Вы находитесь на странице: 1из 5

An automatic comparator for slip gauge calibration

P.J.E. Aldred*
The manufacture of slip gauges is a highly skilled process, the end results of which are batches of gauges which must be measured and, depending on their deviations from nominal size, assigned to the correct British or D IN standard grade. There are altogether ten of these grades for metric gauges and eight for imperial gauges. It is particularly important that the calibration process should be capable of identifying those gauges which meet the requirements of the highest grades since high grade gauges command a premium price. Present calibration methods using mechanical comparators are, however, barely capable of meeting the accuracy required since, for the highest grade, the tolerances on length, flatness and parallelism are 50 nm. TI Research Laboratories have therefore developed, on behalf of Coventry Gauge Limited, the instrument which is described in this paper, with the prime objective of improving the accuracy of post manufacture calibration of slip gauges. An operating range which covers the range of size of metric slip gauges from 1-100 mm and imperial gauges from 0.01 - 4 in and takes into account the difference in cross-sectional area between small gauges and large gauges.

As well as meeting these basic criteria, an automated instrument designed to calibrate gauges after production must measure gauges sufficiently quickly that the calibration throughput matches or exceeds the rate of production of gauges.

I nstrument design
General considerations

Measurement requirements
In order to classify a gauge into the correct grade, its length, the parallelism of its measuring faces and the flatness of each measuring face must be measured. The length is defined (BS 4311) as the distance between the centre of oqe measuring face and a surface onto which the gauge has been wrung for measurement. Parallelism error is defined as the variation in length over the measuring face and the flatness error of each face is the separation between two parallel planes, within which the face can just be enclosed. In practice, measurement at the centre and four corners of each measuring face (positions 1A 5A, 1 B - 5B in Fig 1) permits calculation of these errors. Metric gauges are assigned to one of ten British or DIN standard grades and imperial gauges to one of eight British or American standard grades (see Table 1). The tolerances for small gauges are tighter than those for large gauges, the range of metric gauges being divided into thirteen groups and imperial gauges into five groups. Each group has different tolerances and, of course, for any gauge the appropriate tolerances must be used for classification. Any instrument designed to make the measurements on which the grading of slip gauges is based must satisfy three important requirements: High measuring accuracy on gauge length, parallelism and flatness. The length tolerance on the highest grade (00) in BS 4311 is -+50 nm. Ideally, in order to achieve a good grading efficiency, the instrument should measure to within -+5 nm. Lack of damage to the high quality finish of the measuring faces of the gauges. This is essential since even a narrow scratch (1-2/Jm wide), which may not affect the accuracy of the slip gauge, will show clearly against the background finish and is commercially unacceptable.
*Physics Department, TI Research Laboratories, Hinxton Hall, Hinx ton, Saffron Walden, Essex CB 10 1RH, UK

The initial design consideration is whether the measurement should be absolute or comparative. For this instrument the decision is straightforward since the advantage lies so clearly with comparative measurement and its less stringent requirements for thermal stability and long term mechanical stability. Thermal drift is an obvious problem which must be tackled in the early stages of design. The approach we have adopted is to build large thermal inertia into the machine and to ensure a minimal disturbance of the air surrounding the measuring region. The machine (see Fig 2) has dimensions of approximately 1 x 1 x 2 m, weighs 1 tonnes and is installed in a room with close control of both amplitude and rate of change in temperature. In addition, the gauges must be left for some time after handling to reach thermal equilibrium with their surroundings. The general organisation of the comparator is therefore that each set of gauges is loaded into a cassette together with a previously calibrated master gauge of the same nominal size. The cassettes are

/
I I 02A

~,~';, 2B (-~le
I ,.:, .~_ _ _ 3_B_ _ _

/ 4AO _ ___5_B~L_j

- -

///

01A---

I /
5AO ~ / "

Fig 1 Spatial arrangement o f measured points

Table 1 Classification of slip gauges


Number of different Number of Number of nominal sizes groups grades Metric gauges 184 13 5 10(BS+DIN) 8(BS+American)

Imperial 134 gauges

P R E C I S I O N ENG I N EER I NG

0141-6359/81/020109-05 $02.00 1981 IPC Business Press

1 09

signs, then the flatness error is the sum of their magnitudes. In practice, the measuring heads are more complex since extra probes are included so that gauges with both small and large cross-sectional area can be measured. The probes are steel rods inserted into holes in the measuring head but insulated from the measuring head by a thin film of epoxy adhesive. During measurement the rods are at the same electrical potential as the measuring head and the latter then acts as a guard ring to eliminate fringing of the electric field between the gauge and the probes. The insulating film is kept sufficiently thin to ensure that the probes have the required linearity over their measuring range. As well as the probes, the lower measuring head incorporates three gauge support rods on which the slip gauge rests while being measured.

