Delta steps per mm question

Hi,
Using the most recent firmware.
This is a very basic question that I thought I knew the answer to. I recently rebuilt one of my deltas with some new motors.
I set the steps/mm for the XYZ motors so a move from Z=100 to Z=200 moved the carriages exactly 100mm.
I ran autocal 3x and things looked ok. I then ran meshcomp (20 x 20 grid) and things looked ok.
I than ran a test print of a typical 20mm cube and got an 18.5 x 18.5 x 20.0.
I know that the armlength and deltaradius settings can affect this but the armlength is correct (the value came printed on the arms) and I thought that autocal took care of the deltaradius.
So is my memory of how to set the steps/mm wrong or is there something else at work.
Thanks.
Frederick

Did your tower steps per mm work out close to the theoretical value based on the belt tooth pitch and number of pulley teeth?

@dc42 said in Delta steps per mm question:
Did your tower steps per mm work out close to the theoretical value based on the belt tooth pitch and number of pulley teeth?
Yes assuming I'm doing the math correct.
The values that resulted in the carriage movements I measured were 80 for each X, Y and Z motors and those values seemed familiar.
The height of the cube is correct  does that tell us anything?
Can the autocal procedure determine the correct value for the deltaradius even if the initial value used in M665 is way off?
Thanks.
Frederick

Some additional info.
I placed a steel ruler on the bed and using the jog buttons of the DWC I verified that a 100mm on the X axis was a 100mm move according to the ruler.
So that would seem to confirm that steps/mm settings for the XYZ motors are correct.
So what else could make a cube that is supposed to be 20 x 20 be around 18.x by 18.x
Thanks.
Frederick

@fcwilt said in Delta steps per mm question:
So what else could make a cube that is supposed to be 20 x 20 be around 18.x by 18.x
Different bearing spacings at the top and bottom ends of the rods can do that.
Auto calibration can find the delta radius even if it is a long way off to begin with; although you may need to set the dive height quite high to start with so that all probes succeed without the nozzle crashing into the bed during the travel moves; and you may have to run delta calibration 2 or 3 times. because the effects of the corrections it makes are nonlinear when the corrections are large.

@dc42 said in Delta steps per mm question:
Different bearing spacings at the top and bottom ends of the rods can do that.
Auto calibration can find the delta radius even if it is a long way off to begin with
Thanks.
I can certainly check the spacing. I designed and printed new parts to match the spacing needed for your Smart Effector. If they are off it's going to be by a very small amount.
And the info about autocal is good to have.
Frederick