Peltier Thermopile Cloud Sensor

Transcription

1 Peltier Thermopile Cloud Sensor By Eric Chesak The design of this device is divided into two basic components, the circuit and the detector head. The circuit and final construction is very flexible and can be configured in many different ways, to best suit the builder. The sensor head can also be designed to use small Peltier devices, for small thermal mass and faster response, or large ones for capturing large amounts of energy. The circuit is a fairly traditional bipolar non-inverting operational amplifier (OP Amp). The Author used an LF411CN, which is a low offset, low drift amp, with JFET inputs. It s an older IC, but was readily available. This is pin-compatible with the very common LM741, which could be easily substituted. There is a dizzying array of other op amps that would work equally as well, so it s left up to the final user of this device to decide what is best for their application and what happens to be in their inventory. One unusual aspect of this circuit is the floating output common, which allows the offset potentiometer to be able to offset large signals, either to positive or negative values. The amp takes the inputs of the Peltier module provides gain or unity following, depending on the values of the resistors. The potentiometer connected to the op amp is used for circuit gain. In a separate part of the circuit, the output is off-set by a resistor/potentiometer network. This consists of 2 resistors and a 10-turn potentiometer. Depending on the construction of the detector, gain of the circuit, or background radiation, an offset may or may not be needed to null the signal. This circuit can be as elaborate or simple as the builder wishes. It can be constructed on a prototype board, a perf board or a simple PCB could be etched. It can also be powered by a bipolar power supply, or two 9V batteries (see schematic). The output can be displayed on a DMM, or on a dedicated panel meter. So the final construction is very flexible. The Sensor head is essentially a device that allows one side of the Peltier module to be held at ambient temperature, while the opposite side is allowed to rise and fall as the incident energy allows. It also is designed to help shield the sensor from any stray radiation that may arrive at the sensor face, since this will surely skew the readings, especially in high gain mode, where the heat from one s hand can generate output. This can also be constructed in many different ways. The Author constructed his with a simple PC heat sink and machined aluminum housing. The Details: The circuit was first assembled onto a breadboard to check functionality. It also allows for easy modification of parts, should something need to be adjusted (like a resistor value, or similar).

2 A lab supply was used to power this with +/- 9V, and a DMM was used to measure the output voltage from the circuit. The Peltier is shown edge-on in the lower-center of the frame.

3 Next, the circuit was constructed on a solder perf board. This construction lends itself to quick assembly and moderate ease of modification. Whichever technique is used, the circuit is fairly simple to assemble. Just be sure to use good soldering techniques and work to avoid any flux residue, which can create erroneous readings on sensitive, high gain circuits.

4 In addition to using a dual supply op amp, it is also possible to use a single supply op amp like the LM358, or similar. With type of op amp, there is only a need for a single battery or power supply. Nulling any offset from the Peltier can be more difficult, depending on the environment. But if the sensor head is working properly and both faces of the Peltier are at the ambient temperature, there should be little need for much offset, regardless if a single or dual supply op amp is used. However, if the intended use is as a sky detector, a bipolar power supply will be needed to measure negative temperature readings from the sensor. Also keep in mind that op amps, other than the LF411 can provide a wider operating voltage range. Again, the LF411 was chosen, as this is what the author had readily available. The sensor head is constructed with an appropriately proportioned Peltier cooler. This Peltier was intended for small amounts of thermal radiation, so a smaller Peltier was used. In this case it was a 15mm x 15mm unit with 31 thermocouple junctions. If a larger amount of thermal energy is planned to be measured, then a larger Peltier would be chosen. Also pay attention to the number of junctions that are on the Peltier. The limit in capacity is related to how much energy the Peltier can absorb without heating to a point where the solder joints would become soft. This would require a fairly large amount of energy, however. In general, smaller Peltier coolers will have higher sensitivity and still withstand a fair amount of energy. Peltier s are also available with differing number of junctions. When choosing a Peltier, look for units with the highest number of junctions per unit area. This will give the sensor the most sensitivity. The higher junction density will help the unit to generate more voltage for a given amount of input energy, thus providing more sensitivity. The Peltier is mounted to a heat sink to help in energy removal. The choice of a heat sink is left to the builder, but BGA or CPU heat sinks would be excellent candidates. The key to getting good sensor readings is to keep the cold-side (heat sink side) from changing temperature, when energy in input onto the hot side. This is practically impossible, but with small energy inputs and a decent heat sink, the temperature rise of the heat sink will be much slower than that of the sensor s surface. It s also helpful to minimize the amount of time that the sensor sees incident radiation. This will allow the sensor time to equilibrate. The sensor will need to absorb enough energy to get a stable reading, but once the reading has stabilized, the input energy should be removed, to prevent heat from building in the heat sink. The sensor should also be surrounded with a shield that will be held at ambient, as much as possible. Although this is not absolutely necessary (depending on the level of signal being measured), this will help reduce any energy absorption that hits the Peltier from directions other than the intended source. Since the sensor is sensitive to a very broad range of input energies, even heat from standing near the sensor can create error in the readings, depending again on the level of the readings that are trying to be measured. This shield can be of any form, but should be tied to the heat sink, to assist in heat dissipation. The sensor presented here uses a thick aluminum shield that also adds thermal mass to the heat sink-side, while also shielding the sensor. This could easily be bent sheet metal that is connected to the heat sink. The Author used an aluminum block to add to the thermal mass, to provide a system with a larger time-constant. If the builder wishes to use an aluminum block, but doesn t have access to a machine shop, an appropriate drill hole is just as effective. A link to a supplier of metal stock is provided,

