A MODEL FOR MAINTAINING STANDARDS OF TEMPERATURE IN A COMPANY WITH SATELLITE OPERATIONS

Henry E. Sostmann and John P. Tavener

ABSTRACT

A system is described that allows Corporate headquarters to control and monitor the maintenance of a temperature base at satellite and remote development and production facilities.

INTRODUCTION - THE PROBLEM

Temperature is the most often measured physical parameter, and in most production processes, it is the variable which is most important in the quality assurance of the product. For examples:

For products which must be heat-sterilized, it is necessary to assure that temperatures are high enough to accomplish sterilization, yet not high enough to interfere with the integrity of the product. For many such products, it is necessary to prove to regulatory authorities that temperature is under control and traceable to legal standards.

In the automatic soldering of printed circuit boards, it is necessary to assure that the molten solder is hot enough properly to wet the component leads, and not so hot as to destroy semiconductor chips and other components.

The production of textile fibers requires close control, and particularly close reproducibility, of temperature at the spinnerets.

Other examples could be given, where control is required between relatively narrow temperature limits. The quality of temperature assurance depends upon the accurate calibration of thermometers and process control temperature equipment.

Large companies often have satellite operations; research, development or production facilities which are at other locations or perhaps even in other countries, requiring a uniform base for the quantity temperature throughout the organization. How can total Corporate assurance of that temperature base, via the calibration and control of thermometers, be established, maintained and assured?

AT THE CORPORATE HEADQUARTERS

Most Companies maintain metrology laboratories at central headquarters, whose charge is to realize the units of measurement important to their operations. These may include electrical, mechanical, etc. units, and certainly the International Temperature Scale of 1990 (ITS-90). The Laboratory will be responsible for disseminating the units of measurement to the satellite facilities and maintaining these facilities in adequate control.

ITS-90 is based upon the realization of specified phase equilibria which exist in nature; for example, the co-existence of the liquid, solid and vapor phases (triple points) or the liquid and solid phases at stipulated pressure (melting or freezing points) of pure materials. Primary thermometers are calibrated against these fundamental constants of nature which define the ITS-90, and the central Laboratory should be equipped to realize them over whatever range of temperature is important to Corporate interests. Table 1 shows a number of ranges, and the phase equilibria required to calibrate thermometers over these ranges.

ITS-90 also requires Standard Platinum Resistance Thermometers (SPRTs) which meet the requirements of ITS-90, for use in interpolating between the fixed points of the Scale. NAMAS and ISO-9000 require that the Laboratory have at least two such thermometers so that they can be intercompared. It is much better to have three; more meaningful statistics can be extracted from that number. Two thermometers which disagree provide no information about which is in error. Three thermometers, two of which agree while the third does not, provide at least a hint of which may be in error. If the Corporation's range of temperature extends beyond 961 C, which is the upper end of the ITS-90 Scale for platinum resistance thermometers, standard thermocouples will be required. Although the ITS-90 calls for radiation pyrometers instead of thermocouples to be interpolation instruments above 961 C, we believe that the standard thermocouple will continue to be used as the convenient standard of choice.

The Laboratory's SPRTs will be used for the calibration, by comparison, of working standards, and for no other purpose. The fixed points are used to maintain these primary thermometers in calibration. Depending upon National and other regulations, it may also be necessary that they be calibrated by and traceable to a National Calibration Service, which may be a National Measurement Laboratory (such as NIST, NPL, DKD, etc.) or may be a nationally-accredited Laboratory such as Isotech's accreditation by NAMAS [2].

We are not believers in the requirement that primary thermometers be re-calibrated at specific intervals of time. A thermometer may remain in acceptable calibration for years, or be put out of calibration in minutes, depending upon its conditions of use or abuse. Furthermore, there is a risk, with some constructions of primary thermometer, simply in the process of handling and transportation from the site they serve to and on return from a recalibration Laboratory. It is our counsel to provide frequent checks of the primary thermometers at site, by techniques described below, and transport for recalibrate only when the thermometer calibration is outside acceptable limits.

It is an important function of the Corporate Laboratory to require, maintain and evaluate reports of the checking history of primary thermometers located at the Satellite sites, and to provide recalibration and recertification services when they are required.

IN THE SATELLITE LABORATORY

The Satellite Laboratory must maintain its own set of primary standards. What standards these may be depends upon the range of temperatures required and the permissible uncertainties of measurement. Although such divisions are arbitrary, and far from absolute, we might describe levels I, II and III, ranging from I = simplest to III = most comprehensive. In the following it is assumed that the Satellite Laboratory is not concerned with full maintenance of the ITS-90, but will achieve its traceability to ITS-90 and to National Standards through the Corporate Laboratory.

LEVEL I: The Satellite Laboratory maintains primary thermometers and an elementary means of verifying that their calibration has not shifted. The major function of the Satellite Laboratory is to check and verify the calibrations of plant working thermometers, including such temperature sensors as may be a part of a process control loop, etc.

