|°
Specification standards for coordinate measuring machines Should there be a British Standard? In the past few issues brief descriptions hae been carried of the European (CMMA) 1and USA (B89.1.12) 2 specification standards for coordinate measuring machines (cmms). The British have been biding their time, following the discussions, but now look as though they are on the move. A 'consultative meeting' held at NPL on 14 March 1984 produced a consensus that there was a need for a British Standard in this area and that: • • •
simplicity was the target the timescale for adoption should be short the 'software problem' should be considered at all stages
Introducing the meeting, Dr Peter Campion, Deputy Director of NPL, stated that engineering metrology had been rescued from 'the doldrums often years ago' by the cornbination of three factors: the rapid advances in microelectronics and in the speed of data acquisition and processing; the increased sophistication of production methods, and the demands that had placed on rnetrology; and an increased awareness of the importance of quality, Industry had created the demand for a standard of the type suggested, No one at the meeting - - representatives of all the major UK cmm manufacturers and several large users - - disagreed with the opening remarks; but what were the problems to be encountered, and how had other attempts succeeded (or failed)? The Coordinate Measuring Machines Manufacturers Association (CMMA) was represented by its President, Dr Jacoby (Director for metrology equipment development, Leitz). The CMMA comprises the main European manufacturers plus three Japanese companies. They had set out with the 'brief' that the user should know, and be able to verify, the accuracy of the measurements made with a cram, compared to the figures given by the cmm manufacturer. To this end the user needed definitions and values which could be used in (1) tests prior to purchase, (2) the acceptance test, and (3) verification throughout the lifetime of the machine. In the end
152
the CMMA standard had been aimed at the acceptance test. Short though his presentation was, Dr Jacoby mentioned many of the arguments that must have continued for months during the preparation of the standard - - the deftnitions (all length dependent so that accuracies are given as an absolute value plus a length dependent term), the choice of gauge blocks ('at the time they were the only measurement aids which were traceable to national standards'), and the measurement approach (accuracies valid within the whole volume, measurements must represent shop floor operation). It was conceded that there was room for improvement: rotary tables needed to be considered, and there was a need for a derating factor for temperature dependence. In response to a questioner from Rank Taylor Hobson, Dr Jacoby also indicated that certain questions concerning s o f t w a r e - - eg, relating to alignment of gauge blocks in s p a c e - - had been passed over after a quick evaluation. A German VDI standard, which agreed closely with the CMMA document, had gone slightly further in considering software problems but this also omitted details. The more detailed approach of the American B89.1.12 draft standard was described by Professor Pat McKeown (CUPE). Most of those in attendance were familiar with the proposals it contained and some of the manufacturers were already using it in the USA. Ball bar measurements were the manufacturers main complaint, in that they were said to be unrepresentative of 'the real situation' involving deflection of the probe tip. Despite this it soon became clear that the prevailing mood was one of avoiding the reinvention of the wheel. If two good documents were already on the table, and with the need for prompt action having already been emphasized, the course to follow had to be one of improvement and not of starting from scratch. A British Standard based on an improved CMMA standard was suggested by George Macpherson (Ferranti): 'The CMMA standard ~ a slim document - - is the distillation of the experience of a significant number of cram manufacturers and,
I would propose, is a good start. It has a reasonable degree of international credence and with a lot of work might be better than starting again'. But it was pointed out, if Ferranti themselves are already meeting the ANSI interim Standard in the USA why should the issue be further complicated?
