5.6  Measurement Process

The measurement process of the NAIS is fully automatic. The user only needs to specify the measurement cycle.

The measurement software controls the instrument according to the measurement cycle, receives data from the instrument, processes it and produces measurement records and particle distributions. The data is continuously stored to data files. The software continuously performs diagnostic checks and notifies the user if something might require attention.

Figure 1: General overview of the measurement process of the NAIS software

5.6.1  Connection

The measurement software will always try to establish a connection with the instrument and start measuring unless the program is stopped manually. If the connection is lost for any reason, the software will keep trying to reconnect until it succeeds.

It is safe to switch off or disconnect the instrument while the measurement software is running. The program will automatically reestablish the connection and resume measurements when the instrument becomes available.

5.6.2  Measurement cycle

The measurement cycle definition determines the order and durations of the operating modes that the instrument should run in.

The total period of the cycle is typically 2 – 5 minutes, that is the total duration of all operating modes specified. There should always be one offset measurement of 30 – 60 seconds and one or more particles, ions or other measurements-

Figure 2: Typical measurement cycle of the NAIS. Offset measurements should be done for 20 – 30 seconds about every 5 minutes. Otherwise the user is free to choose the operating modes and their durations.

For example for long term monitoring the cycle “particles 120, ions 120, offset 60” would be recommended.

The instrument continuously measures all electrometer signals and diagnostic channels about 15 – 20 times per seconds and produces raw measurement records. The raw records are processed and turned into averaged records by the measurement program.

The program will always produce block average records: i.e. one average record from start to finish of each element in the measurement cycle. Additionally the user may specify any number of fixed length averaging periods. By default 1 second and 10 second averages are produced in addition to the block averages.

Figure 3: Raw records are measured continuously. Averaged records start when the instrument has settled after an operating mode switch.

When the instrument switches operating modes it takes up to 10 seconds settling time before proper results are produced: the chargers and filters need to stabilize at their new state and the new properly conditioned particles need to reach the measurement electrodes. The settling time is included in the durations given in the measurement cycle. So a 20 second duration in an operating mode will give a bit over 10 seconds of actual measurement results.

Therefore it is recommended that the instrument remains in each operating mode for at least 30 seconds. Otherwise a large amount of measurement time is lost just waiting for settling.

The processed electrometer signals from the averaged records are inverted to produce the particle or ion distributions. The results are produced continuously but at first they are considered preliminary. Final results are calculated after the next offset measurement has completed and the offset signal estimates have been updated (See Current Measurement).

Preliminary results are not stored to output files, they are only visible in the measurement program. The finalized data is saved. When the measurement program quits the preliminary records are stored immediately without waiting for another offset cycle.

5.6.3  Records

The measured data is represented as measurement records. Each records corresponds to a period of time and contains aggregate values of all the data that has been measured during that period.

The averaged measurement records of the NAIS are made up of following fields:

5.6.4  System variables

The status and health of the instrument are indicated by system variables (may also be called diagnostic parameters). All signals measured from the instrument sensors and all signals used to control the instrument are stored as system variables. Common variable types are:

Sensor voltages
The actual voltages output from sensors as measured by the data acquisition system.
Physical values
The physical value of the parameter what the sensor measures.
Control signals
The output signals used to control some component of the instrument (e.g., blower power)
Feedback target voltage
Calculated sensor voltage that is the target value for automatic control channels. For example the flow rate sensor voltage that would match 60 l/min. The respective control signal is adjusted so that the sensor voltage would match the target voltage.
Timing parameters
Timing data about the raw measurement rate.

5.6.5  Diagnostic flags

Many instrument parameters are automatically monitored and warnings will be issued by the software if some condition is not met.

The warnings represented by so called diagnostic flags. In essence, the flags are simply short text messages. The flags may have different severity:

Note A normal event occurred. For example electrometer reset.
Warning The instrument requires attention. It might not be measuring correctly.
Failure The instrument is definitely not measuring correctly.

5.6.6  Spectra

The particle and ion distributions are calculated from the measured electrometer currents using a data inversion procedure.

The distributions are represented by the number density spectrum vector nd.

\[ nd_i = \left.\frac{dn}{d\mathrm{log}_{10}d}\right|_{d=d_i} \qquad \text{(for size distributions)} \]

\[ nd_i = \left.\frac{dn}{d\mathrm{log}_{10}z}\right|_{z=z_i} \qquad \text{(for mobility distributions)} \]

The diameter \(d_i\) or mobility \(z_i\) points cover the measurement range quite densely and they are spaced evenly in the logarithmic scale of size (or mobility).

Fraction concentrations can be found using integration: \[ F_{d_0, d_1} = \int^{\mathrm{log}d_1}_{\mathrm{log}d_0} \left( \frac{dn}{d\mathrm{log}d} \right) \cdot d\mathrm{log}d \]

(or log z for mobility in case in ion spectrum).

The piecewise linear approximation can be used, so that the fraction concentration between consecutive diameter or mobility points is simply the surface area of the trapezoid under the number density function segment: \[ F_{d_i, d_{i+1}} = \frac{\left(y_i + y_{i+1}\right)}{2} \left( \mathrm{log}_{10}d_{i+1} - \mathrm{log}_{10}d_i \right) \]

Figure 4: Number density