Clearpath SDSK steps per inch

Hi, I purchased a machine that was running Mach 3 and after converting my main machine to kflop/kanalog felt like playing with LinuxCNC on this one.

I have a 7i76e.

These motors are capable of accepting quadrature or step/dir input and are set for 6400 pulses per revolution (even though they can only actually position to 800 points per revolution). I have them set up/wired for quadrature input and the mesa stepgen type set to type 2. In the previous Mach 3 config which was using the motors in step/dir input mode, the steps per inch were set at 32512 for a 5mm pitch ball screw. In pncconf this is what I entered.

It was WAY off. 19" of machine travel was read as some fraction of that on the DRO's. I divided by 4 and that got me close but I don't understand why and it is still not quite right. 

Can you share your ini file? Sounds like you may have inadvertantly multiplied your steps per inch by the ball screw pitch or something,

The scale should be 130048 steps per inch
(this will be called STEP_SCALE in the .ini file)

Note that in mode 2 you should set all the step timings short
(say 100,100,100,100 ns)

Also AFAIKR Clearpath suggests not using differential wiring
so connect the 7I76E Step/dir connector +5 to the Clearpath
+ inputs and the 7I76E STEP- and DIR- to the clearpath -inputs
and leave the 7I76E +STEP/ and +DIR pins unconnected

I re-wired the connections for step/dir and changed the input mode on the motor this morning. Now all is good using the same step_scale, I e-mailed teknic about using quadrature mode and will see what they say as I am curious. Since it makes no functional difference for my uses I will just wire the rest up as step/dir. Thanks for the responses!

For anyone who uses ClearPath servos with Mesa Anything IO cards in quadrature mode ...

I've verified with a Teknic engineer that the ClearPath motors count 4 for every complete quadrature cycle - the rising and falling edges of A and B.

Since one step in Mesa stepgen is one complete quadrature cycle. So if the encoder resolution is 6,400/rev (for the enhanced version of the servos) set the step resolution in MSP to 4x6,400 = 24,800, so that every 4 counts (called 'quad count' in MSP) rotates the motor one encoder tick.

The regular version has encoder resolution of 800/rev.

This way, in the math to calculate scale use the servo encoder resolution as the steps/rev. And each Mesa stepgen count equals one motor step rotation. I think this is way easier to visualize and understand. If you run into max rpm issues on the motors, just set the steps/rev to the maximum without rpm limits (say 8000) and in the step/rev on linuxCNC side enter 2000. Tested this on actual hardware and it seems to work correctly.