Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: CNC-Design on May 17, 2013, 02:17:01 PM
-
Hello!
We just bough 3 sets of these drivers+AC servos.
http://www.artisantg.com/info/Oriental_Motor_VextaKXP_Manual.PDF
We have been having some accuracy issues with them.
It says 400steps/rev, so with our 1605 ballscrews it should be 400/5=80 steps per mm, but in order for it to scale properly we have to put 40 steps per mm in motor tuning.
It seems like either the motors react to both the falling steps and the rising steps? Or that we get some kind of double pulsing.
When we cut, holes also become a bit square. Engraving is pretty ugly, especially where the engraving changes direction. It seems a bit 'wiggly'. The motion of the motors can sometimes look a bit sluggish.
The motors also sometimes makes these short 'groaning' noises, when standing still. Is this normal with servos?
Pretty sure it's not the mechanics, as it worked with steppers before.
Could it have anything to do with interference? Seems strange because the errors always seem to occur on the same spot. Even if we rescale things, the errors are the same size.
Any help would be appreciated!
-
Not looked at the manual yet but encoders on servos are normally stated as line count or pulse per rev however the actual steps required will be 4 times that, so 400 line encoder would actually be seen by your drive as 1600 pulses per rev. Now if that were the case then if you have direct coupling then your steps per unit should be 1600/5 = 320.
As you are seem to be the other way I am wondering if you have an electronic gearing set in the drive. What that means is if the drive recieves one pulse from Mach it will tell the motor to move X amountof pulses. For example if you had 2x electronic gearing set each pulse from Mach would be seen by the drive as two pulses.
Hood
-
Thanks for the answer, Hood. We just discovered we can't wrap our heads around!
I can't believe we haven't noticed this before, but when you spin the ballscrew one revolution it goes 10mm! And we are using a ballscrew with a leadpitch of 5mm. Feeling pretty stupid now, but how can this be? Must be thinking wrong somehow.
So the driver was doing 1:1 steps, after all.
-
OK, feel even stupider now. It isn't even a 1605 ballscrew, it is a 1610 ballscrew with dual threads. Didn't even know those existed! Didn't notice the steeper slope when measuring.
It was pretty easy to see when we got a permanent marker and run it through the threads.
Going to get smaller pitch ballscrews now and that should fix things! :)
-
Ha ha these things happen :)
Hood
-
The problem that you are describing is why you should use the steps per calibration routine. Even if they say the screw pitch is 5mm, what if it is really 4.9mm??? When I set up my Bridgeport I did it using the math first and then checked with the calibration routine. It was off by a small amount. It is more accurate the second way.
Mike
-
Will definately check that when we get the new ballscrews, Mike! :)
Thanks
-
Calibration will only be as accurate as your measuring, calculating will be as accurate as your hardware. No harm in doing both but if the hardware is good quality then it will give better results in my opinion.
Hood
-
Hi hood,
I have seen several times where the specs for the parts in question were not perfect compared to the accuracy that we are looking for.
-
They must be very poor if they are ballscrews, especially if they are meant to be c25 or even c7 as there are tolerances they are supposed to adhere to.
Hood