synopsis: (1) residual systematic uncertainty in the sky measurement which depends on orbital position during measurement and space weather conditions. (2) residual systematic uncertainty in the FCS measurement causing an additional systematic photometric uncertainty for the sky measurements.

limitations and applicability:
The dark correction is highly uncertain at the end of a revolution. Not applicable to chopped or differential measurements where the sky background subtraction includes implicitly the dark signal subtraction. The background must be observed together with the target observation to ensure that the dark signals in source and background pointing are identical. The correction is not important for bright sources (cf. Table 1).

description:
The detector/CRE chain as well as the ionising radiation introduce an instrumental signal which causes a zero level offset (``dark signal''). The dark signal depends on the properties of the detector and the time of the observation in the science window. A time dependent dark signal is removed.

uncertainty/noise introduced:
Observations in which a differential flux is measured are not affected by uncertainties in the dark signal correction. Examples are: chopped observations, absolute photometry observations in which the dark signal measurement is included, maps covering discrete sources, etc. Also, for observations for which the sky signal is much higher than the dark signal, the uncertainties are negligible.

Residual uncertainties in the dark signal correction becomes important for faint targets whose signals are close to the dark signal.

Systematic uncertainties in the dark signal subtraction of the FCS measurement affects the responsivity determination for all AOTs except for the absolute photometry AOTs (PHT05, PHT25) where a dark or FCS straylight measurement was performed. The signal uncertainties of the dark and straylight measurements contribute then to the statistical uncertainty.

To be consistent with the reset interval correction (Sect. 3.6, all dark signals except PHT-S are transformed to the equivalent signal for a reset interval of $t_{reset}=\frac{1}{4}$ s. This introduces an additional uncertainty due to the residual uncertainties in the transformation. For some detectors the dark signal after transformation has become negative, this a consequence of the CRE properties and has no deeper physical meaning.

In Table 1 we give the equivalent fluxes for dark signals for each filter. Since the dark signal is time dependent, the average minimum value (column 2), average maximum value (column 3) along the revolution and the $1\,\sigma$ statistical scatter due to space weather effects from revolution to revolution (column 4) are listed for each filter. The dark signals increase along the revolution and the highest values are found for the end of the science window. A typical responsivity is assumed for each detector. For C100 and C200 the dark signals are available for each pixel. Since the dark signal values are similar for all C100 pixels except for pixel 6 we list pixel 6 and the average of the remaining 8 pixels. For C200 we list the average of the 4 pixels.

**Table 1:** Dark signal equivalent fluxes, see text for explanation of columns
	${\bf F_{\nu, min}}$	${\bf F_{\nu, max}}$	${\bf {\Delta}F_{\nu}}$
filter ID	mJy	mJy	mJy
P1_3.29	468	544	323
P1_3.6	93	109	64
P1_4.85	92	107	64
P1_7.3	69	80	47
P1_7.7	352	410	243
P1_10	262	304	181
P1_11.3	929	1080	641
P1_11.5	168	196	116
P1_12.8	385	447	265
P1_16	709	824	489
P2_20	126	254	23
P2_25	121	245	22
P3_60	-196	1144	52
P3_100	-142	827	37
C_50	-36	561	111
C_60	-26	404	80
C_60pix6	573	1245	156
C_70	-27	416	82
C_90	-11	175	35
C_90pix6	248	539	68
C_100	-19	301	60
C_105	-37	567	112
C_120	-97	-64	7
C_135	-63	-42	5
C_160	-50	-33	4
C_180	-105	-69	8
C_200	-318	-209	23

auxillary data:
Cal-G files: PPDARK, PSDARK, PC1DARK, and PC2DARK for the different subsystems and include estimates of the statistical uncertainty.

3.7 Dark signal subtraction