FAQ on FOC Count Rates and Linearity

Q:  For an extended source, can you go above 1 count/sec/pixel in the
    full format and still correct for non-linearity?

A:  The 1 c/s/pix applies to point sources, not extended. For the
full format, the limit is only 0.15 c/s/pix. The correction for
non-linearity is only a first-order correction and is best
suited for extended sources at this point.

Q:  When do serious effects start having an influence, effects such
    as damage to the detectors, residuals from previously saturated
    images,...?

A:  Bright object limits have been set for the FOC to avoid such
damage or residuals. No extended source delivering more than 100
photons per second per pixel over the whole photocathode may be
observed with the FOC. Similarly, no point source delivering
more than 1e+6 photons per second per pixel at the photocathode
will be imaged with the FOC. These values correspond to an 9th
magnitude blue star or an extended source of surface brightness
10 mag per square arcsecond viewed through the F430W filter
with the F/96+COSTAR relay. See Section 6.6 of the FOC
Instrument Handbook for more details on Detector Overload.

Q:  What is the difference between overflow and saturation?

A:  Overflow is what happens in formats that have only 8 bits
available to count photons (i.e., all formats that have 512x1024
pixels). When the count reaches 255 the next count will cycle the
value for that pixel back to 0 from which it continues counting
up. Saturation is the result of nonlinear behavior which
depends on the count rate. It is possible to have overflow
without much saturation or visa versa, or both.

Q:  Can I avoid saturation by taking shorter exposures (like a CCD)?

A:  No! Saturation for the FOC depends on the *count rate* not total
counts.

Q:  It appears I have some overflows in my image. Can I fix them?

A:  That depends. For images that have extended features that are
overflowed, as long as the data have not overflowed more than
once or twice (and is not significantly saturated), it is
usually straightforward to reconstruct the actual data values.
This is usually done in some ad hoc fashion depending on the
details of the image. For overflowed stars or compact sources,
it can be very difficult, if not impossible, to unambiguously
correct the overflows.  BY FAR THE EASIEST WAY TO DEAL WITH
OVERFLOWS IS TO AVOID THEM IN THE FIRST PLACE (i.e., split 8
bit exposures into short enough exposures).

Q:  It appears parts of my image are saturated. Can I correct for it?

A:  Perhaps in certain circumstances, but don't count on it. The
saturation behavior of the FOC is complicated...very
complicated. As with overflows, THE BEST WAY TO DEAL WITH
SATURATION IS TO AVOID IT IN THE FIRST PLACE (no doubt of
little comfort to those of you for whom this advice comes too
late!). This can usually be done by observing the same field
with different ND filters as to be able to analyze objects with
a wider range of count rates rather than trying to obtain one
long exposure.

There are two limited cases where some correction my help. a)
the case of a very extended, somewhat saturated object. The
object must be extended enough so that it does not change in
intensity by more than a factor of 2 in about 5-10 pixels. b)
fairly isolated stars (more than 5-10 pixels from another
source of significant flux) that have mild saturation. In
either case if you see "black holes" or dips from saturation,
you are are outta luck. Saturation an be corrected for only if
less than ~50% of the flux has been lost due to nonlinearity,
and even then, accuracy is greatly reduced. Contact an FOC
specialist to get details on how to correct for these cases.

Q:  Can I reduce saturation effects by using a smaller format?

A:  Absolutely yes. The saturation curve scales inversely with
the number of pixels in a format, hence the 256x256 reaches the
same level of nonlinearity as the 512x512 format at a count rate
4 times higher. The use of zoomed pixels causes a format to reach
the same level of nonlinearity at about 4 times lower count rate
and thus is very costly in that respect.