As mentioned above there are two kinds of models:
They have the dimension of a monochromatic photon flux,
which -- in this package -- is always measured in units of
photons/cm 
/s/keV.
These models have to be handled additively.
They are filter like functions with a range of unitless values between 0 and 1 and have to be used multiplicatively. If a filter function describes the galactic absorption a given red shift is ignored.
With these definitions a more general model f(E;p) can be constructed by adding and multiplying appropriate model components, as e.g.:
where different parameter lists 
are allowed to overlap.
Throughout this package a unit system of cm, s, and keV
with a Boltzmann constant of 
is used. This means that
temperatures are measured in keV. They can be transformed
into degrees of Kelvin by multiplying their values with the
inverse of the Boltzmann constant
1.1604... 
10 
Kelvin/keV.
The normalization amplitude of a spectral model function,
which cannot be normalized, is taken at a certain
reference energy 
, which can be chosen. The default value
of this reference energy is 
= 1keV. In all other cases
the model function is normalized to 1, so that the normalization
amplitude A represents the whole integral of the model spectrum
measured in a unit photons/cm 
/s. Exceptions are thermal plasma
line spectra
[Raymond and Smith1977, Kaastra and Mewe1993]
where the normalization is related to the emission measure.
In the following some standard models are described. Most of them are created in a parameter file with the command CREATE/PARFIL fit <par_file_name>, where they can be edited and combined in a proper way to a more complex model (making also use of the command WRITE/PARFIL.