The present invention relates to an apparatus and method for determining the quantity of an incident flux of electromagnetic radiation.

A requirement exists for an apparatus capable of measuring the level of infra-red radiation present within an industrial oven. Such an apparatus could then be used in qualifying an oven to estimate the power requirements when replacing for example, an existing fossil fuel oven with an electrically powered alternative. The apparatus could also be used to investigate the radiative and convective heating mechanisms associated with a particular oven heating process in order to gain a better understanding of that process, to evaluate and optimise oven design and to provide a bench mark to enable comparative measurements between ovens of different types. In addition, the apparatus could also be used to investigate problems such as non-uniform heating of work pieces and to evaluate and optimise infra-red emitter design.

A necessary feature of any apparatus that is to compare radiative fluxes in ovens with differing source temperatures is that it should give the same response with the same radiative flux regardless of the temperature of the source. However many of the commonly available infra-red detectors include photodiodes and other quantum devices which give a differing response for the same flux depending on the source temperature or rather the admixture of quantal energies. For this reason such devices are unsuitable as a basis for a broad band radiation flux measuring apparatus.

Broad band radiation detectors are traditionally based on thermal detectors such as thermopiles, bolometers or pyroelectric devices. One such detector, known as the

Angstrom Compensation Pyroheliometer, is shown schematically in Figure 1. The detector was designed for the measurment of direct solar radiation flux and comprises two manganin strips 10 and 12 which are blackened on their upper surfaces. The underside of the strips 10 and 12 are attached to the junctions of two thermocouples 14 and 16 each of which is connected to a sensitive mirror galvanometer 18. One of the manganin strips 12 is arranged so as to be capable of being heated by an electric current from a battery 20. A variable resistor 22 interposed between the manganin strip 12 and the battery 20 provides a means of regulating the current and thus the amount of heat that is generated. A second galvanometer 24 provides a means of measuring this current.

Within the pyroheliometer the two manganin strips 10 and 12 are disposed at one end of a cylindrical tube which at the other end is provided with two apertured slots, each in alignment with a respective one of the two manganin strips 10 and 12. In operation one of the apertured slots is covered by a shield and the pyroheliometer tube directed toward the sun. As a result one of the manganin strips 10 is exposed to the suns rays while the other strip 12 is shaded. This generates a temperature difference between the two strips 10 and 12 the first of which is heated by the absorption of the incident solar radiation while the second, being shaded, exhibits no change in temperature. As a result an electric current is generated in the circuit of the the thermocouple. If now a current is allowed to pass through the second strip 12 in order to heat it to the same temperature as the first then the quantity of heat released in the second strip 12 will provide a measure of the quantity of solar radiation absorbed by the first strip 10. The absence of a current in the circuit of the first galvanometer 18 determines the moment when the temperature of the two strips 10 and 12 is equal. Using the second galvanometer 24 it is possible to record the size of the

current necessary to obtain a zero reading in the first galvanometer 18. From this reading the quantity of heat released in the second strip Q may be calculated by virtue of the following equation:

Q=i ² R where i is the strength of the current in Amperes and R is the resistance of the strip in Ohms.

At the same time the quantity of heat received by the strip 10 by virtue of the absorption of solar radiation may be determined by the relation:

Q = S5A

where S is the flux of direct solar radiation, δ is the absorbtivity of the surface of the strip and A is the surface area of the strip.

Thus on the basis of these relationships it is possible to obtain the following expression for direct solar radiation flux:

S = i ² R 5A

Unfortunately the Angstrom Compensation Pyroheliometer does not produce fully reliable results as the above theory does not take into account certain factors that influence the Pyroheliometer readings. As a result the readings obtained by each instrument must be compared and adjusted so as to conform to an International standard.

In particular the Angstrom Compensation Pyroheliometer possesses a systematic error caused by a so called "edge effect" as a result of the exposed strip not being heated uniformly by the absorption of solar radiation. This occurs because its edges are partly shaded by the instruments diaphragm and because the radiation is absorbed by a thin surface layer of the strip whereas the

shaded strip undergoes what is in comparison an even heating by the electric current. Thus the equalisation temperatures of the strips 10 and 12 as observed by means of the first galvanometer 18 actually occurs when Q<S«_A which leads to a lowering of the Pyroheliometer readings.

