Reprinted from Industrial Heating, September, 1995

Editor's Note: This novel pyrometer addresses the range from 500 to beyond 3500°C. It is claimed to require no prior knowledge of the target, including the emissivity, to yield extremely accurate temperature determinations. It provides a unique on-line tolerance, a gauge of the accuracy of each measurement. This feature allows the users to ascertain that the measurement has been successful or alerts them to interfering conditions.

Radiation pyrometry has many advantages over other temperature measurement techniques: it's fast, it doesn't require contact, it doesn't affect the temperature that's being measured, and it needs no consumables. For temperatures above about 1800C/3300F there isn't really a practical alternative. The pyrometer discussed in this article is particularly of value in difficult applications which have traditionally defied solution. These include two broad classes: (1) where the target's emissivity is unknown or changing, and (2) where the environment interferes. A third, smaller class may be added: applications where the temperature is so high that no source of known temperature exists to calibrate the instrument. In many of these applications the reproducibility of some pyrometers has allowed their "temperature" outputs to be correlated with product quality through trial and error. As long as nothing changes, the correlations hold. However, change is inevitable: process scale-up, product and process development, commissioning of new capital equipment, automation, and operator turnover all undermine the reliability of the carefully-developed correlations. Reproducibility is not enough when the target and environment change. Accuracy is needed for the knowledge base to carry over into the new situation. Without it, one may be in the position of the engineering manager who said "every time we change the product 10% we have to do 70% of the engineering over."

A new type of pyrometer, developed and tested in a broad spectrum of industrial applications over the last 10 years, is capable of solving these problems. In the course of its testing it has demonstrated these attributes:

• Independence from knowledge of target emissivity
• An on-line gauge of accuracy (a tolerance describing how well the indicated temperature is known)
• Extreme accuracy and reproducibility accuracy of 0.1% reading is readily achievable, reproducibility of 0.01% is routine
• Unlimited high temperature response
• Ability to cope with absorbing or emitting atmospheres
• Noise immunity
• Adaptability to many situations
• Broad temperature response in one package

CAUSES OF DIFFICULTY

Making a successful pyrometric measurement requires allowance for emissivity and environmental interference. Emissivity, the ratio of the emitted radiation of a real radiator to that of an ideal one, can depend on composition, surface finish, mechanical and thermal history, and the wavelength where the measurement is made. It can also depend on the temperature; then the problem becomes a circular one, impossible to resolve. Environmental interference often is in the form of absorption or emission of thermal radiation by the material between the instrument and the target. This includes both the atmosphere and any optical elements in the path. If the detector is sensitive to the wavelengths where this occurs the result will be affected.

The overwhelming majority of pyrometers available are of two types: brightness or ratio. Brightness devices rely on capturing a known fraction of the energy emitted by the target; the user must know the emissivity to get the correct temperature value. For the reasons shown above this can be impossible.

Ratio pyrometry attempts to circumvent the emissivity issue mathematically. Intensities are measured at two different wavelengths and divided. The resulting representative equation is solved for temperature and the hope is that the division has canceled out the emissivity. This method works if the emissivity is the same at both wavelengths, but this is only certain in an ideal or semi-ideal (greybody) radiator. Concern over emissivity cancellation affects the design of ratio pyrometers: the closer together the wavelengths are chosen, the more likely the emissivities are to cancel; the farther apart, the larger the magnitude of the resultant signal, and the greater the precision.

These problems of relative and absolute emissivity as well as the separate issue of environmental interference are addressed by a new pyrometer. It can detect and avoid the influences of emissions and absorptions in the thermal spectrum and determine the value of the emissivity. It does both of these with no prior information about the target or its environment.

THE NEW MULTI-WAVELENGTH PYROMETER

The idea for this device grew from the frustration developed when several existing pyrometers were used to measure a process temperature. Each gave a different answer. We speculated that if a huge number of instruments could be used and a large fraction of those agreed among themselves, then the result should be trustworthy. It turned out to be not quite that simple. After 10 years, the resulting "tolerant" pyrometer has solved the problems of difficult applications with a variety of targets and environments. These include ceramics, glass, composite materials, metals, combustion, chemical vapor deposition, etc.

