A Vision of the Future ?

Twenty-first Century Calibration

The global market for sales of electronic test and measurement (T&M) equipment has increased in size considerably over the past 30 years and is estimated at 9000 million (US) dollars. Meanwhile, the maintenance costs industry about $2000M of which $1200M is for provision of calibration services; a market that has developed almost entirely during this period.

What factors are influencing this growth and, as the millennium approaches, how might the calibration business develop in the twenty-first century?

Standards

The impact of "quality regulation", particularly, has been immense. Originally to qualify military supply contracts in the 1960s and 70s, manufacturing and inspection standards such as UK Def.Stan.05-21 (later NATO AQAP1) and US MIL-45208 were themselves supported by the calibration management standards DS 05-26 (NATO AQAP6) and Mil-Standard 45662 respectively. The commercial importance of the defense sector meant that its requirements actually influenced the whole of industry. This criteria became a basis for the ISO9000-series of international quality management standards, including ISO10012 for management and calibration of measuring equipment.

In many worldwide markets, certification against ISO9000 has become an initial requirement of prospective suppliers. In the UK alone, 45 thousand entities have such registration in 1996 and, for some, regular calibration of their measuring equipment is a new practice driven by the standard. Yet many instrument users are unfamiliar with the intricacies of calibration and concentrate upon their core competency. Consequently, they look to other companies offering the specialist skills and resources necessary; vendors that are independently assessed against ISO9000 or, less likely, against technical competency criteria based on ISO/IEC Guide 25 (or European Norm EN45001) such as UKAS accreditation.

Additionally, there is broadening legislation concerning products' fitness-for-purpose, environmental impact and safety. This is implemented by requiring manufacturers to certify conformity with technical regulations and often dependent upon reliable measurement.

With such a large potential market, it's not surprising that there are many more such service-providers than ever before. This increased competition especially affects the low-end or basic products, where set-up costs are relatively low and the installed base high. Worldwide, although accredited certification presently amounts to only a small percentage of the total business, a stated preference for this level of service is increasingly apparent in some countries. Limited agency resources result in delay for accreditation of up to a year and the developing market will have to be prudently anticipated by vendors to ensure the availability of accredited certification when requested.

Operating Environment

Customers' expectations are constantly rising but to satisfy them it isn't enough to focus on quality alone; commercial needs are also important. Financial pressure on suppliers will continue relentlessly, aided by increasing competition. One way of reducing delivery-cost is to minimize the engineering overhead per job which seems incompatible with the trend towards complicated instrumentation involving more extensive testing. Automatic test equipment (ATE) can address the complexity and test-time issues so, providing that "smart" software is used, the over-skilled engineer will be replaced by a less expensive operator. Although the scenario looks grim for technicians, the move away from manual calibration provides new opportunity; through increased requirements for test design, method validation, software development, uncertainty assessment and consultancy on application/usage versus extent of test, data analysis, periodicity recommendation, etc..

In general, technology is becoming cheaper. Also, improved circuit theory and design, with enhanced manufacturing precision has led to more reliable and predictable performance. It is quite feasible to produce an instrument that is conservatively specified in relation to its expected capability, such that it need never require periodic maintenance -- calibration for life. Handheld instruments used by field service personnel could be the first of this new breed.

The same engineering advances are seen in the lengthening of manufacturers' initial calibration interval recommendations; 3-6 months was typical 20 years ago, 1-2 years is now common and becoming 3-5 years in the next decade? The computer business well illustrates that technological advancement can itself drive the market. This is apparent in the T&M arena too, with the majority of instruments sold being first introduced within the past 5 years, according to HP. This technology refreshment rate could effectively obsolete periodic calibration, especially for the high-tech industries.

A consequence of all this may be that calibration vendors could see their workload decline if the potential market doesn't grow appreciably. Many already focused their resources to the emerging "calibration" market after repair revenues were impacted by increased mean-time-to-failure times, so competition to maintain a share of the pot may become even more cut-throat. This commercial pressure will heighten the quality-conscious suppliers' expectations of protection from "corner-cutting cowboy operators", through effective policing by the accreditation bodies and registrars.

Constantly dealing with newly introduced equipment presents challenges to the calibration laboratory. For instance, technicians may not become familiar with a particular product before it's superseded, service documentation quickly becomes obsolete and locally-devised calibration procedures need constant development. It all adds up to high ongoing operating cost to keep up-to-date with the escalating technology.

Production pressure and lean inventories are leading to intolerance of today's return-to-bench turnaround times. As in the computer business, downtimes of a few hours are preferable. Calibration carried-out on customers' premises will become increasingly popular, especially with the larger manufacturing companies. Similarly, for logistical reasons, there is a trend toward out-sourcing to a single vendor who either undertakes all work or interfaces with other subcontractors, adding value by dealing with all aspects of the equipment's management on behalf of the customer.

Technology

Developing technology, notably computer-related, will impact the cal lab of the future in many ways.

