### SDAST MEETING #29 ### DSRI Copenhagen 9-10th August 2004 Present: Søren Brandt, DSRI, Copenhagen Carl Budtz-Jørgensen, DSRI, Copenhagen Jerome Chenevez, DSRI, Copenhagen Paul Connell, GACE, Valencia Peter Kretschmar, ISDC, Geneva Niels Lund, DSRI, Copenhagen Sami Maisala, University Observatory, Helsinki Silvia Martìnez-Nuñez, GACE, Valencia Carol Anne Oxborrow, DSRI, Copenhagen Niels-Jørgen Westergaard, DSRI, Copenhagen Absent: Stefan Larsson (sick) Tim Oesterbroek (holiday) Astrid Orr (holiday) ## ISDC News and Views (PK) --> see PK presentation. ## Instrument Status (SB) ## JMX-1 has been activated since March 5th 2004, orbit 170 with the new selection criteria. JMX-2 is dormant since orbit 171 expected for some testing on the gain evolution. 1% increase on the gain per orbit driving to lowering the HV. Gain increase cause could be a change of glass plate conductivity by Chromium. The main consequence of the gain increase is the lost of energy resolution. The changes on energy resolution could be modeled with a non-linear function. More accurately the energy resolution has a linearly increasing term added in square (see handout). The lost on energy resolution is greater at higher energies, the background line at 22 keV is getting broader on time since the line at 7 keV is more or less constant. JEM-X2 has been dormant and reactivated briefly after 26 and 87 days. Rev. 208 a hot stripe appeared. It looks like the plate has suffer "aging" in a short time scale (~ hours). JEM-X2 takes one orbit to reach the gain stable state -- plate state. JEM-X 1 gain evolution: It seems to follow the same trend that JEM-X 2, 1 % per orbit (glass effect) The aging factor is now 1.8 for JMX 1 and 2.8 for JMX 2. The energy resolution trend for unit 1 is more unclear. An increase of 0.1 % per month. Calibrations sources 1 and 3 are showing different energy resolution -- is the spatial gain changing ? AI29_1 NL/SB/CBJ 1st Sept 2004 To provide Peter with a list of bad periods where OPEN JEM-X data in one or both unit may be suspect NEW SELECTION CRITERIA: Hot Spots Since orbit 170 --> the new selection criteria produces artificial hot spots. It looks like that the hot spot activity happens at unphysical energies ! The hot spot usually has a low impact in the TM, but introduces some time variable background. The new selection criteria handles hot spots LESS well, as they are open to low energy noise. JEM-X2 270 orbit re-activation a hot stripe occurred below PHA 50 (< 3 keV). The hot stripe events do not have a good Y position value, positions are random in Y Should check the energy range of the events forming the hot spot in PV data -- revolution 20 --. ARE ALL THOSE EVENTS WITH A PHA < 115 There are two different selection criteria, one below PHA 115 and one above this channel. This kind of criteria introduces a break in the background a channel 115. This break is fixed but the new criteria could introduce gain dependencies around 5-8 keV (?). It seems that we get a 5% more photons below 115 than above 115, BUT are those photons real or just BKG ? The events in the analysis shown are real and coming from the small area in the calibration box where background can be neglected compared to the cal source. The "8 Mhz" incident ... during 1.5 orbits (206, 207 scwid 42) the DFEE CPU was running in the 8 MHz-1Ws state instead of the standard 16 MHz-0Ws. The main effects of running in this mode were: the dead time increased by a factor 2.5 and we got "funny" spectrum at the lower energies (not understood, may be an electronic problem ...). BE CAREFUL IF WE ANALYSE THIS DATA SOME DAY ! Operational wish list: - Finer sampling of the vignetting function (will be implemented soon, to some degree) - Switch off dormant DFEE (JMX-2) to operate active unit at lower temperature (briefly tested, under evolution, since the drop in the temperature could affect also ISGRI). - Raise alert level of On-Event-Message signaling problems with DFEE recovery. - Experiments with VC setting to check possible gas contamination --> may be, better resolution. - May be, a lower cut-off onboard to avoid most of the hot spots --> needs to be discuss ! - Handling the IREM crashes Conclusions and concerns: - Loss of anodes