Gauge transfer
The choice of how large a gap should be left between the gauge and the measuring head surface raises a conflict between the metrological requirement to keep the gap small (therefore reducing the capacitive impedance and the required resolution) and the practical requirement that gauges must be inserted between the measuring heads. The metrological constraint is, however, of dominant importance and a working gap of 20#m has been chosen. Clearly, gauges cannot easily be inserted between measuring heads with a clearance of only 20/]m at each end without considerable risk of damage to the surfaces and therefore the heads must be moved apart during gauge insertion. There are in fact two requirements here: The separation of the heads must be adjusted to suit the size of gauge being measured. This adjustment must be to a precision of a few microns in order to ensure that the gap lies within the range of the probes but need be no more precise than this since the gap is calibrated by measurement of the master gauge. The heads must be separated and then replaced to allow insertion of each gauge in the cassette. The repeatability requirement here is stringent since any variation results in a corresponding calibration error on the gauge.

Fig 2 General view of the Comparator showing storage racks and the measuring zone (centre front)
then placed on one end of a storage rack and a walkingbeam mechanism brings them in turn to the measuring position. The capacity of the storage rack is such that the gauges can come to thermal equilibrium while awaiting measurement. Each cassette, when it reaches the measuring position, passes through the measuring heads so that the master gauge and then the other gauges can be measured. A second storage rack provides a buffer store for measured gauges. The input and output storage racks can be seen at the left and right hand sides of Fig 2, with the measuring position at centre front. The design uncertainty of measurement of the comparator is 5 nm standard deviation on length comparison, with an average throughput rate of one gauge every ten seconds. This measurement uncertainty gives the maximum advantage in grading accurately. Any lower uncertainty in comparison would not give any real advantage since other uncertainties, such as those of the absolute calibration of the master gaugesI or of contact errors when gauges are wrung together z , would become dominant.

The two techniques developed to achieve these two requirements are illustrated i n F i g 3. The upper measuring head is mounted on a ram which can be raised and lowered hydraulically to adjust the separation of the measuring heads to suit the size of gauge being measured. The lower measuring head

Height setting sensors : Capacitive Opt ical

~
[ 1
~ / J . . . . ~

NOddlng t)ead mechanism MeasUrlr~ heods

Transducers
In order to minimise the risk of damage to the measuring faces of the gauges, non-contact capacitive transducers are used. The gauge being measured is positioned between two measuring heads, each of which incorporates five probes to take measurements at the position 1 - 5 indicated on Fig 1. The calculations for length and parallelism errors are straightforward. For flatness errors, readings from three of the corner probes are used to define a plane which is coincident with the surface of the gauge at these positions. Predicted readings at the fourth corner and the centre are calculated and the differences between actual readings and the calculated values are tile distances separating the gauge surface at these points from the plane. If these are both of the same sign, then the flatness error is the larger of the two, and if they are of opposite
Beam p vot ~ . ~

C ssette

m hftlng c y h n d e r

A n n u l a r ciarnl

Probe layout

Fig 3 Sectional view through the measuring zone

110

PRECISION ENGINEERING

measuringl head measuring head

Ram

head is lowered and the gauge can be measured. The primary power source for all the mechanisms is pneumatic, but in order to minimise the generation of heat in the region of the measuring heads, the actual mechanisms are all operated hydraulically via air/oil converters mounted remote from the measuring heads. A general view of the measuring zone of the comparator is shown in Fig 6. The cassette is approaching the measuring position and the lower measuring head can be seen inset in the beam. In the background are the :ontrol console and the desk top computer used for data processing.

Control Fig 4 Schematic view of the beam and ram, showing a gauge at the measuring position
There are a number of mechanisms in this instrument which must act in the correct sequence to avoid damage to slip gauges or the measuring heads. Although a desk top computer has been incorporated for data processing, it is not used for machine control which is handled instead by a purpose-built hard-wired control system. This approach eliminates the risk of inadvertent corruption of control by programming modifications. The control philosophy is that each step of the control sequence is only initiated when a signal from a sensor has indicated the completion of the previous action. Most of these sensors are optical source/detector pairs with the led sources run at low power to minimise heat generation. One exception is the sensor which detects whether the upper measuring head has raised. Here, a microswitch is needed since the sensor is in close proximity to the collapsing strut mechanism and no heat source could be tolerated. For the same reason, the second stage height setting detector (Fig 3) mounted adjacent to the measuring probes is capacitive. Where necessary, fail-safe techniques have been used in order to ensure that in the event of failure of a sensor, no damage will result. An example of this is in height setting where the ram is initially adjusted using an optical sensor mounted on the outside of the ram. This allows full travel of the ram without risk of mechanical contact with the gauge. The ram is clamped while moving upwards so that any delay in clamping will leave extra, rather than less, clearance. The second stage, precise height setting using the capacitive sensor also involves raising the ram