User survey Further manufacturer and customer opinion was provided by Phil Smith (National Engineering Laboratory) who reported the findings of an NEL survey of cmm usage in the UK 3. The survey was undertaken to discover, and define, the potential market for the mechanical space frame that had been developed at NEL4. The principal objectives of the survey were to establish: •
• • • •
the number of cmm used by the general body of mechanical engineering companies in the UK the manufacturers, types and sizes of cram in use the usage from one industrial sector to another and the usage by individual companies the verification practices the level of interest in a national standard to introduce a method of verification available to all cram users
Findings relevant to verification and the need for a Standard are shown on page 153. The survey also showed that Ferranti dominates the UK market, followed some way behind by LK Tools. DEA Soci~td Genevoise and C.E. Johannson are in a group below LK Tools. Motor vehicle manufacturing is estimated to account for between one-quarter and one-third of all cmms in use. This is followed by aerospace, pumps, valves and compressors, metal working machine tools, and scientific and industrial instrurnents - - each with about 10% of all crams in the survey. Although the majority of cram users rely on the cmm manufacturers to verify the accuracy of their machines, a few of the users were dissatisfied with this situation for a variety of reasons, says the report. For instance, the certificate of verification given by a cram manufacturer
J U L Y 1984 V O L 6 NO 3
|'
PItE¢IS is reluctantly accepted by the users' customers who prefer to work to national or equivalent standards. This problem would be solved if a British Standard were available that gave advice and guidelines on verification,
Similarly, other causes of concern among cmm users would also be solved - - a machine's accuracy would be traceable to the national standards of length, and the task of specifying a new machine with the
Results of NEE survey of coordinate measuring machine usage in the UK Verification Practices
Frequency of verification In the NEL survey 64 respondents gave information on the length of time between each verification of their machines. A few stated that verification was carried out 'on demand'; this may mean an inhouse check using a standard workpiece or jig, and when a discrepancy between recorded measurements is noted the user refers to an outside agency for a professional verification. In comparison, another user Interval
No. of users
12 months 6 months 2-6 months
37 16 11
supported a regular three monthly self-verification with an annual one by the supplier. The majority of users, however, had their machines verified at regular intervals,
Means of verification
three Soci~t~ Genevoise machines, Some respondents used both a laser and a mechanical means; in one case the laser was apparently used to check Zeiss 650 while a step standard was used for DEA Iota; in a second case a laser was used for an LK Granite and a Talyvel for a DEA Beta. Two respondents used a cornbination of mechanical methods; one relied on length bars and a calmbrated workpiece, the other used slip gauges and a Talyvel.
Verifying organization Respondents were asked who verified their cram, whether it was their own company, the supplier, or another organization. Of the 93 respondents 89 replied to this question; the information is tabulated below. Eleven respondents supported the supplier's verification by an in-house verification at intermediate periods. Of the remaining four respondents, two used a cmm by subcontract, and two had their own cmm verified by a subcontractor.
In the survey 57 respondents gave information on the means used to
Verifying organization
verify their cmm. The replies showed that a laser interferometer was the single most common instrument used; however, nearly twice as many respondents used specified or unspecified 'mechanical' means than those that used lasers. Two respondents refer to verification by the Johannson method; one of them has a Johannson Cordimet but surprisingly the other has a Ferranti and
User 22 Supplier or manufacturer 56 Supplier and in-house 11 Other 4
Means of verification
No. of users
Laser
20
Step standard Length bar Slip gauge Calibrated workpiece Unspecified
2 4 4 8 20
PRECISION ENGINEERING
No. of users
Introduction of standards The users of cram were asked for their opinion on the need for a British Standard that would introduce a common practice for verifying the accuracy of these machines. Of the 91 cram users who replied to this question 66 felt that there was a need and 23 said that they had no such need. Of the 66 users who expressed a need for a standard method, 40 added a comment based on the experience that justified their opinion. Two non-commital replies were received,
accuracy appropriate to the users' needs would be simplified. Furthermore, the current practice of userverification could continue with the increased confidence of complying with a standard method. The comments are diverse and they include the recognition of the
complexityofverification;thepotential benefits to the user and his customers; the fact that any verification system should be backed by a standard; the need for traceability to the national standards; and the realization that the lack of a standard leaves a gap in the system. Some respondents deal with large organizations (eg MoD, CEGB, CAA, British Leyland, Ford Motor Company) that are accustomed to specifying to British and Defence Standards; part of this group of users stated that the certification of a cmm by the manufacturer was not entirely acceptable to their customers. In the opinion of other users, a standard for verification would allow the intercomparison of cram accuracy; this would be particularly useful when specifying a new machine. Of the 23 respondents who rejected the idea of a national standard, some commented that the present levels of accuracy were acceptable for their type of work.
Repliesfrom Royal Ordnance Factories During the analysis of the replies from industry the survey was extended to include the Royal Ordnance Factories. The reply gave information on the methods used by this organization as a group. The methods of calibration vary greatly and in view of the numbers of machines and the differing methods of calibration used for identical machines these have not been tabulated but are basically: • By contract with manufacturer (at six-monthly intervals) interim check (at six-monthly intervals) by factory quality staff using Sogenique machine calibrator • Calibrated blocks, straight edges and squares with and without lasers at 6 or 12-monthly intervals • Ball frame with lasers and oscilloscope at 12-monthly intervals
153
|'
Most respondents agreed that a standard verification method was required, said Smith, but opinions differed within the same organization. Even within the same factory separate methods are used to verify different machines. Accepting that different standards of accu racy a re required by different parts of industry, the present diversity of methods promotes a situation in which agreement of accuracy between crams is unlikely simply because there are no standard guidelines for verification.
produced very good hardware. Still, however, investment in hardware far outweighed investment in software development and proving, and this position had to be reversed.