One T&M producer has responded to users' sensitivities about the traditional return-to-bench approach by developing a transfer standard that, under computer-direction, allows the customer to calibrate his own multifunction DMM calibrator. The lab that owns the transfer standard does, of course, need to calibrate it before and after use by the client and assess all data in order to issue a calibration certificate. The process, which has been accepted by British accreditation agency UKAS, is reported to work successfully which may encourage further development of remote, co-operative calibration for other products.

There has been a merging of measuring equipment with computers during the past 20 years and this diffusion is likely to accelerate particularly in the fast-moving communications and automotive sectors. Products will present the results of their measurements in terms used in the application, meaning that specifications in the familiar SI units (voltage, time, attenuation, etc.) could become less apparent. Who holds the national standard for the video industry's arcane parameters such as color saturation, illegal frames and "front-porches"?! Calibration in the traditional sense may not be appropriate so perhaps "verification" is the service required; but what is that? Official recognition and guidance is required.

The increasingly popular instrument-on-a-card, requiring mainframe and computer to operate, offers opportunity since existing calibration support is only in its infancy. Some PC manufacturers already build to order using modular components and a possible T&M development is customized equipment that, through hardware modules and firmware options, would offer the features and functionality for the user's application. Existing products are already controlled by firmware that can be configured/upgraded to add new capability to existing hardware. An achievable development of this concept could be remote software-download via the Internet and this telephone help-desk conversation can't be far away......"You want the features of option 006, sir? Okay, connect the instrument's internal modem to your telephone-jack, swipe your credit card through the onboard reader and in a few moments you'll be all set!"

Manual adjustment through mechanical components like potentiometers or trimmer capacitors has largely disappeared from modern test gear in favor of electronic error-correction. However, some manufacturers do not publicize information about updating such memory, which may limit the ordinary calibration laboratory's ability to adjust the product for compliance with a specification. The majority of industrial users consider assurance of compliance as a critical need.

Extensive, internal firmware routines using "self-cal" techniques are becoming more common, having the benefit of providing confidence to the user and, sometimes, reducing the number of external, traceable standards to effect adjustment at routine calibration. Traditionally, performance verification has often required a great deal of equipment but as greater acceptance of self- and artifact calibration develops from supporting evidence, future calibration expectations may become less extensive, thus reducing ownership costs.

Some instruments are principally dependent on their timebase accuracy. GPS-disciplined oscillators might be built into this equipment that, in conjunction with "smart" software, would allow self-contained, traceable calibration simply by attaching an antenna perhaps for a few hours each month.

Developing knowledge of materials science, illustrated by work on "high" temperature (~70K) superconductors, could lead to Josephson / Hall effect devices operating at room temperature or, perhaps, in a small electrically-operated cryostat. Combination with the GPS frequency standard mentioned enables an instrument having continuous, internal comparison with primary voltage/resistance standards.

Test software will be more flexible. In conjunction with the availability of instrument specifications in electronic data-file format, ATE programs will employ algorithms describing the test methodology. Modules will perform specific tests for various equipment-types and classes, using a choice of test gear and new procedures will be "built" from selected, standard modules. Using the algorithm, uncertainties will be dynamically calculated based upon chosen testpoints, equipment used and configuration, environmental factors and even whether the unit-under-test is "noisy". Test results could be parsed through a neural network knowledge database which could, given basic information about the effect of adjustments (which it could "learn" from experiment), advise the technician of the course of action or, since the UUT will be electronically adjustable, make the decision and carry out automatically. Developing this further, the percentage chance of the product being falsely reported as in-tolerance could be reported -- consumer risk analysis -- and customer-defined levels of risk could then drive the optimization schedule. Automatic cost-based, metrology-driven decision making.

Delivery

In addition to their invasion of test equipment itself, computers will revolutionize other aspects of the service delivery process and not just through ATE use.

Since calibration data will be available on-line, test results and calibration certificate will be available for customer view or download via the Internet. Eliminating the need to routinely produce paper hardcopy will reduce costs of supplier and customer, not least for filing space. If wanted, during audit perhaps, the customer simply accesses the supplier's network server using his favorite worldwide web (WWW) browser and prints the material himself as needed. The November 1995 resolution of the European co-operation for Accreditation of Laboratories (EAL) to recognize calibration certificates carrying a facsimile authorizing signature, facilitates the progression away from hardcopy

Although the availability of test data via the Internet is highly desirable, providing it in an easily usable form will be most important so that it can be imported directly into customers' systems. Manually entering correction factors into software from test reports is too time consuming and prone to mistakes. To achieve this, a standard file format is required which, for universality with existing systems and platforms would probably be ASCII. Specialized programs could be developed to read in the data in whatever format the vendor supplies but that would be inefficient and difficult to maintain. The standard format must cover existing test result information as well as being easily extensible to future needs. A tagged ASCII format, similar to the hypertext mark-up language (HTML) of the WWW seems feasible. For example, a power sensor's response might be exported as:-

<MODEL> 8482A
<TYPE> POWER SENSOR
<MANUFACTURER> HP
<SERIAL> 1234A12345
<REFERENCE FREQUENCY> 50E6
<REFERENCE FACTOR> 99.0
<FREQUENCY 1> 100E3
<CAL FACTOR 1> 98.7
<FREQUENCY 2> 5.000E6
<CAL FACTOR 2> 98.6

Tags mean that different vendors could supply the data in slightly different ways (carriage returns or no carriage returns or extra carriage returns), using the <TABLE> tags, or in a different order without concern for how it would be imported by applications.