LFiexure hinge

.\

Measuring head

Fig 5 "Nodding" mechanism for raising the upper measuring head


is in a fixed position and the gauges are brought up to the measuring position by advancing the cassette along a beam which passes through slots in the side of the ram (see also Fig 4). When the upper measuring head is at the correct height, as indicated by sensors which detect the upper surface of the slip gauge, the ram is clamped to the cylinder by applying hydraulic pressure to the annular clamp galleries in the cylinder liner. A second mechanism, shown in more detail in Fig 5, is used to raise and lower the upper measuring head for gauge insertion. The measuring head is supported on two steel tapes, one of which acts as a flexure hinge. The other tape, which has a longer free length, can be deflected by a piston to raise one end of the measuring head. When the piston is withdrawn, the head returns to a position defined by the lengths of the two tapes. Measurements of the repeatability of this 'nodding' action have shown that it is 1.4 nm (standard deviation), which illustrates the excellent results that can be achieved by flexure devices. The sequence for inserting a gauge between the measuring heads is that the head is first raised and then the beam is raised to lift the previous gauge clear of the lower measuring head. The movement of the beam is in fact a rotation about a flexure pivot situated at the left hand end of the beam as depicted in Fig 3. The cassette is then advanced one step and the beam is lowered to place the next gauge on the measuring head. Finally, the

Fig 6 View of the measuring zone with the cover opened and ram raised to show details

PRECISION ENGINEERING

111

Measuring heads

Files Program Master e r r o r s Tolerances

I Multiplexer I

circuit

I -I
v

dvm

9825

Control system

I
J

Valves ~"

+
Code readers Detectors Comparator performance The first stages in the development of the comparator concentrated on answering two critical questions:(i) Can we achieve sufficient tinearity and stability from the transducers and measuring electronics? (ii) Can the nodding head' mechanism achieve good repeatability? The answer to both these questions is yes. Measurement ot transducer linearity showed a maximum error of only 1 nm over a 5/lm range and the short term stability of the measuring system as measured by the repeatability of readings on a single gauge is 1.4 nm (s.d.). The repeatability of measurement when the nodding head is operated between readings is 2.0 nm (s.d.), from which the repeatability of the nodding head itself can be calculated as 1.4 nm (s.d.). Some repeatabilities of length measurement on a range of gauges are listed in Table 2. For the smaller gauges the results for the last gauge to be measured (position 24 in cassette) are as good as those for the first gauge to be measured after the master gauge (position 1 in cassette). The larger standard deviation on the 100 mm gauge in position 24 does, however, indicate the limitations Table 2 Results of repeatability tests on a range of different size gauges Nominal size, mm Position in cassette Standard deviation of length measurement, am 3.0 2.4 2.9 2.2 2.4 2.7 2.9 5.5 Performance required, nm 5 5 5 5 5 5 15 15

Code readers

Fig 7 The measuring and control systems


by a small amount so that, again, any delay in clamping will leave an excess clearance and will not result in damage caused by trying to insert a gauge into too small a gap. The control sequence includes automatic loading of cassettes from the storage rack onto the beam, adjustment of the height of the ram to give the correct gap between the gauge and the upper measuring head, placing of each gauge in turn at the measuring position and unloading the cassette from the beam after all the gauges have been measured. When a gauge has been positioned ready for measurement, the control system hands over to the computer which multiplexes to each probe in turn and stores the readings. When all the probes have been read, the computer signals to the control system that the measurements are completed and the control system then sequences the operations to position the next gauge for measurement.

Data handling
A Hewlett-Packard 9825A desk top computer is used for the data handling since it combines the necessary computing and storage capabilities with ease of interfacing to the other elements of the system (Fig 7). When a new cassette of gauges is loaded onto the beam, an identification code etched on the master gauge is read by fibre optic detectors and the processor must then select the appropriate master calibration and tolerance data from its memory files. In addition, depending on the crosssectional area of the gauge, the correct probes for measurement must be selected. The readings on the master gauge must be stored and then for each remaining gauge in the cassette, the grade and calibration must be calculated and printed. The master gauges have all been calibrated by a standard interferometric method while wrung to a platten. Therefore, although gauges are in the free state when measured in the comparator, a contact error (or wringing film thickness) will, as required, be included in the length measurement.

2 5 10.5 100

1 24 1 24 1 24 1 24

112

PRECISION ENGINEERING

imposed by thermal drift resulting from temperature changes in the laboratory in which the machine was tested prior to its transfer to a temperature controlled room. Repeatability of measurement is not, of course, synonymous with accuracy since systematic errors may be present. I n order to test the accuracy of the comparator, a number of gauges previously calibrated interferometrically by a BCS approved laboratory have been measured. Analysis of the results showed that there were no statistically significant differences between the two sets of measurements. Following extensive laboratory trials covering many thousands of measurements the instrument has been transferred to Coventry Gauge and is currently being integrated into the production inspection system.