5 in case the builder doesn t have a local metal supplier. Many local metal suppliers would consider pieces of these dimensions to be scrap. It is completely acceptable to ask to see and buy pieces from their scrap bin. In general, heat sinks & shields with higher mass will provide a more stable instantaneous reading, and for a longer time period, at the cost of needing to equilibrate longer. Heat sinks & shields with lower mass may not be as stable, in the short timeframe. However, they will also equilibrate faster, allowing for faster turn around. The final use would dictate what would be the ideal combination of mass vs. stability vs. operational up time. The sensor head designed here would be considered to be something of a middle-of-theroad design. The mass is not huge, so equilibration happens within a reasonable timeframe. However it is not so light that it compromises reading stability and absorption time. If the sensor is to be located some distance from the circuit, it best to use some form of shielded cable or coax, to reduce the amount of noise or stray voltages that could be induced in the cable. This will also help create a more stable output from the sensor. If a cable is used, an additional 100nF capacitor can be placed between the input lines to aid in reducing noise. In order to aid in energy absorption, the sensor side of the Peltier needs to be coated with some type of broadband absorber. Several light coats (held at a distance) of Krylon Ultra-Flat black spray paint will make a good absorber. However, the best, but most fragile, is to use a coating of candle soot. This is what is presented here. Prior to coating the sensor face, it s a good idea to mount the Peltier to its final location on the heat sink and perform any calibration operations (see calibration section, later in this document). Mounting of the Peltier is accomplished by the use of a thermally conductive epoxy or thermally conductive tape. If the sensor is coated prior to mounting it to the heat sink, it will be impossible to mount it without destroying the coating. A cheap candle of Paraffin will work fine for producing carbon soot. A small section of ½ x ~1 long diameter copper pipe or similar metal tube is also needed, and a way to hold it in the candle flame. A ½ copper female/female coupler is ideal. A couple turns of stripped wire should be wrapped around this part to allow it to be held, while in the flame. It will obviously get hot. The Author used a work-holding third-hand, to hold it in the proper location of the candle flame. When this stovepipe is held over and around the candle flame, close to the pool of melted candle wax, the flame is enriched and copious amounts of carbon are emitted through the top of the stovepipe. It s best to burn the candle for minutes to have a nice pool of paraffin that is melted. The height of the stovepipe also has to be tuned, for the best carbon production. It can be raised and lowered from the pool to adjust the amount of carbon produced. This is analogous to running an internal combustion engine in a fuel rich setting. There is not enough oxygen to complete the combustion process, and the soot is produced. Note, if the stovepipe is set too low the flame will extinguish.

6 Image 1 Note the column of soot particles emitted from the stovepipe Once you have a nice stream of soot from the candle, pass the sensor back and forth over the soot stream, until the Peltier s surface is nicely coated. It s best to use short passes, rather than holding it in the flame. The heat from the candle could heat the sensor s

7 surface enough to melt the junction solder and ruin the Peltier. Also, you ll get the best adhesion of the carbon soot, if the surface is cleaned with alcohol just prior to coating. It s also a good idea to tape-off the areas where you don t want soot to be coated. This can be a layer of painter s masking tape or similar. This paper tape also helps to gauge whether the flame it too close. If the paper burns or chars, it s too hot. Image 2 The freshly coated sensor surface. Note that the paper appears to be charred. However, the black that is evident on the painter s tape is a layer of soot, not char. In the photo below, the camera flash was intentionally used to create glare off the surface of the heat sink. With this glare, it is easy to see the level of absorption of the layer of soot. The heat sink is black anodized, but still produces some reflection of the incident light. However, the sensor s surface is free of any level of reflections or back scatter, indicating that it has a high level of absorption. The publication from NIST, listed below, indicates that this technique produces absorption on the order of 98.75% (reflection of 1.25%).

8 Image 3 This image showcases the very low reflection coefficient of the sensor surface The sensor head was assembled to the side shield, with an appropriate amount of white thermal grease.. A BNC connector was also mounted, for shielded connection to the sensor. The entire head was mounted on a DIY column and base. These are readily available at several sources dealing with commercial optomechanics (Newport, Thorlabs, Melles Griot/CVI, Edmund Optics, etc). Here is a front and rear view of the sensor. Following are some preliminary views of the sensor. In this configuration, a detached enclosure would have been used between the sensor and the DVM. Or the enclosure could incorporate a panel meter, for direct readings.

