TPI setting - Steps Per Inch Discrepancy

I will do my best to make this as logically sound as possible.

My machine runs on leadscrews which couple directly to a motor which when turning drives the machine on a linear rail. There are no belts or the like to the machine.

Based upon my understanding to calculate the TPI one first measures the threads per inch which in my case is 10. I confirmed this by measurement and manufacturers specs. It is a 5 start lead screw. Thus based upon my past understanding one would take 10/5= 2 turns per inch.

In the Pncconfig when setting up an axis (motor/encoder configuration) I click calculate scale. Initially I entered 2 for leadscrew TPI. This gave 400 steps per unit. It also seemed a rather small figure. In testing/tuning the axis with the recommended settings for my drivers I never could make a 1 inch move equal 1 inch in testing nor after homing and then manually jogging the machine.

I then went and changed the setting within the Pncconfig (motor/encoder configuration) to equal 10 for the leadscrew TPI (edit: which gave me 2000 steps). When testing/tuning the motor it would move exactly 1 inch. After starting Linuxcnc and homing, then manually jogging the machine 1 inch was not one inch on the physical machine. It was clearly off.

I adjusted my stepper drives microstepping and I was able to make, for example a +1 inch move on the x axis within linuxcnc was at last exactly equal to one inch on the physical machine. I went back into Pncconfig to test this new setting and within the test/tune axis when set to move one inch it would move nearly 2 inches. So within testing this setting is off despite while running linuxcnc the settings measure out properly.

My question is this. Am I misunderstanding this process of setting the leadscrew TPI or is there some discrepancy between the test/tune axis and the physical output within Linuxcnc when it is actually running the machine? I cannot help but feel like I am missing something somewhere.

Any advice of where to send me hunting and pecking would be appreciated unless I am worried about nothing.

it's of course possible pncconf's test/tune is not getting the same steps per unit as you entered on the motor/encoder page. What version of linuxcnc are you using - ill check into it.

a 5 start 10 tpi lead screw is indeed equivalent to a 2 tpi thread.

with a 200 step stepper and no micro stepping that would be 400 steps per inch. ( 200x2tpi)

with 5 times micro stepping that would be 2000 steps per inch ( 200 x 5 x 2tpi)

In the end if linuxcnc is moving the right distance that is all that matters - go with your experimentally found numbers.

Chris M

I am using Master 2.8. Thank you for verifying the numbers. When scratching my head seeing TPI in the pncconf I knew it had to mean turns per inch and not threads per inch lol. Threads per inch would leave way to many variables in play and couldnt be a constant without other calculations.

Like you said as long as it works that is what matters. I believe what happened in my calculation is I forgot to account for the microsteps thus using 10 TPI I in the end arrived at the right value just the wrong way of getting there.

Nevertheless the pncconf is giving about 2x the amount of linear motion in testing than when actually running LinuxCnc and moving via gcode. Again one variable at play in this is that my TPI in the setup is 10, not 2 and now I am wondering if that plays into it. Could it be that there is some differences in how the testing is being executed based on the values than when the software is running?

Either way.... it works in the software and thats what matters. Yet I will play with it as well and see if I can make them both agree. The present difference between where I am at and mathematically where I should be, if I remember correctly is about 400 microsteps. This will be a quirk to play with and figure out.