Previous Section: Introduction / Background
Next Section: Results

 

Methods / Experimental (TOC)

Construction of a Blackbody (TOC)
 In considering the design of a thermal radiation source, one must take into account its dimensions, material make up, specific heat, method of heating and cooling, technique for assembly, electrical characteristics, and emissivity of the surface. The radiation source designed in this research was 7 in2 and approximately ¼ in thick. With these dimensions the source is able to fit adequately above the scan mirror in the backscan of the MISI. In this research only one blackbody was constructed. It is desired to have two such radiators in the MISI with their placement shown in Figure 3.1.

      Figure .1 Placement of blackbody radiators used for calibration of the MISI.

       

The blackbodies need to be made of a conductive material that exhibits a relatively low specific heat. In this way the surface temperature can heat and cool with in a minimal amount of time. The material also must exhibit a fair amount of lateral conduction to dampen uniformity issues. The material that was chosen for construction was copper. Copper possesses both of these qualities.

       

      Heating / Cooling the radiation source
      The next issue was how to adequately heat and cool the copper plate. Nine thermoelectric heat pumps5 were placed behind the copper plate in a 3 by 3 matrix. On the other side of the heat pumps was placed an aluminum heat sink. The package resemble a sandwich configuration as seen in Figure 3.2.

         

        Figure .2 Placement and configuration of heat pumps in blackbody.

         

    The choice of heat pump was the CP2-31-10L manufactured by the Melcor Corporation5. This particular model was chosen because of its physical size and response (See Appendix A). Basically, the thermoelectric heat pump is a solid state pump without any moving parts, fluids or gases. The basic laws of thermodynamics apply to this device just as it does conventional heat pumps and other devices involving the transfer of heat energy.

    The heat sink, model 5045 manufactured by EG&G Wakefield Engineering, was used as a means of dissipating heat. The specifications and dimensions of the sink can be found in Appendix B. Forced convection was implemented as opposed to natural convection in order to remove heat at a faster rate. This forced air movement was provided by means of an electric fan (See Figure 3.3).

        Figure .3 Forced convection of aluminum heat sink.

         

        Assembly
        The three layers of the blackbody were held together at the four corners of the plate with 4 #8 hex bolts. This proved to be sufficient in terms of contact and pressure. Between the copper plate and the heat sink, but around the heat pumps, were placed two layers of open cell polyurethane foam sheeting. In this way there would be a minimum amount of heat transfer in the non-contact areas between the plate and the sink, which is unwanted. Two layers were used as opposed to one layer to attain greater insulation. The surfaces of the heat pumps were coated with Omegatherm 201, which is a high thermally conductive paste, to ensure maximum heat transfer.

         

        Electrical Characteristics
        Once the assembly was completed the heat pumps were wired together. Available to the principle investigator was a 10V, 10A programmable power supply (See Appendix F). It was desired that the heat pumps be wired in such a way as not to exceed these supply ratings. It was also desired not to exceed 30W total power. This is a constraint of the MISI system. The nominal rating for one heat pump is 1V @ 3A. With this in mind, the heat pumps were wired in a series / parallel configuration (See Figure 3.4).

         

        Figure .4 Electrical wiring of heat pumps in the blackbody.

         

        If the resistance of each heat pump is known then the over all system power requirement can be found. The individual resistance of each heat pump was 0.5W . In a series / parallel configuration, the equivalent resistance is also 0.5W for the 3 x 3 matrix. From this the voltage and current can be calculated, for the given wattage constraint using equations 3.1 and 3.2.

        (3.1)

        (3.2)

        The overall power required to drive the blackbody would be 3.9V @ 7.8A. These values do not exceed the ratings of the power supply nor do they jeopardize the 30W constraint.

         

         

        Thermal Reflections
        It is desired that the copper plate act like a perfect blackbody. In reality though, this is not possible since there will always be some reflection. It is possible, however, to assess an objects radiant emittance as compared to a perfect blackbody. This comparison is made numerically with a quantity called emissivity. Here the emissivity is the ratio of the actual radiant emittance of the copper plate to that of a perfect blackbody. It is desired that the emissivity be 1.0 or as close to 1.0 as possible. To achieve this, a special paint is sprayed onto the surface of the copper plate. When cured, this paint exhibits a relatively high emissivity. The emissivity for the copper plate was around 0.95 ± 0.02. This value can be determined with a device called an emissometer.

