The programme named “psd-detrend_ECG-COMON.exe” calculates the potential power spectral density (PSD), the current PSD,
and the noise impedance Zn (square root of the ratio PSD V/PSD I) of an EN data file consisting of three columns (time
in s, potential noise in V and current noise in A) separated by tabulators. The programme divides the time record
in one or several sections over which it averages the PSDs. Furthermore it can apply a linear detrend of each section before
calculating the PSDs. It uses the configuration file “config_psd-detrend_ECG-COMON.txt”. These two files must be in the directory
where the data (noise ASCII file) is. Only the configuration file has to be modified according to your data. The
structure of the configuration file is the following:

file_in hanning (0 = no, 1 = yes) nb_section nbpoint_section detrend (0 = no, 1 = linear) file_out

(file_in: name of the EN data file (ASCII file); hanning: Hann window will be used (1) or not (0);
nb_section: number of sections in which the EN data file is divided; nbpoint_section: number of points/samples
in each section; detrend: a linear detrend will be applied (1) or not (0) on each section before calculation of the PSDs; 
file_out: name of the file with the PSD and Zn data)


When the detrend option is selected (detrend = 1), additional information is given in the final window about the 
importance/relevance of the drift in the voltage and current time records:

- SD is the standard deviation of the detrended signal (SD is wrong if calculated on the raw signal because of the 
  influence of the drift),

- aT is the signal drift over the duration T of the time record (a is the slope of the linear drift),

- drift/Vpp (or drift/App) is the ratio aT/(4 SD): 4 SD represents the peak-to-peak amplitude of the 
  fluctuation (indeed, the probability that the amplitude of a Gaussian signal with mean 0 ranges
  between -2 SD and +2 SD is 95.4 %, which is close to 100 %). This ratio compares the signal drift
  and the amplitude of the fluctuations. It only indicates if the drift was important in the raw data
  (roughly, the drift is important when the ratio is higher than 0.1). It brings no information on
  the linearity of the drift that must be checked by looking at the time record.



For example:

	Huet_RR1Mo_10Hz.txt 1 10 2048 1 Huet_RR1Mo_10Hz-PSD.txt

for a data file “Huet_RR1Mo_10Hz.txt” containing 20480 pairs of potential and current values. In that example, the data
file is divided in 10 sections of 2048 pairs of potential and current values, a linear trend removal will be applied and
the Hann window is used in each section. The result is stored in the file “Huet_RR1Mo_10Hz-PSD.txt” as follows:

data file = Huet_RR1Mo_10Hz.txt   hanning = 1   nb_section = 10   nbpoint_section = 2048   detrend = 1
  frequency	    PSD_v	     PSD_i	      Zn
4.882813e-003	1.901942e-011	2.974846e-024	2.528520e+006
9.765625e-003	1.913229e-012	9.364153e-025	1.429385e+006
1.464844e-002	5.789599e-013	5.085475e-025	1.066985e+006
...


Please note:

• no character "blank" in the names of file_in and file_out
• nbpoint_section must be a power of 2
• in the noise ASCII file the values of potential and current noise have to be in V and A (if not, ask F. Huet for a modified programme)



For modifications of the programme, please feel free to ask F. Huet (francois.huet@upmc.fr).