5.2 TIME-TAG vs. ACCUM Mode

COS exposures may be obtained in either a time-tagged photon-address mode (TIME-TAG), in which the position, arrival time, and pulse height (for FUV observations) of each detected photon are saved in an event stream, or in accumulation (ACCUM) mode, in which only the positions of the photon events are recorded.

To optimize the scientific return of COS, we recommend that all observations be performed in TIME-TAG mode, employ the default wavelength-calibration procedures (i.e., TAGFLASH=YES), and with all four FP-POS positions (See Section 5.8) of each CENWAVE setting for the FUV gratings. Observers who wish to employ non-optimal observing techniques must strongly justify their observing strategy in the "Description of Observations" section of their Phase I proposal. A modest reduction in observational overheads will not normally be a sufficient justification for adopting non-optimal observing techniques. Whenever possible, observers should use the recommended techniques, even if this requires requesting additional resources in their Phase I proposal.

5.2.1 TIME-TAG Mode

In TIME-TAG mode each photon is kept as a separate event in a list in COS memory. Each entry in that list contains the (x, y) coordinates of the photon together with the pulse height of the charge cloud generated by it (for FUV observations). Time markers are inserted in the list every 32 ms by the instrument electronics. When data are processed by the ground system arrival times are assigned to the events according to the time marker preceding the event.
COS observations should be obtained in TIME-TAG mode whenever possible because it provides significant opportunities for temporal sampling, exclusion of poor quality data, and, for the FUV, improved thermal correction (by tracking the stim-pulse positions as a function of time) and background removal (by using the pulse-height information). TIME-TAG mode should always be used for exposures that will generate count rates of 21,000 count/s or less from the entire detector (including both detector segments for the FUV). At count rates between 21,000 and 30,000 count/s, TIME-TAG may be used to obtain properly flux-calibrated data, but the loss of some continuous time periods within extended exposures will occur (see the discussion of the buffer time in Section 5.4). At present, TIME-TAG should not be used for count rates greater than 30,000 count/s. ACCUM mode should be used for such high count-rate targets.

We recommend that TIME-TAG mode always be used with FLASH=YES (the so-called TAGFLASH mode) unless circumstances prevent it (see Section 5.7.1).

Doppler Correction for TIME-TAG Mode

No on-board corrections are made for shifts in the spectrum due to the orbital motion of HST while in TIME-TAG mode. This is done later in pipeline processing.

Pulse-Height Data for TIME-TAG

In FUV TIME-TAG mode the pulse height of each photon event is recorded, along with its position and arrival time. Pulse heights are stored as 5-bit words, so their values range from 0 to 31. Post-observation pulse-height screening is useful for rejecting unwanted background events, and can often improve the S/N ratio in the extracted science spectrum. Pulse-height information is not provided by the NUV detector.

5.2.2 ACCUM Mode

In ACCUM mode an image of the detector is stored in a 16-bit memory buffer. As each photon arrives from the detector the location in the buffer at coordinates (x, y) is incremented by one. Each location can hold at most 65,535 counts; the next event will cause the value to roll over to zero. To conserve memory only a 16,384 × 128 region of each segment is stored, as well as the regions around each stim pulse. Timing and pulse-height information are not saved, preventing the application of the data-correction techniques available in TIME-TAG mode. For example, no walk correction is possible, so there could be distortions in the wavelength scale due to gain variations.

ACCUM mode is designed for bright targets whose high count rates would fill the on-board buffer memory too rapidly if the data were taken in TIME-TAG mode. In some instances one may observe a relatively bright object in TIME-TAG mode by using the BOA instead of the PSA, but the BOA degrades the spectroscopic resolution. The BOA is offered as an available-but-unsupported mode (see Section 5.13 for further discussion of BOA usage for FUV science observations). Observers wishing to use ACCUM mode will be asked to justify doing so when submitting their Phase II program.

Observing Efficiencies with ACCUM

FUV ACCUM images do not include the entire detector. To conserve memory photons are collected only from the stim regions and that portion of the detector actually illuminated by the target (1/8 of the full detector area, or 128 pixels in y). Each FUV ACCUM image fills one-half of the total COS memory, so it is possible to acquire two FUV images before dumping the on-board buffer.

NUV ACCUM images cover the entire detector. Because they are smaller, up to nine of them can be stored in the memory buffer. Unlike TIME-TAG mode, no data may be acquired during an ACCUM readout. NUV ACCUM mode is thus most efficient when repeated identical observations are stored in memory, then read out all at once. (Within APT, the Astronomer’s Proposal Tool, one may easily schedule multiple iterations of an exposure using the Number_of_Iterations parameter.)

Doppler Correction for ACCUM Mode

In ACCUM mode, the COS flight software adjusts the pixel coordinates of detected events to correct for the orbital motion of HST. The correction (always by an integer number of pixels) is updated whenever HSTʹs velocity with respect to the target changes enough to shift the spectrum by an additional pixel. This is done via a small table, computed on the ground, that lists the time of each pixel shift based on the orbital motion and the dispersion of the grating in use.
Note that ACCUM mode exposures longer than 900 s that use the G130M or G160M gratings may blur the FUV spectra by 1 to 2 pixels (about 1/6 to 1/3 of a resolution element) since shifts are performed in pixel, not wavelength, space.

Pulse-Height Distribution Data for ACCUM Mode Observations

Some pulse-height information is available for FUV ACCUM observations. A pulse-height histogram, consisting of 256 bins (128 bins for each detector segment) of 32 bits each, is dumped for every ACCUM mode image obtained with the FUV detector. (In ACCUM mode individual pulse heights are stored as 7-bit words, so their values range from 0 to 127, whereas in TIME-TAG mode pulse heights are 5-bit words.) Pulse-height data are not provided for NUV exposures.

Reducing ACCUM data in the COS Pipeline

The CalCOS pipeline does support ACCUM data, but the extraction uses the older "BOXCAR" method, while TIME-TAG uses "TWOZONE."  The use of BOXCAR means that there will be more "bad" columns than might be expected when compared to a TWOZONE extraction of a TIME-TAG COS spectrum at the same location/cenwave.