The adaptation of the Angstrom Compensation Pyroheliometer for use within an industrial oven is clearly no easy task, not least by virtue of the limited solid angle through which the incident radiation must pass before it can be absorbed by the first strip 10. By simply ignoring the radiation that might otherwise impinge upon the first strip 10 from a shallow angle the instrument would not be able to measure all the flux available for heating at a particular surface.

Another thermal detector, known as the Yanishevsky Pyranometer, is shown in Figures 2 and 3. Unlike the Angstrom Compensation Pyroheliometer, the receiving surface in this detector comprises a system of consecutively connected manganin-constantan thermoelements. The thermoelements are disposed either in a rectangular arrangement or radially from the centre of a flux receiving surface, respective opposite ends of the same strip being either sooted black or whitened with magnesia. In this way the receiving surface has the appearance of a black and white chequer board. A hemispherical glass cover prevents the occurrence of wind effects. When irradiated the black and white junctions create a temperature difference approximately proportional to the incident radiation flux. The Yanishevsky Pyronometer is however a relative instrument and so must be calibrated to an International standard. Like other pyronometers the Yanishevsky Pyronometer displays an angular dependance of sensitivity and also a pronounced selectivity of sensitivity for radiation of different wavelengths. This necessitates the introduction of so called angular and spectral corrections

to the readings in the measurement of diffuse radiation which take into account the differences between the anngular and spectral distributions of diffuse and direct solar radition intensity. Again it is not apparent how the Yanishevsky Pyranometer could be adapated for use within an industrial oven.

It is an object of the present invention to provide an apparatus that is capable of measuring an incident flux of radiation and which does not suffer from the drawbacks associated with the devices of the prior art. In particular it is an object of the present invention to provide an apparatus capable of measuring an incident flux of infra-red radiation within an industrial oven.

According to a first aspect of the present invention there is provided an apparatus for determining the quantity of an incident flux of electromagnetic radiation comprising first and second bodies having different coefficients of interaction with the incident radiation, means to monitor the temperatures of said bodies, heating means associated with at least one of said bodies, control means responsive to the monitored temperatures to control said heating means so as to minimize temperature differences therebetween and to provide a signal representative of the power delivered to the heating means, and computing means responsive to said power signal to compute a value for the quantity of the flux of electromagnetic radiation incident upon the bodies.

Advantageously a second and independent heating means may be provided in association with the other of said two bodies. Preferably the or each heating means may comprise a resistive heating element in intimate contact with an undersurface of the body concerned.

Advantageously said first and second bodies are substantially planar.

Advantageously said first and second bodies are mounted on a thermally insulating supporting member and provided with means to monitor the flow of heat through said thermally insulating supporting member.

Advantageously said first of the bodies may be provided on an exposed surface thereof with a material capable of reflecting a substantial proportion of the radiation incident thereon within a selected range of wavelengths and said second of the two bodies may be provided on a corresponding exposed surface thereof with a material capable of absorbing a substantial proportion of the radiation incident thereon within said selected range of wavelengths.

Advantageously said first and second bodies may be contained within a sensor unit remote from said control means and coupled thereto by means of an umbilical cable.

According to a second aspect of the present invention there is provided a method of determining the quantity of an incident flux of electromagnetic radiation comprising the steps of providing first and second bodies having different coefficients of interaction with the incident radiation, exposing said bodies to the incident flux of electromagnetic radiation, measuring the temperatures of said bodies, heating the colder of said two bodies so as to minimize any temperature difference therebetween, measuring the quantity of power necessary to so heat said colder of the two bodies and calculating a value for the quantity of the flux of electromagnetic radiation incident upon the bodies on the basis of said measured values.

Advantageously the method may comprise the additional step of computing the total heat flux arriving at said first or second bodies. Where said bodies are mounted on a thermally insulating supporting member, the additional step of computing the total heat flux arriving at said first or second bodies may include measuring the quantity of heat flowing through the thermally insulating supporting member.

An embodiment of the present invention will now be described by way of example with reference to figure 4 of the accompanying drawings in which there is shown a schematic plan view of an apparatus 110 for measuring an

incident flux of radiation comprising a sensor unit 112 and a controller 114 interconnected by a heat-resistant umbilical cable 116. Within the sensor unit 112 two bodies 118 and 120 are mounted on a thermal insulation board 122. The two bodies 118 and 120 are adapated in a manner to be described so as to have substantially the same thermal mass whilst at the same time having different coefficients of interaction with an incident flux of radiation. In particular the bodies are adapted so that a first of them 118 is capable of reflecting up to 96% of an incident flux of infra-red radiation while the second 120 is capable of absorbing up to 98% of the same incident flux.