strips is not anymore a concern. - Hot spots is not a serious concern. - Gas degradation may be checked by drift voltage. - Aging of the plate is a concern !! AI29_2 SB/CBJ 1st November 2004 Determine whether new `open' rejection criteria OPEN have actually improved data quality (more real events vs. more noise) JC: JMX-1 lost two anodes in 5 months !! ## Calibration and correction improvements for OSA 4.0 (CAO) ## j_cor_gain has a gain smoothing model with 5 separate time zones handling the following: -pre-switch on and full switch on -emergency switch ons -eclipse season warm up -missing pre-switch on -glitches even big ones -smooth variations in gain j_cor_position finds hotspots dynamically -'hot pixels' must be very strong ... - and have strong neighbours --> do not want to remove shadowgram spots, for instance for a off-axis source will have only a few part of the detector illuminated therefore the illuminated area could be consider as a hot spot easily. j_calid_adc includes a better handling of real and outlier events. - Few false alerts and none dead anodes alert - more realistic indications of nonlinearity - the IMOD has been updated accordingly. New gain smoothing model --> each revolution is divided in five time zones. These zones can be fitted usually by a linear fit + 4 harmonic/decay fits. It can handle until 3 emergency switch on per revolution. It seems that glitches in gain evolution are getting bigger on time and therefore leading to a energy resolution degradation ! ## Discussion on spectral extraction below 5 keV ### At the moment we can not determinate Nh with JEM-X data, could this issue be solved ? ## Rejection criteria and their effect on spectral response (SB, CBJ) ## SB has mentioned this issue in the previous point of the agenda. The spectral response is affected by the "break" around channel 115. ## Spectral response (NJW) ## NJW should include spectral resolution changes on the RMF ! The seen "break" by the new selection criteria is affecting the ARF. The ARF is showing a "bump" around ~ 6 keV ! This effect is seen since 4th Crab calibration period (rmf version 0017). Piotr Lubinski has done some analysis of Crab data --> he has found that normalization is 5% too low, else a good match. BUT this issue was due to a different selection effect: rowSelect="(STATUS==0)" VERSUS NJW standard: rowSelect="(STATUS<256)" ! SMN should check which selection criteria she is applying in her analysis. NJW should give an appropriate selection criteria for spectral selection. PK suggests to get a new tool for getting a local ARF for a given set of data. This tool will be necessary for periods with extraordinary status, hot spots, solar activity ... and therefore we will lost a lot of events due to our selection criteria and the obtained flux will be lower than the expected, therefore the normalization will not match ! ## Contents of SVR (NJW, PK, ALL) ## Systematics issue should be clearly defined given numbers ! TO DEFINE THE SYSTEMATICS -- who is going to do this ?? NJW suggests to do some analysis testing with OSA 4.0 software and he has set up an automatic procedure for the analysis. The results of the analysis can be found under DSRI at pub/njw/svr_special ! Test cases: light curve stability, imaging flux stability, detection of weak sources, spectral shape, effect of bkg correction, timing, misalignment and deadtime. ## User support (PK, ALL) ## --> see PK presentation. ## Improved modelling of the JEM-X collimator (NL) ## To improve off-axis response it is necessary to get an improved modelling of the collimator. In this new improved model NL has considered 3 new effects: - Tilt of the collimator lamellae (2 par) - Non-flatness of lamellae (1 par) - Spherical deformation of the collimator (1 par) --> to allow that the collimator is bend. It could be that the collimator has a lower temperature than the microstrip plate leading to a bending of the collimator. To optimize the method NL used Crab data (rev 43-44, 102 and 170) and he obtains with his lc program the strength of Crab in each ScW. The new parameters of this method are used to estimate the effective area. There are 16 parameters involved in the optimization: coll_pitch --> the height of the collimator 7.5 mm coll_width --> the width of the lamellae* The width of the collimator for both detectors are different to the nominal value (0.45 * 0.5 mm) --> optimized parameters are: 1. 