Specifications

EAW Smaart 6 Operation Manual Analysis Modes and Display Types

Vector averaging works directly with the complex data FFT data from the transfer

function calculation. Vector is more effective at rejecting uncorrelated noise and rever-

berant energy and tends to correlate better to the intelligibility and accuracy of signal

reproduction. Vector is more sensitive to wind and speaker/source movement and other

time-variant issues, so it is better suited to indoor venues and/or calmer, more controlled

conditions. Note that Vector is available only for Frequency Response measurements.

Spectrum measurements use RMS averaging in all cases.

Averaging Schemes

The three basic averaging schemes for Frequency Response and Spectrum measurements

are linear First In, First Out (FIFO), Infinite, and exponential (Fast, Slow and variable).

FIFO averaging is a simple arithmetic average of some number (2, 4, 8, 16...) of the

most recent FFT frames with equal weight given to each. The settings for FIFO averaging

are in multiples of two because every doubling of the number of frames going into the

average increases the S/N of the measurement by 3 dB. If the Avg field is set to 1, no

averaging is performed and only the data from the most recent FFT frame is plotted.

Infinite (Inf) averaging also gives equal weight to each FFT measurement included in

the average but rather than including only a fixed number FFT frames, infinite averaging

keeps a running of average of all the FFT data that comes in until the averaging buffers

are flushed. Averaging buffers are flushed automatically when averaging, FFT size, or

sampling rate parameters are changed, when the analyzer stops, display modes are

changed, or by pressing the V key on your keyboard.

Unlike FIFO and infinite averaging, exponential averaging gives more relative weight to

the most recent data going into the average while the weight of the oldest data decays ex-

ponentially. The Fast and Slow options are exponential averaging routines with a fixed

half-life modeled on the characteristics of time integration circuits in standard sound level

meters. The Exp option is similar to these two but has a user-definable half-life. The half-

life for the Exp option is specified on the Inputs tab of the main Options dialog box.

Each doubling of the number of averages increases the S/N of the measurement 3 dB

until the absolute noise floor of the SUT or the measurement system, whichever is higher,

is reached. Note that increasing the number of averages also slows down real-time displays.

As a general rule, the more difficult the measurement conditions, the more averaging

and smoothing is required. Electrical measurements (i.e., comparing the input and output

of an EQ or system processor) typically require very little averaging so reducing the

number of averages allows the display to respond quickly to filter changes. Acoustic

measurements (i.e., those using a microphone) typically require 16-32 FIFO averages.

When making acoustic measurements in noisy, reverberant, or outdoor spaces, either

increase the FIFO averages to 64 or 128 or use the Infinite averaging option.