A first heating element 124 is provided adjacent an under surface of the first of the two bodies 118 between the body and thermal insulation board 122 while a second heating element 126 may optionally be provided adjacent an under surface of the second of the two bodies 120, again between the body and the thermal insulation board 122. First and second temperature sensors 128 and 130 are also located adjacent the respective under surfaces of the two bodies 118 and 120 and their connections 132 and 134 passed through the thermal insulation board 122, together with those of the or each heating element 136 and 138, from whence they pass through the heat-resistant umbilical cable 116 to the controller 114.

In use the sensor unit 112 is located within a flux of radiation. This flux is caused to be incident upon the two bodies 118 and 120 and as a result both bodies heat up, although naturally the more absorbing of the two bodies 120 heats up faster than the more reflecting body 118. During this process the temperatures of the two bodies 118 and 120 are continually monitored by the controller 114 by means of the temperature sensors 128 and 130. In response to a temperature difference between the two bodies 118 and 120 the controller 114 calculates the amount of electrical

power that is required to be fed to the first heating element 124 in order to raise the temperature of the more reflecting body 118 to that of the more absorbing body 120. In this way the temperature difference between the two bodies is minimised enabling their respective temperatures to increase at the same rate. Knowing the temperatures of the two bodies 118 and 120 and the quanitity of the electrical power delivered to the first heating element 124 it is possible to derive a value for the radiative heat flux of the incident radiation. Thus whilst the heat flux incident upon either body may be expressed as:

Te

Total heat flux = H- (1 ⁾ dt

where H is the thermal capacity of the body concerned, T is its temperature and t is the quantity of time that has elapsed, a thermal balance equation for the second and more absorbing body 120 may be written as:

± dT _B α _B F-Aε _B σTg _ C(T _B -T _A ) ₌ ^R — radiative exchange convective loss thermal inertia (2),

Similarly, a thermal balance equation for the first and more reflecting body 118 may be written as:

1 dT _G P _{+ G} F-Aε _G σT _ C(T _G -T _A ) ³ ₌ ^H ~ power in radiative exchange convective loss thermal inertia (3),

where

T G_ = temperature of the more reflecting body 118

T_ B, = temperature of the more absorbing body 120 - = temperature of oven air

P = electrical power fed to the heating element 124

F = total radiation flux incident on the two bodies 118 and 120 A = surface area of each of the bodies 118 and 120 exposed to the radiation flux H = heat capacity of each of the bodies 118 and 120 C = unknown empirical factor for convection α G_ = absorptance of the more reflecting body

118 α_ B = absorptance of the more absorbing body 120

C Gm, = emittance of the more reflecting body 118 ε„ B = emittance of the more absorbing body 120

Subtracting equation (3) from equation (2) gives:

d(T _B -T _G )

- P+ (α _B - _G )F- (ε _B T - e ₀ T#Aσ- C (T _R -T _A ) ³ - (T _G -T _A ) = H- dt

(4).

By ensuring that the temperatures of the two bodies 118 and 120 are the same, the term multiplying the unknown impirical factor C in equation (4) exactly vanishes. Thus under these conditions equation (4) can be re-expressed as:

Using equation (5) the controller 114 may compute a value for the incident radiation flux F from the measurable variables P, T _G and T_.

From equation (5) it can also be seen that under ambient conditions when T = T„ = 296 k and ε = α_ D and c_ = _ the flux F is simply given by the equation:

F = Aσ m

A

Considering the apparatus 110 now in more detail, the two bodies 118 and 120 are designed so as to be substantially planer in nature, each having a substantially uniform flat upper surface 140. In this way the apparatus 110 may be used to provide a measurement of the radiative transfer to a plane. Thus in use, the sensor unit 112 may be placed in any desired orientation provided of course that the two bodies 118 and 120 are at the same horizontal level. This ensures that heat is not convected from a lower to an upper of the two bodies as this would then invalidate the assumption that the bodies have identical convective gains or losses when maintained at the same temperature.