0.45 2. 0.375 ! This parameter handles the non-flatness of the lamellae. mask_height --> NL found the best fit for a non-nominal value (nominally: 3401 mm) --> 3393.0 (distance from Be window to the mask) The mask height influences the found position of the sources. alpha, beta, gamma --> rotation of the mask ? xshift, yshift --> a shift of the collimator over the microstrip plate. c_tiltx, c_tilty --> the tilt of the collimator walls. coll_curv --> the curvature of the collimator mask_scale --> hexa_dim * mask_height phase2 --> twist of Common-X wrt star-tracker offX,offY --> misalignment in pixel dimensions. The efficiency is more or less 50% higher for revolutions 102 and 170 than for 43 and 44. And this different should be due to the different selection criteria, the data from rev.43 and 44 were obtained with the launch selection criteria. Applying the new modelling of the collimator leads to an increase of 20 to 50 % on the strength of the sources(*) found by IROS software and the number of spurious source candidates (> 5 sigma) are reduced by a 10%. (*) the weaker sources normally shows the largest increase on count rate with the new modelling ! SMN should talk with NL about his new software to re-analyse SMC data ! SMN should talk with NJW about the OSA 4. refined imaging software ! SMN should talk with JC about the mosaicking software ! -- Flux extraction of sources -- NL is using 3 methods: 1. Spectral extraction software: it assumes that the position of the source is corrected and it takes the signal in "illuminated" pixels and subtract corresponding signal from "shadowed" pixels. 2. IROS imaging software: in this case the source position is free and it fits source strength with the "best" source model 3. Image analysis: it counts the total number of counts in an array of 6*6 pixels around the source. Normalization problem is not yet addressed, i.e., there are differences in the obtained fluxes between the 3 methods. The most stable method seems to be the first one. NL introduces a constraint in his method to estimate the fluxes of the sources. Once a photon is assign to a given sky position, it can not be assigned to any other position, i.e., it is "considered" anymore by the back projection program. ## Fine resolution imaging (NJW) ## There are different on the shadowgrams output in two imaging methods what it is confused by the users. This issue should be solved for OSA 5. 1 arcmin resolution is getting with this new software in source position determination. The mask is either 0 and 1 and the during the deconvolution you get directly the variance. The image is flatfielding by an exposure map that it should take into account the vignetting effect (?). Next changes for OSA 5 will be: - Source finding and flux determination will be implemented - Get sensitivity as a function of place in the FOV - See how the new mosaicking software works with the new software SMN: Ask PK if makes sense to re-run imaging analysis with fine resolution imaging ! I think that no for the PhD. ## Mosaicking (JC) ## The new version is number 5 and it was distributed last week. This new version handles NJW refine images. This new version includes a "preliminary" vignetting function that it is auto-computed. The program handles 3 different kind of images with a new parameter: <0 NJW inputCode =0 CBJ >0 NL (1,2,3) NL program creates 3 different kind of images Flux determination issue --> the signal (cts/s) from a given map is compared with the signal that corresponds to the Crab. For the future a remapping tool will be implemented to handle geometric deviations (?). PK strongly proposes to contact Roland and gets an official release that it includes JC efforts. ## Background modelling (SM) ## - Major IT meltdown at Helsinki so no regular presentation - SM showed some contour figures of the behaviour of various line components of the background: He, Mo, Fe, U - Major background component is a continuum presumably modulated by the mask pattern, and extracted from the rest of the background using