PART TWO OF A SERIES

Methods and Application of In-Die Dimensional Measurement

By Bob Carabbio, Chief Engineer, Signature Technologies, Inc., Dallas, TX

Part one of this series appeared in the December 1998 issue of MetalForming. It dealt with parameters involved in obtaining in-process measurements and explored the various methods currently used to acquire reliable data. Part two focuses on the types of devices used to effect the various data acquisition methods.

Types of Measurement Devices

In the following examples, the part is shown in an orientation consistent with movement directly toward the reader. Reference surfaces are understood to be stationary and clamping apparatus, which would be required to hold the part in the reference position, is not shown.

Inductive Displacement Sensing

Inductive sensors measure displacement by responding to magnetic permeability of the area in front of the sensor. The presence of conductive material will cause an absorption of field energy in direct proportion to closeness of the conductive object. This allows the sensor to output an electrical analog signal related to displacement of the workpiece, Fig. 6.

Fig. 6--Inductive sensing.

This type of sensor is unaffected by nonconductive coatings and liquids such as drawing compound or stock lubrication. It is affected, however, by the homogeneity of the material and the material's magnetic history. This means that as the sensed material passes by the sensor, even though it may be flat and at a constant distance from the sensor, the output will show dimensional variations caused by localized differences in the material chemistry, and magnetic status.

An inductive type sensor also is "leaky" in that its sensitivity range is much larger than the sensor itself, so the immediate environment of the sensor must be controlled so that other machine elements don't move in and out of the field causing measurement inaccuracies. The material to be sensed also must be larger than the sensing field, although calibrations can be developed for smaller items.

If you are looking to measure to accuracies better than 0.005 in., direct inductive measurement generally won't do it because the inductive sensor will react differently to each individual part. One way around this is to use physical probing where the inductive sensor always looks at the same surface, which is positioned by a probe, Fig. 7. Of course, if you can't touch the material, or have to make quick measurements on fragile parts, this method is not suitable.

Fig. 7--Use of "telegraph" plunger for more accurate magnetic measurement.

Capacitive Displacement Sensing

If the workpiece is nonconductive, magnetic sensing won't work. However, capacitive sensing can operate with plastics. liquids, glass and other nonconductive materials. But, capacitive sensor units are sensitive to environment and contamination. Even the relative humidity of the surrounding air can affect the measurement. Because of this, capacitive sensing normally is disqualified for any type of in-process measurement.

Fig. 8--Capacitive threaded hole verification.

Capacitive displacement sensing can be applied in a free-standing measurement unit, or in manufacturing processes that are extremely clean and operate in controlled environments, Fig. 8.

Fiber Optic Displacement Sensing

Fiber optic displacement sensing units potentially are the most sensitive. Resolutions down to 1 angstrom (0.0000001 millimeter - 0.004 microinches) are available. The unit uses a fiber optic bundle to return light from the target in a manner proportional to the distance from the target. It is available either in a reflectivity compensated form where the material passing by has a variable surface finish, color or reflectivity; or in a noncompensated unit where it always is looking at the same reflectivity, or the same object. The stand-off distance (distance between the measuring head and the workpiece) is quite small in higher accuracy versions.

Fiber optic displacement units will be confused by liquids on the workpiece--measuring to the surface of the liquid. They will be disabled by dirt and oil collecting on the sensor, so are only useful in clean environments. There is an exception--if stationary contact with the material is possible, there are ruby-tipped units that are sealed, and can measure by direct contact with the workpiece.

Linear Variable Differential Transformer Sensing

The linear variable differential transformer (LVDT), is an electromechanical device that features a triple coil assembly with a moveable core piece. The position of the core inside the LVDT controls the level of the unit's output. In the "gauge head" version of the LVDT, the core is connected to a probe assembly that can be used to contact the workpiece in the area where a measurement is desired. Gauge head units are available with either a spring-loaded plunger or a pneumatically actuated one. If the pneumatic unit is used, the probe can be controlled by a valve, and deployed at points in the machine cycle when the workpiece is positioned for measurement (this method is valid for slower speed presses, up to about 100 to 150 spm).