Three primary thermometers are recommended, for reasons given above. They need not be SPRTs, but should be stable, and should be of such dimensions that they can be inserted into the wells of water triple point or gallium cells. The diameter should not exceed 11 mm (0.43 in) if they are to be used only in water triple point cells or gallium cells; 7.6 mm if they are also to be used in fixed point cells; and the length be not less than 37 cm (14.5 in). The thermal conductance of the stem should be such that heat loss due to stem conductance will not be a factor of the calibration.

Level I also requires one thermometric fixed point, independent of the calibration of primary thermometers. This point is used as a quality check on the primary thermometer calibration. It should be used (a) whenever the thermometer has been sent to the Corporate Laboratory for calibration, to assure that no shift or other trauma has occurred in transportation (b) whenever a temperature measurement is made, to assure that the primary thermometer calibration is valid for that particular measurement. The fixed point we suggest is either the triple point of water3 or the melting point of gallium [4]. Either of these points may be realized with an accuracy of better than 0.0002 C. The measurement acquired should be transferred to a control chart specific to that thermometer. A copy of the chart periodically should be sent to the Corporate Laboratory for their permanent records of quality assurance.

Level II: is similar to Level I, except that two fixed points are maintained, which allows some measure of calibration to be performed along some length of the Scale. The two fixed points are the water triple point and the melting point of gallium. Each of these points is realizable with such accuracy that a calibration at these points and no others represents ITS-90 within 10 mK between -80 C and +680 C! (To this must be added the system uncertainties, such as those of the readout instrument, the thermometer stem conductance, etc.) Table 2 provides an estimate of departure from ITS-90 at various temperatures [5]. Records of calibrations and verifications of the specific thermometer must be maintained and sent periodically to the Corporate Laboratory.

LEVEL III: refines Level II, and includes fixed points of the ITS-90 throughout the primary thermometers' ranges of use. It requires primary thermometers, as do Levels I and II, but permits complete calibrations throughout the thermometers' ranges. It assumes that economies of capital equipment are important, but that it is (a) undesirable to transport thermometers to the Corporate Laboratory or (b) not possible to have thermometers absent for the length of time required to acquire calibrations at the Corporate Laboratory.

Isotech manufactures a product line of sealed metal freeze-point cells called "slim cells". These are similar in construction and made with the same purity of metal as its larger, basic-standard, cells, but, largely because the volume of metal is smaller, they cost substantially less. Also, less sophisticated furnaces (e.g., the Isotech "Medusa" furnace) are required to realize the equilibrium points. They are available in all metals shown in Table 1.

The slim cells are as accurate in reproducing ITS-90 as are the larger cells, but:
(a) since the immersion depth is shorter (typically 30 cm) a conventional SPRT used to evaluate the cell's performance will evidence some emergent stem loss, and slim cells therefore cannot be certified to uncertainties as small as those of the larger cells. (b) the duration of the freeze equilibrium may not be as long as that of the larger cells. Within these limitations, slim cells permit calibration of primary thermometers over the ranges shown in Table 1. Using such a system, it may be possible to send only periodic reports of calibration or verification to the Corporate Laboratory, and to transport thermometers never.

THE FUNCTION OF THE SATELLITE LABORATORY

With its temperature base assured by Level I, II or III checks, the Satellite Laboratory will possess primary thermometers (including standard thermocouples, if the temperature range of interest requires these) of assured validity. In general, these thermometers will be used in making calibrations and verifications of working thermometers by comparison, in either liquid baths or metal block furnaces.

Liquid baths have better heat transfer characteristics, and are generally limited, by the limitations of available fluids, to 300 C as a top temperature. To cover a range from -95 C below ambient to +300 C, several changes of medium may be required, due to the variation of oil viscosity with temperature. Temperature gradients on the order of 0.001 C - 0.005 C may be achieved[6]. An alternative to liquid baths, avoiding the need to change media, are fluidized calibration baths, in which finely divided alumina powder is levitated by controlled air flow, and achieves many of the characteristic of a fluid, suitable for the range 50C to 700C [7].

For less demanding isothermal characteristics and lower cost, or for temperatures above the range of liquid and fluidized baths, metal block furnaces are commonly used [8]. Depending upon model, the upper temperature limit may be as high as 1,200 C, taking them well into the range of thermocouples. These comprise in general a dry furnace which heats a metal block, which is provided with a suitable number of suitably sized holes or pockets for thermometers under test, including the primary thermometer. Accuracies using comparison techniques in block furnaces are typically:

                AMBIENT         100 TO  200 TO  300 TO  400 TO  ABOVE

                TO 100          200 C   300 C   400 C   500 C   500 C



	+/-	0.02		0.03	0.04	0.06	0.10	0.25

      

Comparison calibrations are often required between thermometers of very different physical configurations. Primary thermometers are generally between 6mm and 7mm in diameter, and are long, to avoid stem losses. Working thermometers in process measurement or process control situations are configured to suit the site in which they are used. For example, in one textile-fiber application, the working sensor is 3.2 mm (.125 in) in diameter and 38 mm (1.5 in) long, with a mounting thread and a wrench hex 12.7 mm (.50 in) across the flats, and Teflon-insulated lead wires. Calibration of such configurations often require the exercise of some ingenuity. One approach is to prepare semi-standards, or "like" standards, in which a semi-standard is made to be as close as possible to the configuration of the working thermometer, but of such dimensions that it can be calibrated against a primary thermometer in fixed-point cells or in a bath or furnace. In the example cited, the semi-standard was prepared by (a) machining off the hex (b) making the lead wire insulation ceramic instead of Teflon. Once the semi-standard had been calibrated, it was used in a comparison mode to calibrate the working standards, by screwing both into a metal block. Obviously, immersion was inadequate and stem losses were large, but they were the same for both the semi-standard and the working thermometer.

