The Thermopile Cloud Detector

Sloan, Shaw and Williams (1955) have provided data which show that infra red measurements of the sky at wavelengths above m (in particular between m and m) are sensitive to the presence of clouds. The presence of clouds produces an enhanced signal (which corresponds to an approximate black body spectrum at about ground temperature) above that from the clear sky. There is also an effect of atmospheric humidity which can produce enhanced signals particularly at low elevations. Inexpensive infra red sensors are now available which make all-sky or limited direction cloud monitoring possible without detector cooling systems and thus have low current drain with simplified design constraints.

 
Figure 1: Circuit schematic of the cloud detector. A thermistor measures the internal temperature of the detector canister and this is used to provide a reference for the thermopile element. The compensated output thus corresponds to the temperature in the detector field of view.

The cloud detector described by Ashley and Jurcevic satisfied many of the requirements for a monitoring detector but from our point of view had two disadvantages which were (a) its requirement for mechanical chopping and (b) its use of a mirror. These aspects would limit its long term reliability in a remote desert environment. The requirement for chopping is inherent in the pyroelectric infra red element. Such an element responds to an infra red signal with adequate sensitivity but returns to its DC level with a time scale of the order of a second. Thus chopping between the sky and a comparison with a frequency of the order of 1 Hz is required. The concept for the use of a cloud detector with the Auger array is that detectors may be attatched to many of the 3000 cosmic ray detectors at each site. They would operate remotely and have no daytime weather protection. For this purpose, a mirror is not likely to be suitable due to deterioration from dust. A mirror or an infra red fresnel lens might well be suitable for other astronomical applications.

Thermoelectric sensors available now have similar sensitivity to the pyroelectric detectors but will maintain an output voltage proportional to the temperature difference between the field of view and a local comparison within the detector canister. We have now based our cloud detectors on such sensors. We monitor the detector canister temperature with an internal thermistor which allows us to derive an analog signal corresponding to the temperature over the field of view or to later calculate that value from ADC measurements of the output signals. With the data presented below, the correction for the canister temperature is substantial. In situations where the detector is deployed only at night or, as will be the case for the Auger array, it will be shielded from the sun and attached to a massive body with a long thermal time constant, the correction will be smaller and more straightforward. We have operated detectors of this sort for up to six months and have found no problems due to long term drifts. The sensor elements are sufficiently sensitive that we do not require a mirror for monitoring the presence of visible clouds. It is possible that sub-visual cirrus may be detectable with the aid of a mirror as suggested by Ashley and Jurcevic but this was not our present purpose.

We define a field of view by mechanical collimation. In most cases, this was achieved by mounting the sensor element behind a small aluminium tube made to have a diameter of 8.2mm and a length of 14.5mm. With the very small detector sensitive area, this results in a field of view with a full angle of about . We have made detectors with fields of view down to one tenth of this value with the use of infra red fresnel lenses such as are used in security intruder sensors.

We chose to use an EG&G Heimann type TPS 534 thermopile detector which has a sensitive area of 1.2x1.2mm, a responsivity of 42V/W and a filter which passes wavelengths above m. The sensor itself is modest in cost and a complete cloud detector pixel can be built for below AUS$200 in component costs (at one off prices). This includes using low-power precision components with low drifts and offsets, required for an analog correction to allow for the canister temperature.

The circuitry (figure 1) was designed to have low current drain and can operate continuously for well over a month using a 9V alkaline battery (supply current below A).

© Copyright Astronomical Society of Australia 1997