Nod size: Like all infrared spectrographs, observations are taken in pairs (ab) or quads (abba) in order to remove sky emission. For small sources (angular dimensions are much less than the slit length of 14 arcsec), the telescope is nodded back and forth along the slit. For large sources the telescope is nodded between the source and blank sky. For pointlike sources nods are thus very small. If the source is initially centered along the slit, a nod of +/- 3 arcseconds will mean that the spectra are separated by 6 arcsec and are located 4 arcsec from the ends of the slit. Most people specify nods of +/- 2.5 or +/- 3 arcsec.

Offset pattern: The only offset patterns currently supported are AB pairs, ABBA quads, and slit scan mode (multiple positions along the slit).

Read mode: Normally the queue observer will select the read mode, choosing either one or eight low noise reads. In the 1-2.5 µm region individual exposures of faint sources that are more than a few minutes long should be read out with 8 low noise reads. In all other cases a single read should be used.

Slit width: With slit widths ranging from 0.17 to 0.34 arcsec (2-4 pixels), slit losses are an issue, even for the best seeing. Unless the highest resolution is scientifically required, the widest slit should be used.

Slit orientation: Due to differential flexure between Phoenix and the Gemini guide probes, it is highly recommended that the slit position angle on the sky be 90 degrees (slit oriented East-West) whenever possible . With that slit angle the drift of the target is mostly along the slit.

Avoiding array saturation: In the JHK bands try to avoid using calibration stars with JHK magnitudes brighter than ~5, as the acquisition images of brighter stars can saturate the array and have deleterious effects on subsequent observations of faint targets.