PCASP

FAAM CORE Passive Cavity Aerosol Spectrometer Probe

The PCASP measures aerosol size distribution in the nominal range 0.1 to 3 micrometers. It is an optical particle counter (OPC) and the size of each particle is derived by exposing it to laser radiation and measuring the amount of light scattered. The instrument is fitted to the aircraft in one of the PMS canister under the aircraft wings.

History

FAAM owns two PCASP-100X instruments purchased from Particle Measurement Systems (PMS), these are generally referred to with the unimaginative titles PCASP2 and PCASP1. In December 2008 PCASP2 was sent to Droplet Measurement Technologies (DMT) to have its internal electronics brought up to date. This involved removing all data processing electronics in the instrument and replacing them with DMT's SPP200 electronics package. In addition to providing up to date, currently supported and hopefully more reliable electronics this provided the instrument with 30 software programable size channels - a significant improvement on the previous 16 hardware configured size channels. In April 2009 both probes were flown on the MEVEX detachment, during which PCASP1 developed a serious fault. In Early 2010 PCASP1 was repaired and also provided with DMT's SPP200 electronics package. This means that all data from May 2009 onwards is derived from instruments with the SPP200 electronics.

Description and Theory of Operation

The PCASP is an OPC with internal sampling. The two most important points to note are its sampling method and its measurement method. Of some additional note is the internal processing which allocates particles to bins forming a size distribution. These points will be discussed in turn.

Sampling

The most notable feature on the outside of the PCASP is the nose cone. This is a long cone with circular inlet with a diameter of ~1 cm. As air enters the inlet and passes down the length of the cone conservation of mass causes the flow speed to slow by approximately a factor of 10. Near the base of the nosecone a subsampler samples a few cm³ s^-1and the remaining air pases out through the exhaust/bypass at the side. The subsampler is a sharp edged circular cross section tube with inner diameter ~0.4 mm.

Sampling is very important when making aerosol measurements. Because aerosol particles have mass they do not exactly follow streamlines when sampling from ambient air. This can lead to an increase or decrease in the aerosol concentration of your sample. The effect is dependant upon a number of factors including velocity of the sample within the inlet, velocity of the air past the inlet, aerosol diameter, aerosol density and viscosity of the air. It is standard practice to correct sampled size distributions using the relationships derived by Belyaev and Levin (1974) and this process is performed on data provided by the PCASP. It is assumed that at the inlet to the nosecone sampling occurs isoaxially and isokinetically so no correction is required. At the subsampler it is asumed that the air has travelled down the nosecone via incompressible laminar flow which defines the air speed within the nosecone. The air speed within the sampler is derived by measurement of the sample flow rate using a mass flow meter. This mass flow rate is converted to volumetric flow rate using ambient temperature and pressure.

The assumptions made here have significant uncertainties. It is not clear how close to isokinetic the nosecone sampling is and because of the non-zero pitch of the aircraft the sampling is not isoaxial. Flow down the nosecone is expected to be turbulent and the speeds involved may indicate the flow is compressible. This means that the flow at the subsampler is not isoaxial and the direction and speed of the air sample vary with time. The current correction ids based on our current best estimate of the inlet efficiencies of the PCASP but work is ongoing to further characterise this parameter.

Measurement Method

Once an aerosol particle has entered the subsampler it is passed along a straight tube where it is surrounded by a clean sheath of air which focuses the sample. The sample passes through a laser beam where scattered light is collected by a parabolic mirror and focused onto a photodiode. Light is collected over the scattering angles 35-120 degrees plus 60-145 degrees. The light collecting optics are close to rotationally symetric about 0 degrees scattering angle.

Each individual particle causes a peak in the signal from the photodiode and the height and width of each of these peaks are measured. If a peak has a width below a given threshold it is discounted to remove electrical noise, then each remaining particle detection is assigned to one of thirty bins based on the peak height. It should be noted, however, that for data logged on the SPP200 electronics bin 1 is different to all other bins. The lower edge of bin 1 is defined by the peak width threshold not a peak height threshold. This means that bin 1 must always be discounted from any analysis.

Internal processing

As has already been described, the SPP200 electronics assigns each particle to a bin, with bin 1 being unique in that it's lower threshold is defined by the measured peak width not the peak height. In order to assign the particle to a bin the signal from the photodiode is amplified and digitised. For the SPP200 electronics the digitisation is performed by an A-D converter and a processor assigns particles to bins. In the original PCASP-100X electronics the digitisation and bin assignment were performed together with chained voltage comparators. In both cases it is not possible to provide the signal range required with a single linear amplifier. The PCASP instead used 3 amplifiers in parallel which provide high, mid and low gains. If the high gain amplifer saturates then the mid gain amplifier is used. If the mid gain amplifier saturates then the low gain amplifier is used. If the low gain amplifier saturates then the particle is discounted (it is recorded as an oversized particle in the SPP-200 electronics).