In the measurement aspect of a comparison calibration, uncertainty may be reduced, and throughput enhanced, by using a readout instrument containing a reference standard resistor. With the standard thermometer connected in place of the standard resistor, and the working thermometer connected as the unknown, the readout instrument can be made to read the ratio of the working thermometer to the standard thermometer. The indication of the working thermometer can be calculated readily from the known calibration of the standard thermometer, and, assuming isothermality between the two (which can be checked, at least approximately, by reversing the thermometer positions) it is not even necessary to know precisely (within a few degrees) what the true temperature is!

	  TABLE 1



	TEMPERATURE RANGE        FIXED POINTS REQUIRED      

	DEGREES CELSIUS



        -189 TO 0.01  The triple point of argon

                      The triple point of mercury

                      The triple point of water



        0.01 to 30    The triple point of water

                      The melting point of gallium



        0.01 to 157   The triple point of water

                      The melting point of gallium

                      The freezing point of indium



        0.01 to 232   The triple point of water

                      The melting point of gallium

                      The freezing point of tin



        0.01 to 420   The triple point of water

                      The freezing point of tin

                      The freezing point of zinc



        0.01 to 660   The triple point of water

                      The freezing point of tin

                      The freezing point of zinc

                      The freezing point of aluminum



        0.01 to 962   The triple point of water

                      The freezing point of tin



                      The freezing point of zinc

                      The freezing point of aluminum

                      The freezing point of silver



        -38.8 to +30  The triple point of mercury

                      The triple point of water

                      The melting point of gallium







 TABLE 2



TEMP	ITS-90 RATIO	WA-GA RATIO	WA-GA RATIO 

deg C					 MINUS ITS-90

				 	deg C *



-200	0.169754	0.175975	0.68665

-160	0.342638	0.354401	0.12435

-120	0.511547	0.512280	0.04926

 -80	0.676790	0.676901	0.00969

 -40	0.938436	0.939431       -0.00050

   0	0.999960	0.999960	0.00000

  40	1.158530	1.158535	0.00005

  80	1.315171	1.315181	0.00105

 120	1.469898	1.469913	0.00168

 160	1.622624	1.622744	0.00213

 200	1.733663	1.773688	0.00268

 240	1.922729	1.922758	0.00313

 280	2.069933	2.069967	0.00368

 320	2.215285	2.215323	0.00411

 360	2.358789	2.358832	0.00463

 400	2.500441	2.500488	0.00504

 440	2.640233	2.640285	0.00551

 480	2.778149	2.778205	0.00587

 520	2.914168	2.914229	0.00632

 560	3.048268	3.048332	0.00654

 600	3.180425	3.180493	0.00690

 640	3.310618	3.310691	0.00736

 680	3.438842	3.438909	0.00755		

      

FOOTNOTES

[1]: H. Preston-Thomas, The International Temperature Scale of 1990 (ITS-90), Metrologia 27, 1 (1991)

[2:] NAMAS is the acronym for the National Laboratory Accreditation Service of the United Kingdom. As a NAMAS Laboratory, Isotech is supervised by the National Physical Laboratory of England, and is authorized to issue calibration certificates having legal force in most Nations.

[3]: The triple point of water is inexpensive to acquire and is easily realized by the "slush" method, using an Isotech triple point of water cell and an Isotech Model Oceanus block bath. (We mention the Isotech cell because it is the only cell we know which is available with a certificate of verification).

[4]: Isotech Model 17401 Gallium Melt Point Cell can be used in an accurately-controlled water bath or, most conveniently, in the Isotech Model 17402A Gallium Melt Standard apparatus.

[5]: The computation of the thermometer calibration Table is made using the Isotech MS-DOS Daedalus 1.1 program for ITS(90).

[6]: Isotech Model 815 is an example of such a bath.

[7]: Isotech Model 875 is an example of such a bath. Fluidized baths are thought generally to have the disadvantage of creating dust in the Laboratory. The Isotech bath is entirely free from this problem, because of its unique filtration system.

[8]: For example, Isotech Venus, Jupiter, Gemini, Apollo.

[9]: All the required cells and equipment are available from Isotech.

About the Authors

John P Tavener is founder and Managing Director of Isothermal Technology Ltd., and Thermal Developments International. He obtained his Bachelor's degree at the City University of London, and his Masters in control theory at Cranford Institute of Advanced Technology in Bedfordshire. He is a Chartered Electrical Engineer. He worked at Rolls Royce until 1970, when he became co-founder and technical director of Sensing Devices Ltd. In 1980, he left to devote full time to temperature metrology.