The area of the upper surfaces 140 of the two bodies 118 and 120 is not critical again provided of course that it is the same for each of the two bodies. Ideally in use the sensor unit 112 should be positioned in such a way that the two bodies 118 and 120 might be expected to receive identical radiative and convective heat fluxes. By providing the upper surfaces 140 of the two bodies 118 and 120 with a surface area which is small in comparison to the surface area of the radient heat panels typically found in industrial ovens this condition is usually satisfied. Thus in a preferred embodiment the upper surfaces 140 of the two bodies 118 and 120 are each approximately 1" (2.54cm) square.

In order to provide an apparatus 110 in which the more absorbing of two bodies 120 is heated at a rate which is representative of the rate at which a typical product

might be heated the thermal mass per unit area of the two bodies 118 and 120 is preferably not too dissimilar from that of products commonly heated in infra-red ovens. This would suggest that the bodies 118 and 120 should be made of a metal such as steel having a heat capacity of the order of 4.0 x 10 3Jm—2K—1. Other considerations however must also be taken into account in determining the material from which the two bodies 118 and 120 are to be formed such as the need for the two bodies 118 and 120 to be efficient conductors of heat. In this way the two temperature sensors 128 and 130 will not significantly reduce the local temperature of the bodies and nor will there be created a temperature variation through the thickness of the bodies in excess of an acceptable limit.

To understand the factors affecting the choice and thickness of the material of which the two bodies 118 and 120 are to be comprised it will be helpful to consider the effect on the computed flux arising from an inexact knowledge of the quantities that are constantly monitored by the controller 114, namely the power P and the temperatures T„ and T _Q .

Considering first the Power term P, in practice there will of course be fluctuations in the measurement of the power delivered to the or each heating element 124 and 126 corresponding to noise on the associated voltage and current measurements. Whilst the apparatus 110 is clearly designed to reduce this noise, another problem that is inherent in the apparatus and that causes fluctuations in the computed flux is the inevitable time delay between power being applied to say, the first of the heating elements 124, and a temperature change being registered by the associated temperature sensor 128. Furthermore it will be understood this time delay is not of a fixed interval as the temperature rise associated with a pulse of power does not occur at a single instant later in time. In order to

reduce this delay and so improve the sensitivity of the bodies 118 and 120 to pulses of power from the controller 114 it would clearly be advantageous to limit their thermal mass.

Another source of fluctuations in the computed flux arises as a result of fluctuations in the measured temperatures T _R and T _G . In a preferred embodiment of the present apparatus the two temperature sensors 128 and 130 each comprise a thermocouple however, as is well known, the measurement of temperature using a thermocouple involves the measurement of small voltages superimposed on which may be significant amounts of electrical noise. Even after averaging the signal voltage over several 50 Hertz mains cycles fluctuations in the measured temperature are not usually reduced below 0.2°C. Whilst the magnitude of the fluctuation in the power and hence the computed flux arising from a 0.2°C fluctuation in one or both of the measured temperatures T _β and T_ depends, in part, on the control algorithm executed by the controller 114, it will be apparent that the resulting power fluctuation ΔP is directly proportional to the heat capacity of the two bodies 118 and 120 since by virtue of equation 1:

ΔP = H AT At

Thus this source of error would also suggest the benefits of a reduction in the heat capacity of the two bodies. One way of reducing the heat capacity of the two bodies 118 and 120 for a given material would be to reduce their respective thicknesses. However another source of error associated with temperature measurement increases with a reduction in the thickness of the two bodies. This

error arises whenever there is a temperature variation within either of the bodies 118 and 120 and the respective temperature sensors 128 and 130 are not positioned so as to record an average temperature for the body as a whole. This is particularly true of the more reflecting of the two bodies 118 since the more absorbing body 120 will normally receive uniform heating and so one would not expect significant temperature variations across it. The more reflecting body 118 however is, in part, heated by the first heating element 124 which may take the form of a resistor comprising a number of strip elements separated by gaps. Thus for this body in particular there will be a temperature variation between those parts of the body close to one of the strip elements and those parts close to one of the gaps whenever power is being dissipated in the resistor. If the thickness of the body were to be reduced the thermal resistance along the body for heat diffusing from the strip elements would be increased thereby exacerbating the problem and preventing the body from achieving a uniform heat distribution.

