for chopping mode
used in PHT03, PHT22
When using chopped mode in AOTs P03 and P22 observers can vary only a few parameters. The most important is the total on-source integration time. Test measurements have shown that C100 triangular chopping with on-source integration time of 128 sec results in a noise level of order 10-20 mJy. In order to minimise vignetting, a small chopper throw with detector P3 is preferable.
for staring observations
used in PHT03, PHT22, PHT17/18/19, PHT37/38/39
In order to derive the source - background difference signal one has to estimate the background level towards the source as well as to determine the flatfield of the detector array. In ideal cases, 1 background measurement close to the source is sufficient for both purposes. Because of drifts, however, it is recommended to repeat the background observations 2-3 times in order to outline the drift curve.
In case of C100, if the source is fainter than 10 times the noise level, then the contribution of the source to the signals of the edge pixels is negligible. These pixels can be used to determine the background level as well as the measurement noise. In case of a brighter source, observation of a second background position is necessary in order to derive the noise. This is always the case for C200 as the airy disc of the source extends over all four pixels.
The detector should be kept switched-on between the background and source measurements. This can be performed by using sparse maps or 3-point raster instead of 2-3 independent staring AOTs. Otherwise there will be unnecessarily large transients introduced in the middle of the sequence.
One should avoid scheduling an FCS measurement between the background and source measurements in order to avoid drifts. Raster maps fulfil this criterion automatically. In case of a sparse map, the first AOT in the sequence (either PHT17 or PHT37) contains an FCS measurement at the end of its measurement sequence. To avoid the disturbing FCS one should point PHT17 or PHT37 on a background with a short dummy integration time (e.g. 16 sec), remain on this background position and measure the background flux with PHT18 or PHT38 which do not contain an FCS measurement, and then proceed with the rest of the sparse map pointings.
A measurement time of 128-256 sec per position is necessary to obtain a well defined drift curve and also to average out the C100 signal fluctuations which have a timescale of 10 sec. The first observation of the sequence is strongly influenced by transient drifts, therefore longer measurement times are recommended.
for raster observations
used in PHT03, PHT22
As a minimum, a 
raster with C100 and a 
raster with
C200 is recommended. Please be aware that, contrary to what has been written
in the ISOPHOT Observer's Manual, it is possible to use PGA to enter a minimum
stepsize of 35'' with the C100 array and 45'' with the C200 array. In this
mode the raster orientation should be aligned with the detector array. This
is achieved by setting in PGA the raster parameter ``Reference Frame: Y''.
For C100 a raster with 46'' step size with a minimum
integration time of 32 sec per raster step is recommended. Such a raster will
provide the flat-field of the array as an additional result. The source is
seen by all pixels giving a useful redundancy in case of a detection. (Note
that in case of staring or chopping the source is seen by only one pixel.)
For C200 the performance of a raster (DM=DN=90''), with a
minimum integration time of 32 sec per raster step, rather than a staring
observation may provide higher S/N, because the source will be placed at
the centre of each of the pixels instead of the centre of the whole array,
and each individual pixel sees about 2.5 times higher flux compared to a
staring or chopped measurement.