Quick
fitting
This is an example how to fit a segment of diffraction pattern
containing a few overlapped peaks.
- Start the program
- Go to Environment - Directories
and check/modify the working directory. This is not
necessary, if the correct directories are defined in Configuration
file. In principle, only the DAT and PRN directories
must be defined.
- Go to Files - Load - Run (short-cut
FLR) and select the data file from the list
(pathname can also be modified here).
- Go to Graphics (G) main menu
- Select the segment, if necessary, by X-scale
change (XL) in some way - window, explicit left and
right limits. Go back to main Graphical menu (ESC).
- Give zero approximation of background - Corrections
- Background - Global (CBG), select left and right
points by Graphical cursor. Go back to main Graphical
menu (ESC).
- Find the peaks - Lines - Search peak -
Find peaks (LSF), modify the threshold level with
graphical cursor if necessary, give typical line width
for smoothing. Markers of found peaks are shown on the
top. Go back one level in the menu.
- Delete false peaks if necessary with Erase
peak (select a peak with graphical cursor and press
Delete for approapriate peak). Peak parameters can be
viewed and changed under Parameters menu. Peaks
also can be added by Add peak.
- Go back to main text menu (ESC) and go to Fitting
- Fit (TF). Iterations are started and can be
repeated, if unsatisfactory results are obtained. Global
parameters and analytical function can be defined under Global
parameters. Starting profile parameters can be
modified and/or fixed in Peaks.
- Go to Graphics and look at the
results (separated Ka1 lines can be viewed by Alt K, differences by
Alt D, separated doublets can be switched on only in
Fitting-Graphics). If you are unsatisfied, you can
again go to Fit (TF).
- Finally go to Output - Write - File -
Run (OWF). Give filename, select parameters which
will be listed in the table and press Run. Graph can be
saved in HPGL format in Plot - Filename - Run.
This may be the fastest way of fitting the
profiles. As the non-linear regression is applied, zero-estimate
of parameters must be given (background - item 6, profile
parameters - item 7). Always the peaks which are contained in the
latest graphical window are fitted.