Because of slight differences from instrument to instrument each amplifier has an adjustable baseline. The high gain baseline is adjusted so that electrical noise does not cause particles to be falsely detected. The mid and low gain baselines are adjusted to match the saturation point of one amplifier with the zero point of its neighbour. In reality, for the SPP200 electronics, the baselines are adjusted to give some overlap. If the amplifiers overlap slightly then particles which fall in the overlap zone are still counted exactly once by the amplifier with the higher gain, although the first bin in the lower gain amplifier will be narrower than nominal. If, however, the amplifiers do not quite meet then any particles which saturate one amplifier but generate no signal on the next amplifier are discarded. It is certainly better to have one slightly narrower bin which can have its width calibrated (see below) rather than lose particles from the spectrum.

Calibration

The PCASP nominally measures particles from 0.1 to 3 micrometres, however each time the instrument has its optics cleaned and realigned the instrument must be calibrated (this occurs sometimes as often as once a month). The original PCASP-100X instruments were checked with polystyrene-latex (PSL) beads. These are NIST traceable size standards. The check involved nebulising PSL beads and checking they were measured in the correct bins, however, no formal calibrations were performed. Since the SPP200 electronics have been used FAAM have been working hard to develop effective calibration routines for the PCASP. It should be noted that the PCASP does not directly measure the diameter of a particle, it actually measures the particle's scattering cross section integrated over the optics' collecting angle as defined above. For homogeneous spherical particles with a known refractive index the cross section as a function of diameter can be defined by Mie theory. Mie theory indicates that scattering cross section is a highly nonlinear function of diameter and for particles with diameters greater than ~ 1 micron which absorb little light, scattering cross section is not a monotonically increasing function of diameter. This is important to bear in mind when calibrating any OPC.

The small diameter bins are calibrated using a DMA which is used to provide monodisperse particle distributions from 0.09 to 1 micrometer. This allows the position of each bin boundary and the width of each bin to be derived. The low gain stage which cover the larger diameter particles is calibrated using PSL beads. The programable feature of the bin boundaries in the SPP200 electronics is used to "zoom in" on the size distribution from nebulised PSL beads. This allows the cross section f the PSL to be measured with increased accuracy in PCASP internal units (PIUs). Plotting scattering cross section of the PSL beads vs bin centre of the mode bin in PIUs gives a straight line. The equation of this straight line can then be used to define the scattering cross section for any value of peak height in PIUs.

Where only a DMA calibration has been performed the calibration can and has in the past been provided as a set of upper and lower boundaries for each bin defined in terms of diameter of a particle with a given refracive index. It is left to the user to adjust these diameters based on the refractive index of the material they are sampling as described below under Mie Correction. When calibration has been provided above 1 micrometre the concept of a bin boundary defined in terms of diameter is nonsensical due to the non-monotonic nature of the Mie solution. Hence the bin boundaries are now provided as a set of upper and lower boundaries in terms of the parameter directly measured by PCASP - scattering cross section. Again it is the user's responsibility to convert these values to diameter if needed based on the refractive index of the sampled aerosol. A method for performing these calculations is defined below under Mie Correction. This calibration data will be included in the netCDF file in the future but for now contact Phil Rosenberg for the latest version.

Data

Data from the PCASP is available for each of FAAM's flights in the core_cloud_phy v500 netCDF file. As with all FAAM netCDF files the first revision of a file will include r0 in the filename, the number increasing if new revisions are required. To date only data from PCASP2 have been made available in the v500 files because PCASP1 has not flow since the v500 files have been used. All PCASP variables begin PCAS2. The data is referenced against a time variable called PCAS2SPM which provides the time in seconds past midnight on a given date. Note also that another variable called Time is included in the NetCDF which gives integer second past midnight. This variable is provided for easy cross referencing with the nearest data point in the core data file when time offsets of +/- 0.5 seconds are not important and also to allow plotting with NCAR's Aeros software. The data consists of concentrations for each of the 30 bins (do not forget that the data from bin 1 should be discarded) under the variables PCAS2_nn with percentage uncertainties for these concentrations under the variables PCAS2_nn_err (replace nn with the channel number). The uncertainties represent possible offsets and are not independant from one data point to the next, i.e. they cannot be reduced by averaging over multiple data points. The data file also contains the sum of the concentrations over channels 2 to 30 in the variable PCAS2CON the volumetric flow rate within the PCASP in the variable PCAS2_FL and a flag to indicate data quality in the variable PCAS2_FLAG. The flow rate variable can be used to approximately derive counting uncertainties by the equation

counting uncertainty=sqrt(PCAS2_nn/PCAS2_FL)

Note that this is only approximate and an extra variable will be added in the future giving counting uncertainties. Counting uncertainties can be reduced by averaging over multiple data points using the usual uncertainty propogation formulae.

The variable PCAS2_FLAG has the following meaning

0: Data is fine.
1: Absolute concentrations may be incorrect but all bins should be incorrect by the same factor so relative concentrations could be used. This could be because no temperature and pressure data was available to convert measured mass flow to volume flow (in which case the last known good value is used instead) or the flow meter is outside its calibrated range.
1: The ratio of air speed inside/outside the subsampler is outside the range originally tested by Belyaev and Levin for calculating the inlet efficiency.
2: No true air speed is available for performing inlet efficiency calculations. The last good true air speed data has been used.
2: The reference voltage (used as a measurement of the laser power) is low. This could be caused by low temperatures, poor alignment or dirty optics. The effect will be to allow electrical noise to be falsely generate particle events creating false particles in the lower bins.
3: The data is unusable or does not exist. In this case the data are replaced with the value -9999. This can occur if data has not been recorded for this time (e.g. during preflight or postflight), the total flow through the instrument is so high that it is likely to cause turbulence in the optical cavity or the true air speed is low indicating we are onthe ground (in this case the inlet efficiency calculations are not valid and actually tend to infinite for still air).

Nominal, and calibrated lower and upper bin boundaries with uncertainties are also provided in the netCDF file in variables called PCAS2_D_L_NOM, PCAS2_D_U_NOM, PCAS2_D_L_CAL, PCAS2_D_U_CAL, PCAS2_D_L_ERR and PCAS2_D_U_ERR. The refractive index of the aerosol used to generate the calibrated boundaries is given in the variable PCAS2_D_L_RI and PCAS2_D_U_RI. All these variables are indexed against the bin number in a variable called PCAS2CH.

Mie Corrections

As discussed above PCASP measures the scattering cross section of particles rather than directly measuring their diameters. For homogeneous spherical particles of a known (complex) refractive index the relationship between scattering cross section and diameter exposed to light of a known wavelength is defined by Mie theory and will be referred to here as a Mie curve. For boundaries defined in terms of diameter it is important to correct the values for the refractive index of interest. These corrections can be large especially for particles which absorb light (those with non-zero imaginary parts of their refractive index). For boundaries defined in terms of scattering cross section the boundaries may need to be converted into diameter in order to derive further parameters. The following method is suggested.

If boundaries are defined in terms of diameter then convert them to scattering cross section. Boundaries greater than approximately 1 micrometre should be regarded with some suspicion due to the non-monotonic nature of the Mie curve above this diameter.
Define a Mie curve for the refractive index of interest. Use this to determine which diameters fall within a particular bin. There may be a number of sub ranges with gaps in between for diameters greater than 1 micrometre.
Determine the mean diameter of particles which would fall into each bin. This is referred to as the bin mean diameter.
Sum the width of all the sub ranges of which fall within the bin. This is referred to as the bin width or bin range.

The bin mean diameters and widths can now be used to plot size distributions and to define other properties such as area or volume concentration or can be used in radiative transfer calculations etc.

In order to aid the user with such corrections FAAM can provide two software tools. Mie Scattering Table Generator uses Mie theory to define a table of scattering cross sections as a function of diameter based on given instrument optical geometry, instrument laser wavelength and particle refractive index. Cross Section to Diameter Converter reads a table of scattering cross section data and uses this to determine the bin mean and bin width from bin boundaries determined in terms of scattering cross section. Cross Section to Diameter Converter can use scattering data created by Mie Scattering Table Generator or other data provided by the user (if for example Mie theory is not applicable). Both programs are provided as Windows executables, however, C++ and FORTRAN source code is included and should compile on other operating systems (but requires the wxWidgets GUI libraries).

Both Software packages are available from the FAAM Documents -> Software section or via these direct links

CStoDConverter_1.1 (Cross Section to Diameter Converter)

MieConScat_1.1.1 (Mie Scattering Table Generator)

Please Note, if you have version 1.0 of CStoDConverter please download the latest version as a bug in version 1.0 causes slight errors in the output. Subsequent versions also allow calculations where uncertainties in the calibration data are set to zero.

References

Belyaev and Levin, 1974, Techniques for Collection of Representative Aerosol, Aerosol Science, Vol 5 pp. 325-338