Development of the first dustsonde began at the University of Minnesota in 1961. The first sounding was conducted in August of 1963, and the first publication appeared in 1964 (Rosen, 1964). In the late 1960's the dustsonde program moved to the University of Wyoming and has continued there until the time of this writing (2014).
An important original design goal of the dustsonde was to focus on achieving as close a forward angle scattering as possible without extraneous internal scattering from the walls or from other parts of the optics, with the background being determined by molecular scattering from the internal volume. This led to an instrument with an average scattering angle of 25 deg. (ranging from ~ 12 to ~ 35 deg.) and a background light level that was 98% molecular scattering as determined from low pressure tests. A White light source was used from a straight filament incandescent bulb operated at an over voltage giving it about a 4 hour lifetime.
At lower pressures the flow pattern through the sensitive scattering region can take on a very different pattern, passing only partially through the illuminated region. In the development process, visual observation of the flow pattern (using smoky air) was accomplished using a glass bell jar and windows in the scattering chamber. Adjustments to the intake and sheath flow were made so that virtually all the sampled air passed through the defined scattering region. Furthermore, it was found that additional adjustments needed to be made to the sheath air geometry and flow rate to prevent escape of incoming sampled particles which would cause spurious counts. In a continuing effort, during flight aerosol free air was periodically introduce into the sample stream to detect and eliminate such possibilities.
The counting efficiency of the dustsonde vs altitude was tested in an environmental chamber. The usual method involved first operating at ambient conditions with a surrounding non-volatile aerosol. When the pressure was reduced relatively quickly by a factor of two, a resultant factor of two decrease in the counting rate should also be observed for 100% efficiency. This incremental change in pressure was done at successively lower pressures.
In addition to the scattering chamber geometry and tests, several other aspects are relevant for documentation here:
The pumps drawing air through the scattering chamber needed to be smooth flowing (non pulsating), be able to produce the required head pressure, have a flow rate independent of altitude, and require low power consumption. A type of gear pump with high tolerances was developed for this application.
Photomultiplier tubes were employed in the light pulse detecting system which required high voltage power supplies. The presence of high voltage (greater than ~300 volts) on balloon flights can lead to numerous types of spurious signals, sometimes imitating the data itself. The effort put into encapsulating and low pressure testing high voltage light detector units was a major part of dustsonde construction.
Initially the scattering chambers had significant volume "dead space" which contained expanding air as the balloon rises and must be taken into account in the sample flow rate. This effect was greatly reduce by using a smaller scattering chamber with the same optical characteristics.
In part, the traditional dustsonde is characterized by the employment of just 2 channels for particle size discrimination. The smallest size was determined by setting the detector threshold as low as possible without an excessive background counting rate. The largest size was set so that a significant number of particles would be counted as constrained by the pump and flow rate. Also, as discussed below, the threshold for larger particles could not be significantly larger because of the double response region of the optical scattering relevant to the dustsonde. In addition, for larger sizes more consideration would need to be given to sedimentation and impaction losses in the sampling geometry of the intake system.
The initial dustsonde flights indicated the presence of a mysterious background counting rate that needed to be subtracted from the aerosol counting rate. This background was determined to be from Cherenkov radiation (cosmic rays passing through the glass in the photomultiplier tubes and lenses) which had a maximum in the stratosphere from secondary production. The problem was eliminated with the use of a coincidence system requiring two independent photomultiplier tubes (including power supplies) to produce simultaneous pulses. This resulted in a huge improvement in reduced background counting rates from all sources but considerably increased the complication and expense of dustsonde construction. Coincidence counting started with the flight on April, 28, 1965 ( the first 6 flights did not use coincidence) and allowed the pulse height discriminators to be set at a lower equivalent size level, as is documented in the NDACC files. In addition the first two dustsonde flights from Minneapolis reported in the NDACC data base only employed one particle size.