Caveats about Vignetting sign error bug
![[IMPORTANT]](../imgs/important.png)
This bug was fixed in CIAO 4.14.
Summary
The CXC has identified a sign error when applying the mirror vignetting correction to the effective area in the following tools:
- mkarf: used to create on-axis point source ARFs.
- mkgarf: used to create ARFs for grating dispersed spectra.
- mkinstmap: creates instrument maps, used by mkexpmap to create exposure maps.
It does not affect mkwarf nor eff2evt. The specextract default is to use weighted ARFs (weight=yes), which uses mkwarf.
The bug causes an incorrect azimuthal angle, Φ, to be used: 180-Φ. This can introduce an error in the effective area up to several percent. The error is both spatially dependent and energy dependent. It affects both ACIS and HRC responses (exposure maps and point-source ARFs) as well as grating ARFs. This bug has always been in CIAO (not something recently introduced).
Examples
The variation in the vignetting component of the effective area is relatively small, generally a few percent peak-to-peak, as shown in Figure 1.
Figure 1. Vignetting vs. Azimuthal Angle
![[Plot showing vignetting vs. azimuth for 3 off-axis angles]](imgs/vign_vs_phi-01.png)
Figure 1. Vignetting vs. Azimuthal Angle
This plot shows the 1.5keV vignetting correction to the effective area vs azimuthal angle (Φ) for 3 different off axis angle (theta) values.
and varies with off-axis angle as shown in Figure 2.
Figure 2. Vignetting vs. Energy
![[Plot showing vignetting vs. energy for 3 off-axis angles]](imgs/vign_vs_e-01.png)
Figure 2. Vignetting vs. Energy
This plot shows the vignetting correction to the effective area vs energy for 3 different off axis angle (theta) values. The separate curves at each off axis angle are different azimuthal angles (Φ).
Taken together we can see the effect in an example exposure map for ObsID 5532, ACIS-2367. In Figure 3 we see the exposure map at 1.5keV. The aimpoint is on ACIS-7, with two -I array chips turned on. In Figure 4 we see the fractional difference between the exposure before and after the sign correction is applied. We can see that near the aimpoint the fractional difference is very small (less than 0.2%). At large off axis angles, the difference increases up to +/-2%.
Figure 3. Exposure Map for ObsID 5532 at 1.5keV
![[Plot showing vignetting vs. energy for 3 off-axis angles]](imgs/sign_err_expmap.png)
Figure 3. Exposure Map for ObsID 5532 at 1.5keV
The 1.5 keV exposure map for ObsID 5532. The aimpont is on ACIS-7, the chip with the highest effective area (red and white colors). The two separate chips are ACIS-2 and ACIS-3, part of the -I array.
Figure 4. Fractional Difference in Exposure Map
![[Plot showing vignetting vs. energy for 3 off-axis angles]](imgs/sign_err_frac_diff.png)
Figure 4. Fractional Difference in Exposure Map
The fractional difference in the 1.5 keV exposure maps before and after the sign correction. On axis the error is small (less than 0.2%), while further off-axis the error can increase to +/-2%
We can see the energy dependence by computing the ARF at an off-axis location 4 arcmin away from the aimpoint. Note: mkarf should generally only be used for near on-axis point-like sources; we used it here for illustrative purposes. Figure 5 shows the ARFs before and after the vignetting sign correction bug has been fixed. Figure 6 shows the percent difference in the ARFs. As we can see there is slight energy dependence to the percentage difference at this location.
Figure 5. An example ARF before and after bug fixed
![[An example ARF before and after bug fixed]](imgs/sign_err_arf_absolute.png)
Figure 5. An example ARF before and after bug fixed
The ARF on ACIS-7 for a location 4 arcmin away from the optical axis, at 10 degrees azimuth. The two curves are for the original ("Bad") vignetting correction (orange), and for the corrected code (blue).
Figure 6. Percent difference in ARFs
![[An example ARF before and after bug fixed]](imgs/sign_err_arf_percdiff.png)
Figure 6. Percent difference in ARFs
This plot shows the percent difference between the two curves shows in Figure 5. We can see that there is a slight slope to the percent difference vs energy below 6keV.
However, remember that typically for a source located this far from the optical axis, 4', users would normally use mkwarf which is not affected by this bug.
Conclusions
As we can see from these examples, the magnitude of the error is generally small, especially near the aimpoint.
Furthermore, the CXC Calibration team has done a comprehensive review of existing calibration products and have determined that this bug does not affect them. That is the current calibrations are not already in some how compensating for this bug.