9 Image 4 Preliminary front view of the sensor

10 Image 5 Preliminary rear view of the sensor Calibration of the sensor can be done in several ways, depending on its final intended use. If the sensor is to be used for something to measure visible light (ie laser or light power meter), it is best to use a calibrated sensor for the wavelength similar to the wavelength of the intended use. The intended light source (ie laser or lamp) can be directed into the laser power meter, and the reading recorded. Then the same light source can be directed into this sensor. The gain potentiometer is then adjusted to set the power level to that which was read from the calibrated sensor. For calibrating for mid to far infrared wavelengths, a 1 ohm surface mount resistor can be attached to the face of the Peltier (prior to coating with lamp black). Thermal paste would need to be used to insure good thermal contact. By applying power to the resistor, the power can then be calculated and the sensor electronics adjusted accordingly. For

11 example, with current of 10mA going through the resistor, the resistor would be dissipating 100mW. This is calculated with Ohm s Law (Power = Current 2 x Resistance). The gain potentiometer can be adjusted accordingly. This method may also work to calibrate the sensor for visible wavelengths. However, if the lamp black absorption is not the same at visible and far infrared wavelengths, there would be some level of error. According the NIST paper below, lamp black can vary as much as 10% in absorption, when measured at visible vs. (thermal) IR wavelengths. This sensor turns out to be incredibly sensitive at detecting heat. Simply holding one s hand in front of the sensor is enough to change the sensor s readings. Knowing this, the builder should be aware that if small signal levels are being measured, heat from ones hand or body can skew the readings, providing erroneous results. At the last minute, it was decided to change the design slightly, with the addition of an enclosure that is incorporated as the base plate of the sensor stand. This was done for no other reason but to make the unit more compact, while still having some flexibility for external power. Two ½ plates were machined with matching recessed pockets, to house the electronics, as well as several of the connectors. Although the trim-pots are enclosed in the base enclosure, it would be generally more flexible to have these readily accessible (at least the zero pot). Image 6 The enclosure to house the electronics, which doubles as a base for the sensor Following are a couple images of the completed sensor, ready for testing. As mentioned earlier, this version was intended to be powered by a bipolar lab power supply, rather than batteries. However, the current draw on the circuit is very small, so battery operation is also easily done.

12 Image 7 Final version of the sensor front view

13 Image 8 Final version of the sensor rear view

14 Image 9 - An image of the whole system, except for the power supply. The Use The intended purpose of this device was to measure the sky temperature. This may seem like a strange thing to measure, but generally when the sky temperature is low, the sky is clear. When the temperature is closer to the ambient temperature, it is cloudy. The fixture was mounted sideways so that the open portion of the sensor was facing the sky. It was also covered with a ½ thick aluminum plate, the same size as the sensor frame. This was used to cover the sensor and take valid zero readings. The sensor and the cover plate were all left outside to equilibrate to the outside air temperature before taking any readings. Small amounts of high level moisture will affect the sky temperature enough to distinguish a reading difference between clear and dry vs. clear and moist. When measuring cloudy skies, the sky temperature (or cloud temperature) will be closer to the ambient temperature, as the clouds emission of IR radiation. This is important for astronomy and especially astrophotography. The more transparent the sky, the less the scattering will affect the propagation of light. The best astrophotography data is collected under truly clear skies.

15 The following is data collected from the sensor under various sky conditions. Note how the recorded sensor voltage drops with clear skies. The lower voltage reading corresponds to lower sky temperatures, and generally clearer skies. Also note the data from the partly cloudy skies, and the fact that it is positive, indicating a much higher temperature than the readings with the clear skies. Different Peltier sensors, different gains and sensors of different design will read different voltages, but should have similar proportional readings. Thin Clouds Clear Clear Thin Clouds Partly Cloudy Very Clear Sky Condition Zero Reading Sky reading Final Reading Table 1 A set of sky readings from the Peltier Cloud Sensor. These readings were all recorded under similar temperatures (15-20 C). Drastically different ambient temperature readings between sensor readings will likely need some type of offset. There has bee some difficulty reproducing these exact readings on other occasions. However, readings in similar conditions were proportional to the readings above. Clear skies always are shown as negative readings and cloudy skies as a positive reading. The Other Uses Although the primary purpose of this project was to detect clouds, it turned out to be a fairly interesting project for measuring other sources of radiation. It was tested with other optical sources such as lasers and flashlights, with equal success and sensitivity. The sensitivity of the sensor is quite amazing. When in close proximity, the heat of one s hand is easily detected. The author found this sensor design to be particularly sensitive, easy to fabricate and a worthwhile endeavor. It is well worthwhile investigating these other uses as well. The Section for Further Reading: