Anomalies seen in the Plots created by rad_compare updated 5/21/01 dd ------------------- Key: ** - anomaly not yet explained or investigated OK - anomaly is explained !! - anomaly is or could be a "real" measurement change :) - no unexplained anomalies ------------------- MA1025 (010301/001207) OK - zero jump at 0.93 keV (high zo Chi2) 4/4-OK: SSD/fitting problem, reanalyzed, lowered Chi2 OK - first second-order point consistantly ~12% low in all regions 5/8-OK: fits look good. Error on ref ratio is 6% one sigma - ref stat error could create this easily. Note that the Chi2 is not large for the regions. :) MA1026 (010301/001101) :) MB1148 (010302/001117) !! - zero-order change at MB1148m4 +15% at 0.93 keV 4/4 data look alright, real change or just stat fluke? OK - first-order 0.93 keV 10% high in m4 and all regions !! 5/4 Conclude that it's probably either: i) some ref/PSPC thing common to '1148 and '1158 ( and also showing up in some HEG tests...) OR ii) it may be a rad effect that BOTH '1148 and '1158 experienced, if they were both rad-exposed. MB1158 (010302/001117) OK - second order Chi2 large - poor fitting of small signals 4/4 exclude poor fits/measurements :) MA1022 (010307/001116) :) ME1409 (010307/001127) OK - large second-order Chi2 in m=-2 Mg in region p8 5/8 low number of second order counts gave rise to a poor fit (115 counts in ROI yet gaussian fit gave 134...) :) HC2214 (010308/001101) OK - large second-order Chi2 5/7 3 regions OK, 2 regions CHi2 < 6.0. :) HC2259 (010308/001101) OK - minus 20% change in first-order 0.93 keV m=+1 effic in p8, !! other regions too 5/4 improved HEG Cu-L line fitting and effect is more like only -10% averaged over the five regions. Unfortunately we don't have the 001101 m=+1 Cu-L measurements for HC2214 which was tested at the same time as '2259... This effect feels like the MB1148 problem of Cu-L measurement systematic effects... 5/7 The HX220 ratios are within 3% for 001101 and 010308, so most of the 10% effect is from the grating... and counting uncertainty of few per cent. OK - second-order large change in region m8 at one energy, 1.49 keV 5/8 A complete mis-fit of the 2nd order 010308 distribution: 634 ROI counts yet Gaussian fit is only 167. HB2104 (010309/001113) !! - first-order in p4 region looks 5-10% low over all energies -- this is isolated to one test region of the grating -- don't see anything like this in other gratings/regions, what's the cause? Same effect is reducing the first two second order points (1.25 and 1.49 keV) in p4? 5/21 Given the way the measurements are made (an Energy is selected, then the regions on each grating are tested), it is unlikely for an instrument effect to show up in all Energies but for a single grating/region. Will check the grating itself to see if the p4 region has obvious blemishes, etc. that would make it sensitive to a small offset when re-inserted in the equipment. OK - large zero-order Chi2, 4 of 5 regions have 5% drop in effic at 1.25 keV 5/21 SSD had mystery peak in MgO spectrum - could explain this small consistant change. OK - large second-order Chi2 in region p0 at m=-2 4.5 and 6.4 keV !! 5/21 2nd order 4.5, 6.4 keV measurements OK at a glance for HB2104 and HX220 reference grating; these measurements can have large errors due to fitting few counts plus continuum... HE2426 (010309/001115) OK - large second-order Chi2 in m8 (Al-K +1 order) 5/7 re-analysis reduced the Chi2. :) HE2412 (010312/001115) OK - large zero-order Chi2: 5% drop in effic at 1.25 keV 5/7 OK: looks statistical... :) HF2517 (010312/001114) OK - first-orders at 0.93 keV: one higher, one lower by ~15%. !! 5/4 re-analysis reduced the effect a little... a la '2259. 5/7 The reference measurements are only 4% different, so it appears to be mostly in the grating... OK - large zero-order Chi2: ~5% increase in effic at 1.25 keV ! 5/21 The measurements/analysis look OK; but the SSD was showing a mystery peak in MgO Direct spectrum, so could also be an instrument effect... OK - second-order Chi2 at one energy in m8 region 5/21 This is m=-2 order at 4.5 keV: analysis plots show the 010312 data were fit with too narrow a Gaussian causing the lowered "after" value. -------------------------------------------------------------------