Pyreos ezPyro SMD, User Manual, Page: 12 / 18

ezPyro
TM
Flame Detection Evaluation Kit User Manual
Page 12 of 18 © Copyright Pyreos Ltd. 2020 ezPyro
TM
User Manual Rev. 3.1
Window Size controls the speed of the algorithm vs. the stability to momentary fluctuations. A value of 256 is the
lowest value available, please see
Table 1
in section 10 for a justification of this lowest value. A lower value of window
size gives faster response, but also less immunity to noise, both from the sensor and electronics and from the scene
seen by the sensors. A higher value increases stability and glitch rejection.
FFT Bin Width this informs the user of the current frequency groupings imposed by the algorithm parameters. For
example, a 0.977Hz/Bin means that if a frequency bin were centered at 10 Hz it would contain frequencies in the
range 10 ± 0.977/2 Hz. So, this bin would contain signals of both 9.8Hz and 10.2Hz.
Time Length of Window This value is indicative of the response time of the algorithm. It is simply the sample rate
(figure 14 shows 1ms) multiplied by the FFT Window Size (1024).
Bin Centers inform the user of the first and last FFT bin centers being used by the algorithm. Since the value input by
the user is an integer it will never be a perfect match for the bin centers. Therefore, the algorithm uses the first bin
that includes the frequency input by the user.
High Freq Cutoff This is field is set to the value for the upper limit of frequency to be used by the algorithm.
Low Freq Cutoff This field is set to the lower limit of frequency to be used by the algorithm.
Threshold Ratio allows the user to set the threshold ratio. The example value of 2.0 requires that a flame channel’s
post processing signal strength must be twice that of the rejection channel for a flame event to register.
Signal Multiplier allows each channels signal strength to be modified. This should be used for calibration purposes to
account for differences between the sensors on each kit.
7.5.3 FFT Related Parameters Quick Guide
The input parameters that the user can change that affect the performance of the FFT are
Sample rate/low power mode and FFT window size affect both the maximum frequency detectable and the time
duration of a window of certain size. Please refer to
Table 1
in section 10 for the effect of these parameters.
High Freq Cutoff/Low Freq Cutoff These parameters allow the user to select which frequencies are included in the
algorithm. This is easily visualized by showing the response of the software to a change of these parameters.
Figure 16: 1Khz Sample Freq - 1024 FFT window - 4 to 16 Hz Summation Range
The above figure shows the bins being used by the algorithm coloured in red with the bins being ignored coloured
black. To see the above graphs of the FFT of each sensor follow the instructions of section 7.5.1 for the Fourier
transform view.