Acknowledgements
The author wishes to acknowledge the major contributions to this work made by colleagues at TI Research Laboratories, particular D.M. Marsh, T.S.R. Potter and Mrs P.T. Humm, and to thank the Chairman of Tube Investments for permission to publish this paper. It is based on a paper presented at NELEX '80 held on 7 - 9 October 1980 at the National Engineering Laboratory, East Kilbride, Glasgow G75 0QU, UK.

R eferences
1. Birch K.G. Uncertainties in the Measurementof Gauge Blocks by Interferometry. NPL Report MOM 29, 1978 2. SiddallG.J. and Willey P.C.T. Flat-SurfaceWringing and Contact Error Variability. J. Phys. D (Appl. Phys.), 1970, 3, 8-28

Open-bridge cmm
Open bridge configuration has been adapted to provide both convenient loading of large engineering components and structural rigidity for the recently introduced cnc Mauser KMZS co-ordinate measuring machines. Travel of the largest version of the Mauser KMZ-S is 3000mm in the x axis, 2000mm in the y axis and 1600ram in thez axis. When greater measuring capacity is required, the design is altered, and the machine is then designated KMZ-B. The two types of machine, however, are similar in their equipment, accuracy and performance. Resolution in each axis is 1/1m and repeatability is said to be better than 3/~m. In the open-bridge configuration, the horizontal beam of the travelling bridge is supported at one end by a raised guideway, which is itself a bridge. The axes are claimed to be perpendicular to one another to within 2 arc-seconds. Transistor-controlled dc motors acting through pre-loaded ball screws drive the machine. Measurement in each axis is performed by a Zeiss Mini-Phocosin optical scale. The reference scale is an interference grating on glass, and machine travel is registered by reader heads with light emitting diodes. The vertical slide carries a universal probe head which senses inductively the probe deflection in three axes. With the probe in contact with the work-piece, the machine moves to bring probe deflection to zero. Data is collected only when the machine is at a standstill, which is claimed to improve accuracy by eliminating errors due to machine motion. Direction and force of probing are controlled by a logic link between probe head and computer. This active probing system gives self-centring under computer control, easing measurement of tooth spacing of gears and

The processor is a HewlettPackard HP 9825 T. Results are recorded by high-speed printer and x-y plotter. Although data are normally recorded concurrently with measurement, the machine cycle can be speeded by using an off-line computer system for printout. Mauser specify the measuring uncertainty in mm for the largest KMZ-S per co-ordinate in one plane at 20 C as: L Ugs = A + - 250 000 Where A, the standard deviation, is 0.006mm in the x axis, and 0.004mm in the y and z axis. Hahn & Kolb (GB) Ltd, Leicester Road, Rugby, Warwickshire CV21 1NY, UK or Mauser-Werke Oberndorf GmbH, Postfach 1349 und 1360, D-7238 Oberndorf, FRG

Mauser KMZ-S coordinate measuring machine location of centres of small bores, for example.

Transducer miniaturised
Controlling the machine
Stored in the memory of the computer is the Zeiss/Mauser software known as UMESS-S. Developed to aid the operator in communicating with the machine, this software also reduces the data collected by the machine into easily interpreted results. UMESS-S is designed to measure combinations of features such as plane surfaces, cylinders, ellipses, cones and spheres. Additional software packages are available for specific applications including the measurement of spiral bevel gears, cams, turbine blades, impellers and generalised curved surfaces. Illustrating the power of the software is its ability to recognise the attitudes of spatial axes and planes as well as to differentiate between internal and external circles and cylinders. Setting up a work-piece is quickly done because its alignment can be adjusted automatically within the computer. Substantial reduction in the dimensions of the transducer head of the Fastrac 2000 linear measuring system have been achieved without loss of resolution or traverse speed. The cylindrical 552-20 transducer head, which is 20mm in diameter by 20mm long (compared with 27 x 37mm approx) detects infra red radiation reflected from a chromium on glass grating. At traverse speeds of 1m/s, resolution and claimed repeatability are 1/~m. Special systems to resolve to 0.5/lm at traverse speeds up to 20m/s can also be supplied with the 552-20. The manufacturers, who describe drift and stability characteristics of 55220 as excellent, suggest that the equipment is particularly suited for application in coordinate tables, inspeczion equipment and photogrammetry tables. Metronic Technology Ltd, PO Box 404, Kings Langley, Hertfordshire WD4 8ND, UK

PRECISION ENGINEERING

113

Вам также может понравиться