Cosmic Origins Spectrograph Instrument Handbook for Cycle 17

7.2 Safety First: Bright Object Protection

---

Photon-counting detectors are vulnerable to physical damage if hit with too much light at one time. Well before that flux level is reached, excess light leads to poor results because the electronics cannot handle the high event rates (the dead-time correction). In the case of COS' FUV detector, the very high gains mean that over-illumination of an area on the detector leads to charge depletion that can permanently impair the sensitivity of the detector at that point. This is also true to a lesser degree for the NUV MAMA detector.

For all these reasons COS has stringent count-rate limits that all observations must conform to.

7.2.1 FUV bright object protection

There are five levels of protection for COS FUV:

1. At the lowest level are the screening limits imposed on observers in order to provide a margin of safety for the instrument. The screening limits are set at a factor of two below actual risk levels, and we expect observers to work with us to ensure these limits are adhered to. They are determined by estimating the expected count rate from an object, both globally over the detector, and locally in an emission line if appropriate.
2. At the next level, within COS the "Take Data Flag" is monitored during an exposure. If an event occurs that causes the TDF to drop (such as loss of lock), then the COS external shutter is commanded closed.
3. Next comes local rate monitoring. It is possible to permanently damage a localized region of the micro-channel plates without necessarily exceeding the global rate limits. This could occur if an object with bright emission lines were observed, for example. The flight software in COS takes the FUV spectrum, collapses it in the cross-dispersion direction to create a one-dimensional array, and then bins the data by 4096 pixels at a time. These bins are then examined to see if they exceed a threshold value. The limit set is 75 events sec^-1 per resel.
4. Global rate monitoring is next. The COS flight software monitors the total event rate for both FUV detector segments. If that rate exceeds a threshold, the high voltage to that segment is shut off and the external shutter is closed.
5. At the highest level, the instrument is protected by the software sensing an overcurrent condition in the high voltage; this shuts down the detector entirely.

7.2.2 NUV bright object protection

The same five protections also apply to the NUV MAMA with two exceptions. With the MAMA, the local count rate check is performed by taking a short exposure, with the flight software then examining that exposure for groups of pixels with excessive counts. For the global rate monitoring, the software checks to see if the total count rate exceeds 77,000 in 0.1 sec and if so the high voltage to the MAMA is turned off. The highest level of protection is done by the detector electronics, which performs a Bright Scene Detection. If the total counts within 0.138 msec exceeds 17,000, then the MAMA high voltage is turned off and the external shutter is closed.

Note that all of the higher level safings that shut off high voltage and close the external shutter also turn off any calibration lamps that may be on.

7.2.3 Screening limits

Screening limits are the count rate limits that we at STScI expect observers to adhere to, in order to provide a margin of safety in instrument operations. Screening limits of two kinds - global and local - and both limits must be adhered to. These limits were provided in COS Quick Reference Guide, in the discussion of the detectors. They are:

Table 7.1:

Table 7:

Detector	Source type1	Type of limit	Limiting count rate2
FUV	predictable	global	15,000 per segment
		local	40 per resel
	irregular	global	6,000 per segment
		local	40 per resel
NUV	predictable	global	30,000 per stripe
		local	75 per pixel
	irregular	global	12,000 per stripe
		local	75 per pixel

1"Predictable" means the brightness of the source can be reliably predicted for the time of observation to within 0.5 magnitude. 2Entries are counts per second.

COS count rate screening limits

Bear in mind that these are "screening limits," which means that if a target is predicted to cause counts in excess of these rates, then a more thorough check must be made. There are two higher limits that are important. First, a factor of two above the screening limits is the practical operation limit, the level we will not knowingly allow an observation to exceed, so as to provide a margin of safety for COS. In addition, if the FUV detector is used in TIME-TAG mode, significant data drop-outs occur when the count rate exceeds 21,000 per segment. The highest of these rate limits are those specified in the HST Constraints and Restrictions Document (CARD). If the CARD limits are exceeded on-orbit, the software and hardware within COS turn off the high voltage to the detector.

If a target is too bright to observe in the Primary Science Aperture (PSA), it may be possible to observe it with the Bright Object Aperture (BOA), which attenuates flux by a factor of approximately 200. However, the neutral density filter in the BOA also degrades the optical quality of the source image, reducing the effective resolving power for a point source by a factor of 2 to 3.

If a target is safe to observe in the PSA but is too bright for a straightforward acquisition with ACQ/IMAGE mode in the NUV channel, it is possible to acquire with an attenuating mirror, with the BOA, or with both; this is described below.

7.2.4 Risks from nearby objects

It is not sufficient for just a potential target to be safe to observe, for nearby bright objects pose a risk as well. There are three scenarios:

• Given errors in the initial pointings of objects with HST, even with good coordinates, an unintended source may end up in the PSA. With good coordinates, objects beyond 5 arcsec should not pose a risk.
• Even without errors, a bright object could unintentionally end up in the other COS science aperture. This is a particular risk if the BOA is the aperture in use because a fainter-but-still-bright object could end up in the PSA. An very bright source could possibly fall in the BOA when it is the PSA that is in use, of course.
• Finally, a nearby source that is very bright could throw enough light into the PSA to cause problems. Here we adopt the same criterion as used for STIS. The region of concern is an annulus that extends from 6.5 to 15 arcsec from the center of the PSA. Any object falling in this annulus may not produce a global count rate per second in excess of 1 × 10⁵ per segment for the FUV or 2 × 10⁵ per stripe for the NUV, nor a local count rate over 200 per resel (FUV) or 400 per pixel (NUV). These limiting count rates are those estimated with the ETC as though the source were in the center of the aperture.

To guard against the risks imposed by these scenarios, observers are required to use the tools in APT to certify that no potentially UV-bright objects lie within a zone that could cause problems. In some cases it may be necessary to choose a specific ORIENT for the observation to ensure that nearby bright objects cannot fall in a COS aperture.