       

Feedback Control System (TOC)
In order to control the temperature of the blackbody a system was designed that could sense the temperature of the copper plate, report the temperature to a computer, and have the computer make a decision as to what the power supply output should be in order to maintain a given temperature set point.

A typical temperature control system is a group of components, or subsystems, combined to regulate the temperature of a mass. The system may be described by a functional block diagram, such as that shown in Figure 3.5, where the blocks represent black boxes that receive certain indicated inputs and emit designated outputs.

      Figure .5 Block diagram of the control system.

       

The control system implemented was made up of 5 black boxes and a temperature sensor, which acted as the feeding back element. These blocks are described in detail in the following sections.

 

    Transducer and Signal Processing
    A transducer, for all practical purposes, is a devise that converts one kind of energy (or signal) to another.5 Here a thermocouple converted a radiant heat signal into an electric voltage for input into an amplifier. The thermocouple used for sensing temperature was the AD590L manufactured by Analog Devices (See Appendix C). Basically the AD590 is a two-terminal integrated circuit temperature transducer which produces an output current proportional to absolute temperature. When wired as a voltage divider in series with a 1kW resistor, the transducer is capable of producing a change of 1mV / °C. This is far too small a change to be adequately sampled. Therefore it was necessary to process the signal by means of amplification. A 741 operational amplifier was employed to enhance the signal by a scaling factor(See Appendix K). The schematic of the entire transducer and signal processing circuit can be seen in Figure 3.6.

         

        Figure .6 Schematic of transducer and signal processing circuit.

         

    An explanation of the circuit diagram in Figure 3.6 follows. The resistor in series with the AD590L sensor was 1kW (R3). As mentioned before this produces a change of 1mV / °C across the sensor with the remaining 5V dropped across R3. This remaining voltage goes to the non-inverting input of the op-amp. The room temperature voltage at R3 was about 0.322 V. This is effectively the offset voltage for the given room temperature.

    The closed-loop gain of the op-amp was 40. Gains above 40 were difficult to employ because of high signal-to-noise. As a result, it became increasingly difficult to sample signals at the larger gains. The addition of 0.1m F capacitors (C1 ,C2 and C3) at the output and across the supply helped to filter out some of the noise.

    The values of R1 and R2 can be found by using equation (3.3) which is the gain relationship for non-inverting op-amps.

         

        (3.3)

    It is desired to have the input impedance as large as possible. In this way the op-amp will not load down other components in the system. At the same time it is desired to have the output impedance be as small as possible in order to have maximum voltage output. From equation (3.4) R2 was 107.9kW and R1 was calculated to be about 2698W . This produced a gain of around 40.

    Normally the inverting input of the op-amp is tied to ground. In this case, however, it was necessary to induce an offset at the input. The reason being is that with Vin = 0.322v for a given ambient, the output of the op-amp will be (0.322 x 40) 12.88 V. This output exceeds the sampling range of the data acquisition system which is 0-10V. By adding in a positive or negative bias factor, the output voltage can be controlled to accommodate sampling. The values of R5 and R7 were 1kW while R6 was a 2kW trimmer. R4, which was 475W , was used as a current limiter to the inverting input.

     

        A/D Conversion
        After the analog signal was amplified it needed to be sampled so as to be useful in a software control algorithm. The sampling was performed with an RTI-800 Personal Computer interface board manufactured by Analog Devices (See Appendix D). The RTI 800 was used to provide a direct interface between an IBM PC and the analog/digital world. In general, the boards are used for a variety of data acquisition, analog output, and digital I/O functions. For the analog-to-digital conversion, 12-bit sampling resolution was employed. This provided digital counts in the range of 0 to 4095. The 12-bit converter on the RTI was unipolar configured so as to receive analog input voltages within the range of 0 to 10V. The sampling rates were on the order of 25m s which were fast enough for this application. The conversions were initiated with software commands as described in section 3.2.3.

         

        Computer
        At the heart of the data acquisition system was an IBM compatible 286 PC. This system was running under MS-DOS 5.0 and was configured with 2Megs of RAM. The CPU clock speed was 16Mhz. Included in the systems files was Microsoft’s GW-BASIC interpreter. BASIC is one of many programming languages understood by the RTI board. For this research all the algorithms to perform control tasks were coded in BASIC. The RTI-800 series software package provided a convenient and powerful software interface between the RTI-800 series I/O board and the IBM PC. Included with the software package were I/O and system routines, written in BASIC, that enabled communication with the RTI boards (See Appendix J). Theses routines were used in addition to the standard BASIC commands.

