Data reduction for high precision photometry is challenging due to both instrumental and astrophysical effects. This website focuses on the known instrumental effects which prevent observers from achieving poisson noise limited observations. 

The strongest instrumental effect is the combination of an under-sampled PSF coupled with gain variations within a single pixel.  We first discuss the ability of the Spitzer spacecraft motions in reference to holding a certain position, which will minimize the effect of the gain variations on photometry.  Once the spacecraft is able to repeatedly point to a single position, and hold that position, it was possible to map out the gain variations for one particular pixel on each IRAC array.  Techniques for removing this source of correlated noise are referenced.  We show methods for measuring the improvement in photometry achieved by removing correlated noise. 
The majority of the work surrounding the intrapixel gain effect has been done for the brightest stars using short subarray frame times (less than or equal to 2seconds).  However, as exoplanet and brown dwarf science progresses, we are seeing more requests for longer frame times.  We include a discussion of the expected instrument performance at these longer frametimes.
The relative strengths of instrumental systematics are shown in Figure 1 (taken from Krick et al. 2015), and discussed in this section. We plot all known noise sources in units of electrons as a function of binning scale in number of frames on the bottom and in minues on the top, assuming 2s frametimes.  This figure includes the Poisson contribution from an assumed source at half-full well as the slolid diagonal black line and the background noise, along with the readnoise, as the solid diagonal gray lines.  The level of the intrapixel gain effect for both channel 1 and channel 2 are shown as the top horizontal gray lines (This is referred to as pixel phase effect in some literature).  An estimate of the level of uncertainty caused by bias pattern effects and persistent images (latents) is indicated with dashed gray lines.  Lastly, for reference, we show the rough levels expected for a hot jupiter transit signature, eclipse signature, and a 70ppm superearth transit event.
This plot shows why we work so hard to remove the intrapixel gain effect from the data, otherwise we would not be able to detect astrophysical signal.  It also shows what the next level of noise sources are which influence measurements of lower signals.
Figure 1: Noise in electrons as a function of binning scale showing the relative strength of instrument sytematics.  Noises from the source and background are shown as diagonal lines.  Other known instrumental systematics are shown as horizontal lines.  For reference, we show also the strength of a 1% transit, 0.1% phase, and 70 ppm superearth.