The user-friendly WWW, in conjunction with automated work-scheduling and the possible test software developments mentioned, could offer almost unlimited potential for customer interaction. Imagine.

A customer accesses the calibration lab's web server and selects Calibration Services. Pop-up screens make it easy to log his details and qualify that calibration is offered for the specific product. A few more mouse clicks selects the certification type and service level. He opts for Accredited Cal and Per Manufacturer's Recommendation and depending on available resources (personnel and other scheduled work), he reserves a time-slot. This is a "live" system so as time-slots fill up, they're no longer offered but depending on production capacity, it may take several orders for the same time-slot to fill it up. Finally, the system confirms the order and provides instruction concerning shipping arrangements.

Now, stretch your imagination. The metrologically knowledgeable customer logs another product -- say an HP8562B spectrum analyzer -- but this time wants to define the extent of testing so chooses Custom Cal. Now the screen fills up with a list of test names. He points and clicks on the tests of interest; he chooses, say, Input attenuator accuracy, IF gain, Resolution bandwidth, Scale fidelity and Frequency response. He confirms his requirement and the system quotes the applicable custom price. He accepts, chooses time-slot, etc. When the unit arrives at the lab it's already logged as Custom Cal, with encoded information that causes the ATE system(s) to automatically select the required tests. Even more valuable to the customer would be customization of the actual calibration testpoints (delete/add points within upper and lower boundaries). This allows the testing to be consistent with the usage made of the instrument and, significantly, avoidance of unnecessary recall evaluations due to out-of-specification failures of unused ranges/functions. Now this is rocket science !

Conclusion

When, or if, these prophecies become reality is pure conjecture; there is one certainty, that the service / support environment will continue to evolve. As ever, successful calibration laboratories of the future will be those who are responsive to, or better, anticipate those needs. The forecast is written from the viewpoint of a large calibration vendor operating in multinational markets and it is not intended to suggest that current practices will entirely disappear -- calibration as we know it will continue well into the next century. But in 50 years time......?

Appendix

It seems an excellent example for how proper measurement underpins the correct and reliable operation of any manufactured product, so was the 2 micron error simply due to inadequate calibration practices as has been alleged? The real cause is, predictably, not quite so straight-forward.

The telescope is fundamentally made-up from two matching mirrors having hyperbolic surfaces. Unlike spheres or paraboloids, hyperbolic surfaces have no intrinsic focus and are therefore difficult to test directly during polishing. To assure correct grinding to prescription, a "null-lens" is manufactured which can be set-up using testable optics; the combined null-lens/hyperbolic does focus light and so can be tested together, as a unit. The null-lens has to be carefully tuned to the particular design of the mirror to be manufactured. If this is done incorrectly, the hyperbolic surface will be erroneous but since the null-lens and mirror test perfectly together, the problem won't be detected until the null-lens is discarded and attempt made to use or test the whole assembled telescope which includes the, equally fined-tuned, secondary mirror.

A component of the null-lens assembly used for HST's primary mirror was a "field-lens" that needed to be accurately positioned within the device. A special metering rod was made to achieve this that reportedly had a rounded end. In order to assure that the correct location on the end was optically referenced, a special cap with a central hole was placed on the metering rod. Unfortunately, this had a speck of paint missing which led to the resulting reflective point on the cap being referenced rather than the end of the metering rod. The field lens was consequently positioned about 1 mm away from its correct location, leading to a compensating error being inadvertently ground into the primary mirror.

Final testing of the assembled telescope was forsaken due to schedule pressure and the cost to build the necessary facilities. With hindsight, comparatively cheap tests could have been designed which would have identified the flaw, if the problem had been looked-for. In fact, tests made using a lower quality, back-up null-lens did suggest a problem but the results were apparently dismissed as unreliable in the rush to meet deadlines. Fortunately, the mirror's smoothness is outstanding and since the resulting telescope was so "perfectly wrong", the true prescription could be determined and corrective optics designed and manufactured. Performance now exceeds the original expectations but at an additional cost of $40M, three years of "lost" astronomy.... and, of course, dented reputations.

Acknowledgment
I am grateful to Dr. Peter Jakobson of the Astrophysics Division of the European Space Agency in Holland for providing the simplified interpretation of the HST problem described in the Appendix.