm single wall cylinder and it was 200mm tall but 19.4mm Diameter.
Did the calculations for carriage movement with +-100mm X travel and as close as I could measure they were correct.
Went back to square one and started measuring everything and bam, there it was.
When I upgraded the steppers, metal corners, and electronics to Duet Wifi I printed new carriages.
I had changed my effector in the past to a stainless shapeways printed "Umbee" so I had 2 carriage files in my folder and I printed the wrong one which has the wrong ball centers!
Sometimes it's the simplest and most embarrassing things that make one bang their head against a wall.

@dc42:

Here are some possibilities:
.
.
2. You have some other geometrical error that is distorting prints. One possibility is unequal bearing spacings at top and bottom of a pair of rods.
.
.

dc42

I'm glad you sorted it!