The choice and thickness of the material from which the two bodies 118 and 120 are to be formed is therefore a ^' balance between conflicting factors. In a preferred embodiment however the two bodies 118 and 120 are each comprised of aluminium nitride and have a thickness of 0.025 inches (6.35 x 10 -4m) . The advantage of aluminium nitride is that it combines the properties of being both a good electrical insulator and a good thermal conductor. At approximately 165W/mK aluminium nitride has a thermal conductivity almost ten times that of stainless steel which enables the temperature profile across each of the bodies

118 and 120 to be as even as possible whilst ensuring that any heat applied from either of the heating elements 124 and 126 is dispersed quickly and efficiently. Aluminium nitride also exhibits excellent thermal shock properties enabling the apparatus described to be designed to

withstand 5000 cycles to a temperture 200°C in excess of the maxiu um anticipated oven temperature, which for most finishing ovens is typically of the order of 350°C.

In order to provide the two bodies 118 and 120 with different coefficients of interaction with an incident flux of radiation their upper surfaces 140 are coated with a suitable reflecting or absorbing material.

Considering first the more reflecting body 118, it is important that the reflective coating has a reflectance close to 1.0 over a wavelength range for the incident radiation from 1x10 —6m to 9x10—6m. Gold is widely recognised as having the best reflecting properties within this wavelength range. For this reason in a preferred embodiment the upper surface 140 of the more reflecting of the two bodies 118 is provided with a coating of gold ink of the type more usually used in the fabrication of conductor tracks in thick film hybrid circuits.

By contrast the best infra-red absorbing materials within the wavelength range from 1x10 —6m to 9x10—6m tend to be based on either black organic paints or copper oxide. Whilst copper oxide offers the best high temperature stability, for ease of construction in a preferred embodiment the upper surface 140 of the more absorbing of the two bodies 120 is provided with a layer of black ink of the type normally used to fabricate thick film resistors.

Turning now to the fabrication of the heating elements 124 and 126, the importance of ensuring that the power dissipated in the heating elements is quickly and efficiently dispersed to the bodies 118 and 120 as has already been highlighted. For this reason the heating elements 124 and 126 are preferably mounted an under surface of the bodies concerned in such a way as to be in intimate contact therewith. One advantage of fabricating the bodies 118 and 120 from an electrically insultaing material such as aluminium nitride is that the two heating

— —

elements 124 and 126 may simply comprise a resistive layer which is printed or otherwise applied to the under surface of the body concerned. In this way the thermal lag previously identified is minimised thereby simplifying the task of the controller 114 in minimising any temperature differences between the reflecting and absorbing bodies 118 and 120. As a result the apparatus as a whole 110 may respond more quickly to rapid changes in the incident flux while on a more local level thermal runaway within the heating elements is prevented as a result of some parts of the strip element becoming hotter than others.

In a preferred embodiment the resistive layer which is applied to the under surfaces of the two bodies 118 and 120 is based on a silver/palladium mixture which is normally used within thick film circuits as a conductor or termination having high solderability. In the present application however the mixture is chosen for its higher than normal resistivity and its lower than normal Temperature Coefficient of Resistivity of between approximately 300-400ppm. As a result the resistance value of the layer does not increase too rapidly with temperature thereby facilitating the adaptation of the apparatus 110 for use with battery power where the voltage available to drive current through the heating elements 124 and 126 is limited.

Whilst it will be apparent to those skilled in the art that a heating element need only be provided in association with the more reflecting of the two bodies 118, the provision of a second heating element 126 in association with the more absorbing of the two bodies 120 does provide additional advantages. Firstly, it does of course mean that the two bodies are of substantially similar construction, differing only in the coatings that are applied to their upper surfaces 140, thereby minimising any differences in their respective heat capacities. Secondly however, by providing a second heating element 126

in association with the more absorbing of the two bodies 120, the apparatus described is able to compute a value for the incident flux after the source of that flux has been removed. In this situation both bodies 118 and 120 will initially be at some elevated temperature and will begin to cool, the more absorbing of the two bodies 120 cooling faster than the more reflecting body 118. Under these circumstances, the controller 114 operating under the algorithm so far described will look to take energy from the reflecting plate 118 rather than adding to it. However instead of trying to cool the reflecting plate 118, by providing a second heating element 126 it is possible to heat the more absorbing body 120 so as to once more reduce the temperature difference between the two. All that is required is to adapt the controlling algorithm in such a way that the controller 114 interprets a negative power as a signal to deliver the magnitude of that power to the more absorbing body 120.