Fig. 9--Laser triangulation sensing.

Laser Triangulation Sensing

This is done by "bouncing" a laser beam off the target and picking it up on an array of photocells in the laser unit. The position at which the beam hits the photocell array is proportional to the distance between the laser and the target, Fig. 9.

The laser unit must be kept free of diffractive material such as clear liquids or oils. This is done best by encasing the laser unit in an enclosure that is filled with low-pressure clean, dry compressed air. The laser will respond to liquids on the surface of the workpiece.

Laser Interruption Sensing

This method operates by aiming a laser/receiver pair so that the laser beam to the receiver is interrupted by the workpiece in a way that is directly related to the parameter to be measured. In other words, the laser beam should be oriented so that the measured quantity varies at 90 deg. to the beam axis. The output from these units is an electrical analog proportional to the area of the beam that reaches the receiver. This results in a nonlinear, roughly sinusoidal response, which is relatively linear around the 50-percent beam blockage point.

Fig. 10--Beam interruption laser with aperture plate.

The laser unit must be kept free of diffractive material such as clear liquids or oils. Again, this can be accomplished by encasing the laser unit in an enclosure filled with low-pressure clean, dry compressed air, Fig. 10.

This type of measurement is potentially quite simple to do. The laser and its receiver can be located outside the tooling and all that is needed is an optical path so that the laser beam can be interrupted by the part feature that is being measured. Receiver alignment is relatively unimportant as long as the beam from the laser strikes its sensitive area. The critical alignment requirement is in the aiming of the laser so that its beam is masked by the workpiece feature.

In cases where the laser is a round-beam type, the beam should be aimed so that it is approximately 50 percent interrupted when the dimension is at its nominal value. Round beam lasers have a nonlinear output over the full range, but are fairly linear around the 50 percent point, which also is the point of greatest sensitivity.

Fig. 11--A round aperture.

Fig. 12--A slit aperture.

The measurement resolution can be enhanced by using an aperture plate between the laser emitter and the part to be measured, and increasing the gain of the laser receiver to compensate for the lost beam area. The simplest aperture is a round hole that typically can be as small as 0.020 in. and still allow full range on 0.039 in. beam lasers, Fig. 11. A slit 0.010 in. wide and 0.04 in. long also can be used to allow a wider linear range of measurement, Fig. 12.

Where the laser is a "sheet beam" type, larger ranges of position can be measured at a lower resolution by orienting the beam with its long axis in the direction that the variation occurs. Alternately, a wide object can be given an "average" measurement by orienting the plane of the beam parallel to the edge to be measured.

Fig. 13--Differential interruption laser measurement.

In the same fashion, the measurement also can be made in differential fashion, using two laser heads--one on each end of the part--so that small differences in the position of the part can be cancelled out of the measurement, Fig. 13. In this case, with the part (or gauge piece) at nominal value, both lasers must be adjusted for 1Ž2 scale. This is done by adjusting the laser gain for full scale with the beam unblocked, then inserting the part (or gauge piece), and physically adjusting the aim of the laser sources so the signal levels fall to 50 percent of full scale.

Scanning Laser Sensing

A more advanced version of the interruption laser is the scanning laser device. In this device a laser beam is scanned across the workpiece, and the edges that interrupt the beam are recognized as measurement points, which then either are reported digitally or fed into a data-acquisition system via analog signals.

Fig. 14--Scanning laser measurement.

In the illustration (Fig. 14), the letters "A" through "E" are significant points where "A" is the beam origin. The acquisition system then can post-process the information it receives to allow the creation of values for:

"B--E" (which is overall part width),

"C--D" (which is the dimension of the round hole in the part). Since the laser beam scans the hole as it passes, the system will "profile" the hole. This will, when considered along with the feed progression, give the longitudinal location, and diametric measurement of the hole.