INDEPENDENCE FROM KNOWLEDGE OF TARGET EMISSIVITY

The independence from emissivity is a result of the pyrometer's most unusual feature: an on-line gauge of each measurement's individual accuracy. Its effect is that each measurement is reported as a value plus or minus its tolerance. Tolerances are typically less than 0.5% of the reported value. A tolerance in this range indicates that the measurement is complete; the material of the target does not even have to be known. In the rare cases when the tolerance grows large, it indicates that the measurement was less successful than the norm. If the resulting value is not accurate enough for the process under investigation then there is more work to be done to overcome the factors affecting the result. The nature of the instrument is such that the information from an unsatisfactory measurement points the way to making a satisfactory one.

An example illustrates how "information" with a long history can be refuted. Fig. 1 shows a refractory ceramic material being cast. Manufacturing lore was that the material was semi-transparent. Many pyrometers had been tried without agreeing among themselves or impressing the operating personnel. Rather, the operators had become convinced that pyrometry had no future in their facility. Measurements with the new multi-wavelength pyrometer presented a different picture. Temperature values agreed with mass balance and eutectic point calculations (to the accuracy of the calculations). Reproducible temperature values and gradients were found, along with an interesting artifact. The artifact can be seen as the streamline in Fig. 1, delineated by the two dark boundaries and representing a stream of locally colder material. The measured values showed that stratification of temperatures in the melt explained the phenomenon. As a result, confidence in pyrometry was restored. Finally, since the value of the emissivity had been determined, the manufacturer was routinely able to measure the melt temperature.

Figure 1

Liquid ceramic being cast. The streamline indicates a cooler flow existing within the larger one. The temperatures measured explained that the source of the cooler material was stratification within the melt.

EXTREME ACCURACY, UNLIMITED HIGH TEMPERATURE RESPONSE

Every pyrometer must be calibrated against some standard. This usually means running some known source of temperature through a range of temperatures and in some fashion forcing the output of the device under test to agree with the standard. This limits the accurate range of the device to pretty much the range that can be reproducibly achieved in the laboratory.

The new pyrometer avoids this problem by using an unusual calibration method. It is not adjusted to a series of chosen temperatures, which would then require that all others be interpolated or extrapolated. It is, however, tested against selected temperatures in its intended range after the calibration. In this way two different standards are always employed in the calibration routine; if one has drifted since its last use this fact will immediately be caught.

As a result of the novel calibration we use, the new pyrometer has accurately measured temperatures up to 3500C. These ultra-high temperature values were determined to be correct by the internal gauge of accuracy and consensus with two commercial devices of different manufacture at the highest temperature they reached. Error analysis has shown that our technique is 10 times as accurate as a corresponding temperature calibration. Accuracy of 0.1%, as measured against an NIST-traceable blackbody source, has been achieved. Reproducibility of 0.01% is routine.

ABILITY TO COMPENSATE FOR INTERFERENCE (EMITTING, ABSORBING ATMOSPHERES)

A major use of high temperature in industry is to make chemical changes in a workpiece. Unfortunately this often results in driving off gases that can interfere with the optical determination of temperature. Data from a case in point are shown in Fig. 2. Note the substantial absorption centered around 625 nm and another at 770 nm. If the manufacturer were to use a vanishing filament pyrometer (which operates at 650 nm) or any brightness or ratio pyrometer whose range of sensitivity includes the affected wavelengths, the determination would be incorrect. But as is illustrated, the technique described here extracted the correct temperature in spite of the absorptions. Proof was obtained by purging the absorbing gas from the sight path: the original temperature determination, with the offgas, was 3253.7 ±45.3°C. After the offgas was purged from the field of view the determination was 3246.4 ±20.9°C. The difference between the before and after temperatures, 8°C, is insignificant. If a vanishing filament pyrometer had been used in this case, the difference in its indications before and after the purge would have been 410°C (presuming it gave the correct indication of 3250°C with the purge, it would have indicated 2840°C without it).