In a preferred embodiment the electrical connections 136 and 138 of the two heating elements 124 and 126 are each fabricated from a length of gold tape 0.002 inches

(5.08 x 10 -5m) thi.ck by 0.025 i.nches (6.35 x 10-4 ) wide. The gold tape is attached directly to the silver/palladium layer using a thermo-compression technique and is capable of carrying up to a 2A heater current at temperatures up to 500°C. Alternatively the gold tape may be attached to a small gold connecting pad provided at each end of the silver/palladium layer. By using a thin gold tape in this way the mechanical stresses in the connections

124 and 126 are kept small in relation to the contact area of their respective joints. A further advantage of this type of connection is that the gold is resistent to oxidation in air at the temperatures involved.

As has been previously stated, each of the two temperature sensors 128 and 130 preferably comprise a thermocouple. In particular, the two temperature sensors

128 and 130 preferably comprise a matched pair of 0.5mm diameter Inconnel sheathed type K thermocouples each of which is cemented into a blind hole drilled in the centre of the under surface of the bodies concerned using a high temperature adhesive. Alternatively one or both of the temperature sensors 128 and 130 may comprise a resistive temperature sensing element printed directly on to the under surface of the two bodies 118 and 120. As a consequence this alternative arrangement would simplify the manufacturing of the apparatus whilst at the same time improving the thermal response of the computed flux measurement.

As has been previously described, the connections 132 and 134 of the two temperature sensors 128 and 130 pass through the thermal insulation board 122, together with those of the or each of the heating elements 136 and 138, from whence they pass out from the sensor unit 112, through the heat resistent umbilical cable 116 to the controller 114.

Since it is intended that the apparatus 110 should only measure the radiative heat flux incident upon the upper surfaces 140 of the two bodies 118 and 120 and since the apparatus 110 is to be used in ovens which typically have radiant panels on each of their internal surfaces, it follows that the two bodies 118 and 120 should preferably be thermally isolated from a flux of radiation that would otherwise be incident upon their respective under surfaces. For this reason the rear face and sides of the sensor unit 112 are preferably made of a reflective metal having a sufficient thickness to provide the walls of the sensor unit 112 with a reasonable thermal conductivity in their lateral directions. In this way the high reflectance will reduce the amount of heat that is conducted to the internal surfaces of the sensor unit 112 while the high conductivity will ensure that different parts of the sensor unit 112 are not at widely differing temperatures.

In addition to measuring the radiation flux the apparatus 110 may also be adapted to measure the total heat flux incident upon the two bodies 118 and 120. In order to achieve acurate measurements however it is clearly important that the two bodies 118 and 120 be thermally isolated as far as possible from the walls of the sensor unit 112. Whilst it is thought that an air gap would probably provide satisfactory thermal isolation, construction considerations suggest the easiest way of mounting the two bodies 118 and 120 is by using a thermal insulation board 122. Even using the best thermal insulation boards however the quantity of heat conducted away from the two bodies 118 and 120 and through the board 122 can remain a significant correction to the total heat flux measurement. For example, where a microporous thermal insulation board is used having a thickness of 2cm, three minutes after the sensor unit 112 is exposed to a flux of

20kWm -2 a heat flow of 2kWm-2 may be measured from the bodies 118 and 120 to the insulation board 122. This heat flow if not corrected would give rise to a 10% error in the computed total heat flux measurement. In order to take account of this heat flow it is necessary to measure the temperature of the under surface of the thermal insulation board 122. To this end the sensor unit 112 may include a third temperature sensor 142 which, like the first two temperature sensors 128 and 130, may comprise a thermocouple whose connections 144 are fed to the controller 114 through the heat resistent umbilical cable

116.

Having knowledge of the thermal conductivity and specific heat of the thermal insulation board 122 it is possible to estimate the heat flow to the insulation board to within 10% by means of a one dimensional heat flow calculation. Under these circumstances the error in the computed total heat flux may be reduced to as little as

1%. Thus, once provision has been made to calculate the

heat flow through the thermal insulation board 122 it is no longer necessary that the board be chosen so as to have the lowest thermal conductivity. Thus, whilst microporous thermal insulation boards generally have the lowest thermal conductivities, their mechanical strength limits their usefulness as a construction material. By contrast ceramic fibreboards such as "Morganite Strong Board" are more suitable for the mounting of the two bodies 118 and 120 even though its thermal conductivity at 300°C is

0.075Wm —1K—1 which is about three times higher than that of microporous thermal insulation board at the same temperature.