"B--C" (which is the dimension from the hole edge to the part edge.

Typical scan widths are 40 mm (1.575 in.) and 120 mm (4.72 in.). The narrower units are more accurate.

The scan rate is 1.2 Khz, which is not super fast, but generally adequate at medium press speeds. The scanning lasers get more accurate if multiple scans are averaged, but will have a base resolution of 0.010 mm (0.000394 in.) with a single scan, yielding a useable accuracy of 0.001 in. for the 40 mm unit.

Pneumatic Differential Sensing

Where there's a lot of contamination, and particulate matter likely to be in the measurement area, air pressure sensing is an alternative, Fig. 15. While among the highest cost alternatives, it's also the most robust, having no moving parts or electronics in the measurement site.

It's not the fastest, however, requiring as much as 0.3 seconds under worst-case conditions to stabilize a measurement from a large step change. Once the workpiece is in position, measurements can be taken at a 0.020 second rate. Readings are accurate to 0.001 in. Readings from the pressure sensors--being highly nonlinear--require extensive post-processing to arrive at a linear indication.

Inferential Sensing

In cases where there is no room to implant sensors, or where the constraints of the tooling do not permit access to the part, the measurement sometimes can be inferential. In other words, the dimension is measured by an indirect method that returns an electrical analog signal, which varies as the desired dimension varies, but may be based upon force or pressure, and also may exhibit a nonlinear but definable relationship to the desired measurement.

Inferential Dimension--In the example shown in Fig. 16, a push-on terminal is being produced. The clearance between the ends of the curl on the sides of the female terminal and the base surface of the terminal is an important parameter.

Since the part can be held down against a reference surface, and there are narrow slots in the back of the connector, we can insert narrow blades through the back of the connector and "infer" the dimension of the curls by the force they apply to the blade tips.

There is an adjustment provided for each of the fingers so that they can be adjusted to a known value at the nominal dimension through the use of a gauge block. Insufficient curl clearance is indicated by too much force being registered on the finger, while too much clearance results in too low a force.

Fig. 17--Load cell punch mount.

Inferential Tool Condition--Another variable that can be measured by inference is burring caused by loss of punch sharpness, or wear. Fig. 17 illustrates a load cell/ball-lock punch mount that would provide bipolar information about the operation of the tooling.

What can be derived from the punch signature, while not quantitative in the direct sense, will indicate whether the tooling is operating normally or whether:

1) The cutting edges have become worn, and are causing burring on the punched hole.

2) The surface finish of the punch is degrading.

3) There is a lack of adequate lubrication being applied to the material.

4) There is a jam in the slug channel.

5) The punch has broken or chipped.

Experience has shown that automatically detectable changes will occur in the signature due to tooling wear long before the tool has degraded to the point where burring gets to an objectionable level.

Fig. 18--Analog proximity hole detection.

Inferential Punch Confirmation--The presence or absence of holes in a workpiece can be determined with a magnetic pickup, which responds not only to the closeness of the sensed material, but also to the amount of material in the sense field, Fig. 18. Missing holes in the workpiece or embedded slugs mean there is more material there, and will register as a higher output from the device.

The part must be presented to the analog proximity sensor consistently so that the only variable is the presence or absence of holes in the workpiece. The hole should be sized so the change in signal from the sensor is at least 10 percent of the total measurement range in cases where the hole is not present. This can be tested by presenting the part to the sensor with and without the holes and noting the difference in response. The smaller the proximity device, the smaller the hole that can be detected. As has been stated before, analog proximity sensors are sensitive also to material consistency, and a small variation in measurement can be expected from identical parts.

Part Three

The third installment will appear in the February 1999 issue of MetalForming magazine. It will discuss closing the loop and using the test bench, as well as present material showing how sensors are being used for in-die measurement and data acquisition in the field. MF

Copyright 1999 by Signature Technologies, Inc., Dallas, TX. Used with permission. All rights reserved.