Axis Reversal Errors

Reversal Errors are a loss of motion when an axis changes directions.  It is also referred to as lost motion.

There are a number of things that cause reversal errors.   These things fall into two main categories which tend to overlap each other.   One category is backlash, the other is machine hysteresis.

Backlash

Backlash coming from a drive train could be defined as the amount of free play or clearance between the components of the drive train.   The "deadband" that occurs when the direction of motion is reversed.

Imagine putting two nuts and a washer on a bolt.  Put the washer in between the two nuts.   Leave the nuts backed off from the washer so that the washer can be moved back and forth 1/8".   Hold the washer in between your fingers so that it is the stationary part.   Push the bolt toward the washer until one nut contacts the washer.   There is now an 1/8" gap on the opposite side of the washer between it and the other nut.   Reverse directions on the bolt.   The bolt will not try and move the washer until that 1/8" gap is closed.   This would be 1/8" backlash on a machine axis if the washer where the axis and the bolt the drive train.   A machine with a leadscrew, ballscrew, or a gear type drive train assembly would fall into that category.   Backlash in a drive train is the result of things like loose bolts, gears, or a worn out ballscrew assembly, etc.

Measuring For Reversal Errors

Normally, measuring for reversal errors can easily be performed with a dial indicator setup on a magnetic base.   It is also one of the errors collected when checking for linear displacement errors with an Interferometer Laser or collecting data with a Ballbar.   One would measure for reversal errors by moving the machine axis in one direction a few inches or so, then setup and zero the indicator to read inline with the axis direction of travel and to measure from the axis to a stationary part of the machine.   With the CNC Control, input the command for the axis to move 0.001" in the reverse direction from the original direction.   If the indicator still says zero then so far we have 0.001" backlash.   If we moved the axis 0.010" and the indicator only moved 0.002" off from zero then that axis has 0.008" backlash.   That's the general idea.  It doesn't have to be performed exactly like that as long as one measures the error upon the reversal.   Try that at different feedrates too!  You might get different results due to stiction.

Angular Deviations

Reversal errors also show up from angular deviations such as Yaw when the point of measurement is offset from the pivoting point of the angular change.   This angular change can come from a loose guideway system.  This is sometimes referred to as skewing.   In the case of skewing, this also occurs mainly upon a reversal of direction and the amount of reversal error will be proportionate to the distance the measurement is taken offset from the pivoting point.   The further away, the greater the error.   In other words, if one where to measure for reversal error at the pivot point of the angular change, one probably would not see any error.   On the other hand, if the measurement is taken offset two feet away from the pivot point and the resultant error is 0.005" then at four feet away it will be 0.010"   Angular deviations not only occur from looseness, they also occur due to machine hysteresis.

Machine Hysteresis

Machine hysteresis concerns the rigidity of the machine when the structure of the machine is put under a load.   Suppose we were to setup and zero an indicator on a small machine to measure for reversal error then push the axis by hand away from the indicator gradually letting up on the pressure until there is no pressure.   If that axis has no lost motion due to backlash and the indicator did not return to zero then the amount the indicator is off from zero is due to machine hysteresis.   The point is, just moving an axis in opposite directions can cause a reversal error due to hysteresis because it takes opposing force to do that.   As a side note, be careful with the indicator setup because it has hysteresis too.   You can check it the same way as the axis.

Gravity

Everything on earth has a huge force applied to it all the time.  That force is gravity.   Is is easier to go down than it is to go up because of it, isn't it?  Of course that applies to vertical axes too!   Reversal errors do show up with vertical axes due to the effect of gravity.  How much error depends upon a number of factors.   One of the factors is the method used to compensate for gravity.  That factor is the vertical axes counterbalance system.

Counterbalancing a vertical axis to reduce reversal error is a very tricky business because no method to date is perfect.   The heavier the axis the more complicated the effort.   As an example, let's say the machine has a hydraulic counterbalance system.   The type of system we are referring to has two pressure settings.   One setting is the moving upward and holding pressure setting either coming from a pressure compensating pump adjustment or a reducing valve.   The the other setting is the downword pressure setting coming from a relief valve.

The downward pressure setting has to be higher than the upward pressure due to the nature of the refief valve.   The relief valve provides a path for the hydraulic fluid to excape back to the tank.   If the relief valve wasn't there, the axis would not be able to go down because there wouldn't be anywhere for the fluid which filled the cylinder as the axis went up to go.   The relief valve must have a higher setting than the upward pressure or it will constantly relieve fluid creating fluid friction due to flow under pressure over the valve.   This in turn generates heat which reduces the lifespan of the hydraulic fluid and components and wastes energy at the same time.   This difference in pressure causes forces or loads which cannot exactly balance for gravity because gravity is one constant value not two.   That load imposed against the machine structure causes reversal errors due to hysteresis.

Actual cases of this involving two fairly large CNC Machines are, one with 0.002" and the other with 0.003" reversal errors.   Both manufacturers of those machines said, there is nothing that can be done about it, it is the normal amount for that machine.

Position Feedback Devices

With the advent of the CNC came positioning feedback devices.   Basically, these devices feedback the axes positions to the CNC Controller and the controller adjusts those as dictated by the part program.   With modern CNC Machines these are typically encoders.   There are linear encoders sometimes referred to as scales and there are rotary encoders.   What does this have to do with reversal errors?

How those effect reversal errors depends upon the type of encoder used and how it is attached to the machine.   Take a linear axis for example.  A linear axis just travels straight back and forth.   Commonly, this linear motion is generated by a rotary motion in which a servo motor turns a ballscrew.   It may have a gear train between the motor and ballscrew or not.   If the position feedback device for that axis is a rotary encoder in the motor and there is any looseness anywhere in that drive train between that encoder on one extreme end and the axis on the other end then the CNC Controller can send out commands unaware of the backlash.   The backlash will eat away at the tolorance of the part being machined from the axis coming up short of the targeted position every time the axis reverses directions.

On the other hand, if the axis uses a linear encoder for a position feedback device the CNC Controller may still be completely unaware of the backlash in the drive train but not come up short upon axes reversals at all.   This is because of the way that a linear encoder is attached to the machine.   The scale part of the encoder is bolted to the stationary part of the machine and the moving part that is reading the scale is attached to the moving part of the axis.   It is not dependant upon the drive train's integrity at all because it is not attached to the drive train as in the case of the rotary encoder.   The only amount of reversal error that will show up in the case of the scale comes from either hysteresis or the amount of one resolution of the scale.   Modern scales can have a resolution of one micron or less.  This leads us to the next subject, backlash compensation.

Backlash Compensation

Backlash compensation is a feature of many modern day CNC Controllers.   It simply involves measuring the reversal error then inputting the appropriate value for the error into the control's parameter for backlash compensation.   From then on the CNC Controller will not include that lost motion as a position change as far as the part being machined is concerned.

This is a great feature but also can be a bad thing if used inappropriately.   It can be a bad thing if the reversal error comes not from machine hysteresis which we may not be able to do anything to correct but rather from backlash in the drive train or a loose guideway system.   Loose parts should be tightened or the tendency for the loose part will be to keep loosening from the effect of the loose parts rubbing and bumping into each other.

The next thing to consider is, how much reversal error do we want to compensate for before we decide we should look into the cause?   That's a tough question to answer.  Consider these questions.  Is the cause of the reversal error damaging the machine?  Is the machine making good parts?

That's for you to decide!
Admin~

This is a printer friendly website.
Copyright © 2007-2009 CNCexpo.com,  All Rights Reserved