In order that the one dimensional computation of the heat flow through the thermal insulation board 122 approximates to that which is actually observed the insulation board 122 on which the two bodies 118 and 120 are mounted preferably extends laterally beyond the bodies by an amount greater than the thickness of the board. Furthermore, the area of the thermal insulation board 122 at the front of the sensor unit 112 and around the two bodies 118 and 120 is preferably covered with a metal plate having a similar thermal capacity to that of the bodies themselves and is preferably coated with an absorbing material.

Whilst the two bodies 118 and 120 are preferably thermally isolated from the back of the sensor unit 112 it will be apparent that it is not essential that they are thermally isolated from the front. Having said that however by thermally isolating the two bodies 118 and 120 from the front of the sensor unit 112 it is possible to determine their respective heat capacities and to calibrate the difference in the emissivity of their two coatings by electrically heating the bodies using the heating elements 124 and 126. However since this heating need only be through a few tens of degrees the two bodies may simply be isolated from the front of the sensor unit 112 by a gap of

approximately 4mm. At the same time it is to be noted that the possibility of calibrating the two bodies 118 and 120 in this way arises as a result of their both being provided with a respective heating element which is a further reason for providing the more absorbing body 120 with its own heating element 126.

In a preferred embodiment the heat resistant umbilical cable 116 interconnecting the sensor unit 112 and the controller 114 comprises a stainless steel flexible conduit of typically 4 metres in length. By contrast the controller 114 which is preferably battery powered and microprocessor based is enclosed within a housing capable of withstanding an external temperature not exceeding 300°C for at least 30 minutes. In this way should the need arise, both the sensor unit 112 and the controller 114 may be passed through an oven on a continuous conveyor. On other oaccasions, provided the umbilical cable 116 is long enough, only the sensor unit 112 need be placed in the oven enabling the controller 114 to remain outside.

As has been previously described, the function of the controller 114 is to take measurements from the sensor unit 112, to calculate and provide a controlled amount of electrical power to one or other of the two bodies 118 and 120 to minimise the temperature differences between the two, and from these measurements to compute a value for the radiative incident flux. To this end the controller 114 preferably comprises a microprocessor, interface electronics, a suitable storage medium and means for down loading the data stored therein for further processing. Advantageously the controller 114 may also incorporate a display means so as to enable a user to analyse the data stored in the storage medium without having to first download it.

In order to allow for the inherent time delay previously identified between power being applied to one or other of the heating elements 124 or 126 and a temperature

change being registered by the associated temperature sensor 128 or 130, the controller 114 preferably operates under the control of an algorithm that in response to a detected temperature difference between the bodies 118 and 120 delivers a slowly varying power to the cooler of the two bodies. During this period the flux computation is derived from a weighted average of the applied power levels. A suitable weighting function associating temperature rise to the application of power might be derived from the temperature record after a short duration pulse of power. For example the weighting function can be regarded as the proportion of the pulse that is realised as a temperature rise in time intervals after its application while the time to realise half the energy of the pulse can be defined as the delay time. Clearly the delay time will depend on the thermal resistance between the heating element and the associated temperature sensor. By contrast the weighting function may be established from a state in which both the bodies 118 and 120 and the thermal insulation board 122 are at the same temperature. Whilst the weighting function thus derived cannot be expected to be exactly appropriate when the temperature of the sensor unit 112 is different, the small extent to which the weighting function becomes inappropriate can be regarded as an uncertainty in the average delay time associated with non-repeatability.

From this it will be apparent that 1% repeatability when the power term P is significant will not be achieved unless the power level varies by less than 1% during a time equal to the uncertainty in the delay time. However when the flux changes by a substantial step the power delivered by the controller 114 will need to change by a similarly substantial step in order to minimise the temperature difference between the bodies 118 and 120. Following such a step change there will inevitably be a time period during which the power is changing too rapidly for two digit accuracy to be maintained. The length of this period

however depends on the control algorithm and ultimately on the delay time. For this reason the control algorithm preferably comprises a three term function incorporating both integral and differential elements. In this way the difference in temperature between the two bodies 118 and 120 may be critically damped so as to prevent an excessive oscillation in the temperature of the body to which power is being delivered both above and below that of the other of the two bodies.