the old (coarse) vignetting function. - Outer most edges of detector can't really be fitted - Above 26keV see clearly effect of onboard curtain gain correction: in some strips upper energy events are simply missing - XSPEC fits the line components and Sami plots these as function of radius and as contour plots across the detector area. - Xe line intensity rises by factor of 2 towards edge of detector and is single strongest line component. CBJ notes that of 20 cts background only about 1 cts is Xe - Ni drops slightly towards edge of detector - Mo is more or less uniform across detector, as expected by collimator position - Uranium daughter products at ~ 13keV - All SM's models should be taken into count by SL's programs but SL not present to discuss this AI29_3 SM 1st Sept 2004 SM to make zipped package of all his background spectra OPEN and XSPEC fits, and send it to all SDAST members AI29_4 SB/CBJ 1st Nov. 2004 To make a wish list of things they'd like to OPEN see presented for background modelling for the next meeting ## Shared software (PK) --> see PK presentation ## JEMIROS status (NL) ## NL is working to have a JEMIROS running inside the ISDC system. PhC gave to him a version based on an old midisky version and NL is working with his latest version. NL has installed OSA 4 in his laptop from the binary package. He will try to do from the source code. ## Vignetting problem ## (ALL) --> the vignetting problem is still on-working. NL will provide to the rest of the team a set of functions to handle the vignetting issue. ## Event status filtering (ALL) ## Which is the appropriate event filtering selection for spectral extraction ? We are discussing which event flag values should be used for the different analysis levels. J_COR_BAD_ANODE_SEG --> REMOVE. J_COR_BAD_OBT --> A.I. CAO to contact Nicolas Produit to discuss which are the appropriate event flag for OBT problem. SB also will be involved in this discussion. J_COR_BAD_ANODE --> 16 OK. It changes to 32. J_COR_BAD_GAIN_PARAMS --> it was defined to handle problems in j_cor_gain to estimate the gain. With new version of CAO software this flag could be obsolete. J_COR_BAD_GAIN_PIXEL --> to handle individual pixels that were defected. REMOVE. J_COR_BAD_XPOS_BIT --> " _Y" they are obsolete. REMOVE. J_COR_BAD_CAL_SOURCE --> It is OK. J_COR_PRE_SETUP --> OK. CAO flags all the events at the beginning of a ScW but without checking if their time stamps and therefore without knowing if the events belong to the given ScW or the previous one. If they belong to the given ScW they can be included in the deconvolution analysis. J_COR_OUTSIDE --> OK. J_COR_HOTSPOT --> CAO mentions that the hotspots are only presented in the low energy range. CAO can change the flag to handle the energy of the event that falls in a hotspot region. J_COR_BAD_NEIGHBOUR --> OK. It changes to value 16. NJW/CAO suggest to count the number of events that are threw out with the different flags. VIP. We really need to be able to throw away as few events as possible since the determination of flux will be defined by the received number of events since the effective area is a constant in the software. And another way to handle the problem will be to estimate the effective area for each scw !! PK suggests that the order of status flag should be changed to be more consistent. AI29_5 PK/CAO 1st Sept. 2004 Check whether L-OBT jumps caused by HK rollover OPEN are handled by NP's j_prp_obt programs AI29_6 CAO/JC 1st Oct. 2004 Make spectra of the different types of bad OPEN events (esp. bad anodes and bad anode neighbours) to see just how bad these really are. AI29_7 CAO 15th Aug. 2004 Update jemx.h with new STATUS flag values OPEN and distribute to all SDAST Should we have a shared function to determinate the effective area for each ScW? NJW should make RMF with STATUS==0 for filtering of the events. Everyone should CHECK THE LATEST STATUS FLAG TABLE BEFORE RUNNING A NEW ANALYSIS. ## Science Results (NL) ## Sgr A* Galactic Center (Goldwurm et al) --> it is not possible to solve the mystery of the GC, it can not solved if there is one or more than one source. In this observation they are looking for X-ray flares from Sgr A*. NL shows mosaicking with ISGRI of the GCDE