Caveats about ACIS Onboard Event Filtering
Posted: 26 June 2008
For a variety of reasons, many events are never telemetered to the ground from the spacecraft. The following set of filters are applied, in the order listed, to discard events that may have little or no scientific value.
If the bias value (see acisf*bias0.fits) associated with a pixel has a value of 4094 or 4095, then events are not reported for the pixel. A value of 4094 indicates that a bias-parity error has occurred. Typically, no pixels have bias-parity errors. A value of 4095 indicates that the pixel is included in the onboard bad-pixel list or that the pixel has been identified as hot by the onboard software. For TIMED exposure mode observations, the pixels that have bias values of 4095 usually include
CCD_ID CHIPX CHIPY ------ ----- -------- 0 71 1-1024 0 72 1-1024 0 73 1-1024 0 303 253 1 303 1-1024 1 409 1-1024 5 383 585 5 792 351 7 496 441-1024 7 497 441-1024 7 498 441-1024
This list, which is not user adjustable, can differ from the bad-pixel list in the CALDB that is used for processing the data on the ground. An observation-specific bad-pixel file (acisf*bpix1.fits) includes pixels for which the bias values are 4094 or 4095, bad pixels and columns in the appropriate CALDB bad-pixel file and other bad pixels (see the ahelp file for acis_build_badpix). The number of bias-rejected events is not included in the telemetry stream.
If an event has an acceptable bias value, but the summed pulse height (PHA) is not within the globally acceptable range, then the event is discarded. The acceptable range for an observation, which is user adjustable, is described by the keywords PHAMIN and PHARANGE in the header of a Level 1 mask file (acisf*msk1.fits). For example, if PHAMIN = 39 and PHARANGE = 3710, then the pulse height of the event must be at least 39 adu and cannot exceed 3749 adu (PHAMIN + PHARANGE). Otherwise, the event is not telemetered. The lower limit of the range must be greater than the event threshold (typically 20 adu for BI CCDs and 38 adu for FI CCDs) and is often dictated by the threshold. The upper limit cannot be greater than 32767 adu and is often set to 3749 adu. Note that it is possible, after reprocessing on the ground, for events to lie outside the specified range by a few adu. The number of pulse-height rejected events per CCD per frame is recorded in the column DROP_AMP of a Level 1 exposure statistics file (acisf*stat1.fits).
An event that has acceptable bias and pulse-height values may be subsequently rejected because the FLTGRADE for the event is not one of the acceptable grades. A bit-encoded list of acceptable FLTGRADEs, which is not user adjustable, is described by the keywords GRADEMA* in the header of a Level 1 mask file (acisf*msk1.fits). An example of how these keywords are used is included at the end of this document.
For most observations, the set of bad FLTGRADEs is 24, 66, 107, 214 and 255. Events with these FLTGRADEs are often dominated by the charged-particle background instead of astrophysical sources. The number of FLTGRADE rejected events per CCD per frame are recorded in the column DROP_GRD of a Level 1 exposure-statistics file (acisf*stat1.fits). The values in the columns DROP_AMP and DROP_GRD are mutually exclusive. Events are not double counted.
An event that has a valid bias value, that is within the acceptable pulse-height range and that has an acceptable FLTGRADE may be subsequently rejected because the event does not satisfy one or more of the conditions associated with an optional, user-specified window at the chip location of the event. Within a rectangular window the global acceptable pulse-height range can be modified and/or a fraction of the events can be discarded. The pulse-height range of the window cannot extend beyond the lower or upper limits of the global acceptable range. It is only possible to narrow the acceptable pulse-height range. The information about the optional windows used, if any, are described in the binary tables of a Level 1 mask file (acisf*msk1.fits). For example, one mask file for ACIS-S3 includes
CHIPX CHIPY ------------- ------------- SHAPE (min) (max) (min) (max) SAMP_CYC PHAMIN PHAMAX ---------- ----- ----- ----- ----- -------- ------ ------ rectangle 186 325 12 149 1 23 2292 rectangle 1 1024 1 1024 0 21 3749 !rectangle 186 325 12 149 1 23 2292
The first line indicates that all (SAMP_CYC = 1) events with 186 <= CHIPX <= 325, 12 <= CHIPY <= 149 and 23 <= PHA <= 2292 are telemetered. The second and third lines specify that no events from the rest of the CCD should be telemetered because the SAMP_CYC = 0. If the SAMP_CYC = 2, 3, ... or N, then every second, third, ... or Nth event in the window is telemetered. The others are discarded. Often no windows are used. In this case, the binary tables of the mask file indicate that all but the outer edge of a CCD are valid. It is not possible to detect events with CHIPX = 1 or 1024 or with CHIPY = 1 or 1024. If the DATAMODE is VFAINT, then it is also not possible to detect events that have CHIPY = 2 or 1023. The number of window rejected events per CCD per frame are recorded in the column DROP_POS of a Level 1 exposure-statistics file (acisf*stat1.fits). The values in the columns DROP_AMP, DROP_GRD and DROP_POS are mutually exclusive. Events are not double counted.
Example: how the flight grade keywords are used
The following is an example of how the keywords GRADEMA* are used to describe which FLTGRADEs are acceptable. If the hexadecimal values of the keywords are
GRADEMA1 = FEFFFFFF GRADEMA2 = FFFFFFFF GRADEMA3 = FFFFFFFB GRADEMA4 = FFFFF7FF GRADEMA5 = FFFFFFFF GRADEMA6 = FFFFFFFF GRADEMA7 = FFBFFFFF GRADEMA8 = 7FFFFFFF
then, the binary representations are
GRADEMA1 = 11111110111111111111111111111111 GRADEMA2 = 11111111111111111111111111111111 GRADEMA3 = 11111111111111111111111111111011 GRADEMA4 = 11111111111111111111011111111111 GRADEMA5 = 11111111111111111111111111111111 GRADEMA6 = 11111111111111111111111111111111 GRADEMA7 = 11111111101111111111111111111111 GRADEMA8 = 01111111111111111111111111111111
(i.e. 7, B, E and F correspond to 0111, 1011, 1110 and 1111, respectively). Since the mapping between the FLTGRADEs 0-255 and the keywords GRADEMA1-8 is
GRADEMA1 = 31 ... 0 GRADEMA2 = 63 ... 32 GRADEMA3 = 95 ... 64 GRADEMA4 = 127 ... 96 GRADEMA5 = 159 ... 128 GRADEMA6 = 191 ... 160 GRADEMA7 = 223 ... 192 GRADEMA8 = 255 ... 224
the five zeros in the binary representation above are for FLTGRADEs 24, 66, 107, 214 and 255. This set of FLTGRADEs is the set that is most often identified as "bad." The events associated with these FLTGRADEs are typically dominated by charged-particle background instead of astrophysical sources.