         

        D/A Conversion
        Because the RTI-800 was not capable of digital-to-analog conversion, an external D/A converter was implemented. This process was conveniently perform in a single 16 pin DIP. The device used for the conversion was the AD558KN manufactured by Analog Devices (See Appendix E). The AD558KN is a complete voltage-output 8-bit digital-to-analog converter. No external components or trims were required to interface the 8-bit data bus to the analog output of the computer (See Figure 3.7). In addition, the chip was setup so as to have an output voltage range of 0 to 10V.

        Figure .7 Schematic of digital-to-analog conversion.

         

         

        Since the data bus was 8-bit, the 10Vmax output was quantized into 255 levels, including zero. This gave a resolution of (10V / 255) 0.039V/digital count (DC). This also means that one volt is equivalent to 25.5 DC. From this a look up table was employed, as shown in Table 3.1, to convert from voltage to digital counts (DC) and visa versa.

         

        Voltage (V)

        Digital Count (DC)

        0

        255

        1

        229.5

        2

        204

        3

        178.5

        4

        153

        5

        127.5

        6

        102

        7

        76.5

        8

        51

        9

        25.5

        10

        0

        Table .1 Look up table for voltage-digital count relationship.

         

        This relationship can be summed up in the following equation:

        (3.4)

         

        Programmable Power Supply
        After the D-to-A conversion the signal was then feed into a programmable power supply to electrically control the heat pumps. The choice of power supply was the MPB Series 10V/10A supply manufactured by Hewlett Packard (See Appendix F). At the rear of the supply was a series of terminals that enabled remote voltage programming. This means that the voltage stays fixed for a given condition while the current varies depending on the load. To employ voltage programming the following strapping pattern was used (See Figure 3.8):

      Figure .8 Remote programming for constant voltage.

      In this mode, the output voltage will vary in a 1 to 1 ratio with the programming voltage (reference voltage) and the load on the programming voltage source will not exceed 25 microamperes. The impedance matching resistor (Rx) for the programming voltage source was 1kW to maintain the temperature and stability specification of the power supply.

       

       

Visual Means of Assessment (TOC)
To examine the temperature and uniformity of the blackbody an IR camera was employed. The Inframetrics 600L Infrared camera was used for all visual experiments (See Appendix G). The camera was setup such that it was pointing directly at the blackbody approximately 4 feet away (See Figure 3.9). With the inframetrics camera, the blackbody surface could be seen in real time. The output of the camera was connected to a VCR for image capture. This video of the blackbody "in action" was then digitized using a frame grabber. The software/hardware used to capture the video images was IMAGE LAB manufactured by Werner Frei Associates. These captured images were save in a TIFF format and then imported into Photoshop 3.0 for evaluation.

 

 

Sensor / Themistor Placement (TOC)
As previously mentioned in section 3.2.1, the sensing element used for the majority of the experiments was the AD590 integrated circuit temperature transducer. This sensor was imbedded into the front of the copper plate via a milled hole. This drilled hole was slightly smaller than the diameter of the sensor. In this way the thermal lag between the sensor and the copper plate could be minimized. To help in the coupling Omeathem 201, which is a high thermally conductive paste, was used around the sensor.

The placement of the sensor was determined after examining preliminary surface profiles of the blackbody. From this the sensor was placed as shown in Figure 3.9.

Also used for sensing temperature was the YSI Series 400 thermistor (See Appendix I). These were used on the front surface as well as on the cooling fins of the blackbody as shown in Figure 3.9. Since the thermistors exhibit a change in resistance with a change in temperature, they were plugged into a standard ohm meter.

 

 

      Figure .9 Diagram showing the placement of thermistor, AD590, and usage of IR camera.

       

       

      Use of Laboratory Standard Blackbody (TOC)
      A laboratory standard blackbody was used as a reference for calibration and as a comparison standard for uniformity issues. The model used was the SR 80-7A (7" square model) manufactured by CI Systems (See Appendix I). This model consists of a radiation surface plate that is microprocessor controlled. In general, it has a temperature accuracy around 0.01 °C with an emissivity of 0.97 ±0.02. The temperature sensors in the unit are precision platinum resistance thermometers (PRT’s) in which the difference between them is controlled.

       

      Previous Section: Introduction / Background
      Next Section: Results