Introduction
In today’s manufacturing environment, it is more and more commonplace to utilize laser measurement systems for the calibration of CNC machines.  After all, companies routinely calibrate the gages they use for verifying dimensional information so why not go ahead and do the same for the equipment that create the dimensions?  The benefits of a preventive maintenance program that includes the calibration of CNC machines are many.  Reduced scrap, reduced rework at the detail and assembly level, and increased shop floor flexibility are just a few.  Add to this the ability to know if a machine is adequate to manufacture a given part before putting it on the machine (or getting the work) and it all adds up to give you significant insight into the manufacturing capabilities of an organization. 

However, there is a problem with the process.  Today’s shops have a myriad of machines with many different types of controls that have different capabilities, communication methods, and varying approaches to compensation.  These variables combined with today’s leaner and meaner maintenance organizations present quite a challenge to those who need to utilize the electronic compensation inherent in the controls.  Maintenance groups across the country (and world) struggle with the issue of compensation. 

Not too long ago, I was working for a rather well known laser manufacturer and had the unique opportunity to travel all throughout North America talking to end users of the equipment.  By and large, the most common question was, well the laser is great, but, "How do you comp this machine?"  The more I traveled, the more I heard the same question.  I even had conversations with machine tool control manufacturers of many varieties and explained to them that the great compensation they were making available in their controls was fantastic but that unfortunately, nobody knew how to take advantage of it – it just wasn’t clear.

In this article, we will seek to get some foundational theory underneath the various types of compensation and further attempt to give examples of its formulation.  For simplicity sake, we will focus on linear compensation.  Future articles will begin to deal with aspects of straightness and squareness compensation.  Ultimately, we will touch on the aspects of volumetric compensation. However, first things first:  let’s start with linear.

Background
Before we dive in, please let me say that it is not my intent to endorse any specific type of compensation in this series of article or further more any brand of control.  As will become clear in future articles (and most of you likely already know this), they all have their strengths.

There are two basic types of compensation available in today’s controls:  Absolute and Incremental.  Incremental compensation is typically found in Japanese controls and Absolute compensation is found in most others.  As the details of these types of compensation are investigated, it may be seen that there are advantages in each of the approaches.

Absolute Compensation
With absolute compensation, individual points stand alone.  For the most part, the value entered into the control is the opposite value of the error observed, however, this depends on the control.  Consider the error profile depicted below where the first column represents the measured point and the second column the measured error.  This is typical of the linear data obtained from a laser measurement system:

The resulting compensation for an absolute control is then:

Based on the absolute model, it should then be expected that the graphs of the error versus the compensation to show an inversion from one another as shown below:

Graph of the error versus the absolute compensation.

Incremental Compensation
Basically, incremental compensation operates under the premise that each point influences the other.  That is to say, changes in a previous point (from a given point) will influence all points from there on to the end of the string.  While initially this may seem somewhat cumbersome, it can be quite useful.  In situations where some relatively modest scaling may be seen, the modification of the first compensation value will be felt through the entire curve.

When working with incremental compensation, the values are additive from one point to another.  Going back to the same error profile mentioned in the discussion of the absolute compensation model, the differences may be seen.  Again, the first column represents the point number with the second column representing the associated error as measured by a laser.

The first point would be -.002 (to get to zero), the second, another -.002, the third, .002 and so on.  Note here that the compensation value is the change from one point to another as opposed to the absolute value as explained earlier in the absolute compensation discussion.  The final compensation table would then look like:

To further illustrate, review the graph shown below.

Graph of the error versus the incremental compensation.

To compare the two methods simultaneously, please see the graph shown below.  This shows the overall error, the absolute compensation, and the incremental compensation graphs overlaid.

Graph comparing absolute and incremental compensation.

Exercises
A sample set of exercises has been included (along with answers) with this article so that this may be practiced. 

Summary
Certainly it is said by many users of controls, laser, and associated software that the absolute compensation method makes more intuitive sense.  As mentioned before, this type of compensation is seen on mostly non-Japanese controls however the reality is that, once understood, both compensation approaches offer unique abilities.  Either way, with today’s shops, one is bound to find both types of controls.  Learning to manipulate both types of compensation can not only extend the life of a given machine but also ultimately result in improved manufacturing flexibility, accuracy, and quality.

Sample Exercises
The graph below formulates compensation values based upon both incremental and absolute controls.  Note the dimensional data is given in millimeters.

Bi-directional Linear Error Plot

 

 

 

 

 

 

 

 

 

 

The Raw Data:

Position	Forward Error
0	-0.000991
25.4	0.000103
50.8	0.000484
76.2	0.000151
101.6	0.002265
127	0.00063
152.4	0.001902
177.8	0.002967
203.2	0.003343
228.6	0.00499
254	0.004481
279.4	0.008745
304.8	0.00625
330.2	0.008284
355.6	0.00981
381	0.011361
406.4	0.013371
431.8	0.011046
457.2	0.012257
482.6	0.012499
508	0.012693
533.4	0.015842
558.8	0.015212
584.2	0.017441
609.6	0.017538
635	0.019621
660.4	0.020493
685.8	0.022915
711.2	0.024853
736.6	0.023836
762	0.02495

To begin your exercise, fill in the compensation values.  For convenience, values have been rounded to the nearest micron.  First, assume an incremental compensation approach.