The absorptions observed in this environment were a function of temperature, as might be expected with the various vaporization points of different species. This means that unless the locations of all the absorptions are known, the choice of an appropriate standard pyrometer for similar applications will depend entirely on luck.

Figure 2

The data show a substantial absorption centered at 625 nm, and another at 770 nm. The absorptions are a function of temperature. Without knowing this a pyrometer sensitive in these regions could easily be used with incorrect results. Despite the absorptions, the new pyrometer measured the temperature accurately; this was verified later by purging the offgas.

Not all interfering environmental effects stem from target offgas. Fig. 3 shows a weld seam almost completely obscured by steam.

Figure 3

In this pipe mill steam almost obscures the weld seam - the temperature measurement target.
The pipe seam produced by the welder (Thermatool) is located just to the left of the vertical cylinder in the center foreground. Radiation pyrometer measurements must also contend with dust, smoke, flame, and target offgas.

IMMUNITY TO NOISE

A noisy signal, distorted by optical or electrical interference, can also cause problems. Andreic reports that noise of as little as 1% can render multi-wavelength techniques unsuitable due to the large error that results. He observed a 5% error resulting from 1% noise. The following example shows that noise of 10% has a small effect on the new measuring technique. Fig. 4 represents the data collected in an industrial environment, complete with small absorptions at 588 and 765 nanometers; Figure 5 shows the same data with 10% noise added. The temperature calculated for the original data was 2887.0 ±10.3°C; after the noise addition the calculated temperature was 2899.7 ±23.2°C. The error resulting from the 10% added noise is 0.4%. Interestingly, the tolerance has increased 12.9°C, accurately predicting the 12.7°C change in indicated temperature.

Figure 4

Data collected in an industrial environment. Temperature calculated was 2887.0 ±10.3°C. Noise is added to these data to create Figure 5.

Figure 5

The same data as in Figure 4 with 10% noise added. The temperature determined is 2899.7 ±23.2°C. The error resulting from the noise addition is 0.4%. It was accurately predicted by the increase in the tolerance from 10.3 to 23.2°C. Others have reported error of 5% resulting from 1% noise.

ADAPTABILITY/BROAD TEMPERATURE RESPONSE IN ONE ELECTRONIC UNIT

The unusual way of calibrating makes it easy to adapt various optics to the pyrometer. Everything from bare fiberoptic cable to commercial camera lenses has been used to get a view of the intended target. The different optical pickups can even be fitted in the field if desirable. The adaptability of the methodology can be shown by a list of some successful applications. These applications, Table 1, used the basic electronic unit with a variety of optical inputs:

Table 1 Variety of Successful Applications of New Multi-Wavelength Pyrometer

• ceramic manufacturing
• glass processing
• fiber drawing
• chemical vapor deposition and infiltration
• carbon and carbon/carbon manufacture and testing
• combustion
• ultra-high temperatures (>3300C/6000F)
• exotic metal alloys (oxides vary with thermal history)
• reflective environments (temperature gradients inside heated objects)

The temperatures measured in these applications ranged from 500 - 3500C. This does not represent the upper limit of temperature; as we mentioned earlier, high temperature response is essentially unlimited.

SUMMARY

A new multi-wavelength pyrometer has been used to solve formerly intractable problems. Unknown or changing emissivities and environmental interferences have been readily overcome. The new instrument, which is suitable for temperatures ranging from 500 to beyond 3500°C (900 - >6300°F), is capable of exceptional accuracy to within 0.1%. It requires no prior knowledge of either the target or the environment. It provides a tolerance for each temperature measured so the operator knows that the measurement was successful. In some cases standard instrumentation can be used after the new pyrometer measures the emissivity and locates the environmental interferences. In our experience problems of long standing can be resolved rapidly when the new instrument is used.