After effects
Software It was no great surprise to most present at the meeting that algorithms and software were the subject of much discussion. Both users and
'Computer-Aided Dimensional Engineering Metrology'. This cornmittee will be charged with producing three standards:
manufacturers had their say; the former calling for algorithms to meet specifications in the same way as accuracy and for access to fundamental definitions, the latter maintaining their right to protect their 'stock in trade', The question of definitions used in the algorithm base was raised by several people. A first stage, said Professor McKeown, was to investigate the algorithm base - - how do we define the displacement between two holes, or planes, or flatness? Dr Maurice Cox, a member of the NPL team investigating algorithms and software in metrology s, suggested that considerable advantages would accrue from devising, implementing and using good algorithms within the general principles and procedures of a 'complete measuring system'. It was necessary to develop mathematical conceptual methods, and in the first instance work would of necessity be confined to geometrical quantities - - straightness, flatness, cylindricity, conicity. George Macpherson agreed that the underlying mechanical definitions were not sufficiently specific and that this had led to problems both of comparability and of
(1) Methods for verifying the mechanical performance of a coordinate measuring machine. (2) Formulating definitions of geometrical form suitable for use with computer aided equipment, (3) Methods for verifying the software used with computer aided equipment,
154
References 1
Following the NPL meeting its organizer, Dr K.F. Poulter 6, has held further discussions with senior BSI staff and formally requested the Institute to start new work in this field. BSI have agreed to form a new committee in the MEE series entitled
reliability of results. Several users indicated that this had now led them to write their own specifications and MoD QAD at Woolwich were even using their specifications at the acceptance stage. Sixteen years experience have been gained at Woolwich and for the last six there has been general satisfaction with the hardware. In the last two years attention has shifted to software and the instability of algorithms. Dr Cox agreed that the manufacturers now
already prepared by other groups it is hoped that a draft standard will be available by September 1985.
Trade Associations and Professional Institutions are being asked to nominate people to serve on the committee. It is intended that the committee will tackle tasks (1) and (2) in parallel and that (3) will follow in due course. By making use of the documents
2
3
4
5
6
CMMA. Accuracy specification for coordinate measuring machines. 1982, Coordinate Measuring Machines Manufacturers Association, c/o BCM, 27a Old Gloucester Street, London WC1N3XX, UK. See also: Precision Engineering, April 1983, 5 (2),86-87 ASMEB89.1.12 Working Group. Method for performance evaluation of coordinate measuring machines. March 1983See also: Precision Engineering, July 1983, 5 (3), 125128 Shelley M., Cook M., Smith P. and Harvie A. A survey of coordinate measuring machine usage in the UK. Publication EAU M39, March 1984, Economic Assessment Unit, National Engineering Laboratory, East Kilbride, Glasgow, UK Harvie A. and Beattie J. S. NEL's coordinate measuring machine calibration service for industrial application. Production Engineer, June 1982 Cox M. G. and Ja©kson K. Algorithms and software for engineering metrology: A statement of need. NPL Report MOM 65, June 1983 Dr.MetrolF.ogy,K. Poulter,DivisionHeadof-MechanicalEngineering and Optical Metrology, National Physical Laboratory, Queens Road, Teddington, Middlesex TWl I OLW, UK
Fixture makers often have to use levers to gain access to bores, which can be expensive, difficult to manufacture and prone to error. The Mercer miniature lever probe, however, can be implanted into ground slots in the bore mandrel or fixture to allow direct measurement of bores. Other applications, such as the gauging o f lengths and the measurement of narrow /ands, can also make use o f the miniature probe. The probe consists of three main mechanical parts - - a steel body (about25 m m long), a probing head with a 2.00 m m carbide ball inserted, and a leaf spring which connects the head to the body. Inside the probe is an armature and an inductance coil which produces a gauging signal for use with standard or microprocessor based electronic units. Tip force of the probe is 150 to 220 g at electrical zero and repeatability is to +_0.0001 ram. The measuring range is +_0.250 m m and the total travel of the tip is 0.650 m m with 0.050 m m pre-travel. M a x i m u m linearity error is +_0.2% of range. Thomas Mercer Limited, Eywood Road, St A/bans, Hertfordshire AL 1 2ND, UK
J U L Y 1 9 8 4 V O L 6 NO 3