in 2003 and 2004, there are different between both images. IGR J19140+0951 --> A&A 2004 Hannaikainen paper. Two different spectral states have been detected for this source. INTEGRAL observation of 3EG J1736-2908 A&A 2004 Di Cocco et al. Discovery of hard non-thermal pulsed X-ray emission from the anomalous X-ray pulsar 1E 1841-045. Kuiper et al. astro-ph 0404582 2004. Core programme during AO-3: 1.3 Ms Galactic Centre 2*5*5 pt dither at 45 deg - 2.17 deg spacing* *F. Lebrun requests to have a wider step on the pattern for removing code mask. 1.4 Ms Scutum Arm 10*5 points, 2.17 deg spacing 0.8 Ms Norma Arm 10*5 points, 2.17 deg spacing 1.3 Ms Galactic latitude scan 5*16 points, 2.0 deg spacing ? Ms Galactic Plane scans every 12 days, step size 7 deg |b| < 6.45 deg 1.0 Ms Deep Extragalactic Survey 5*12 pointings, 2.16 deg spacing (Virgo, 3C273, 3C379) VIP. AO-3. IF WE WANT TO ASK AGAIN FOR AN OBSERVATION OF THE SCUTUM REGION WE SHOULD ASK FOR A DIFERTHING PATTERN WITH STEP SIZE ~ 1 DEG TO ALLOW JEM-X TO HAVE ALL THE FIELD INSIDE THE FOV. ## Science Results (NJW) ## Survey work --> NJW is running all the accessible data on the archive to create detected source catalog, spectra and lc. Perseus cluster (SN1006) GRS 1915+105 A.I. DISCUSS WITH PK HOW TO SAMPLE OR NOT THE SPECTRA AND HOW TO OBTAIN THE SYSTEMATIC ERRORS. ## Science Results (JC) ## 18 detected sources in the GCDE in the energy range 14 to 26 keV in a mosaicking with a total exposure time ~ 300 ks. ## Science Results (PK) ## --> see PK presentation ## Plans for OSA 5.0 ## PK --> see peter.sxi CAO --> j_calib_gain_fitting: -- anti-glitches (noise) -- add HV values to the gain history table j_cor_gain -- flag all events with PI below 3 keV j_calib_adc -- handle odd ANODE_SWITCH values j_dead_time_calc -- handle mixed 8 MHz data (?) -- make IMOD values of wait factors for modes slower than the fastest. NJW --> To implement source finding and flux determination for j_ima_fine_resol. To update j_ima_shadowgram to not use more than once time a given event. Various SPRs and SCREWs New IMOD files Collimator modelling* --> it should be included in the IMOD table PK suggests to include NL modelling collimator parameter values should be included as keywords in the IMOD table. Hopefully better vign.corr. Get rid of skew shadowgrams New RMF files SM --> new models will be delivered. But Sami needs inputs for Stefan using his current models to be able to improve them. AI29_8 CAO/SMN 15th Aug. 2004 Compile minutes of meeting and put them on the OPEN DSRI document page ## Action Items ### AI29_1 NL/SB/CBJ 1st Sept 2004 To provide Peter with a list of bad periods where OPEN JEM-X data in one or both unit may be suspect AI29_2 SB/CBJ 1st November 2004 Determine whether new `open' rejection criteria OPEN have actually improved data quality (more real events vs. more noise) AI29_3 SM 1st Sept 2004 SM to make zipped package of all his background spectra OPEN and XSPEC fits, and send it to all SDAST members AI29_4 SB/CBJ 1st Nov. 2004 To make a wish list of things they'd like to OPEN see presented for background modelling for the next meeting AI29_5 PK/CAO 1st Sept. 2004 Check whether L-OBT jumps caused by HK rollover OPEN are handled by NP's j_prp_obt programs AI29_6 CAO/JC 1st Oct. 2004 Make spectra of the different types of bad OPEN events (esp. bad anodes and bad anode neighbours) to see just how bad these really are. AI29_7 CAO 15th Aug. 2004 Update jemx.h with new STATUS flag values OPEN and distribute to all SDAST AI29_8 CAO/SMN 15th Aug. 2004 Compile minutes of meeting and put them on the OPEN DSRI document page AI29_9 PK 1st Sept. 2004 Provide binned spectra created with assorted OPEN ARFs ## Agenda ## SDAST Meeting #29: DSRI, Copenhagen 9-10th August 2004 Monday 9th August 2004 13.00 Welcome and adoption of agenda (NL) 13.10 ISDC News and Views (PK) 13.40 Instrument Status (SB) 14.10 Calibration and correction improvements for OSA 4.0 (CAO) 14.30 Shared software (PK, ALL) 15.00 Refreshment break 15.20 Rejection criteria and spectral response (SB, CBJ) 15.50 Spectral response (NJW) 16.20 Contents of SVR (NJW, PK, ALL) 16.40 User support (PK, ALL) 17.00 End of day 19.00 Supper at `Los Flamencos' Tuesday 10th August 2004 9.00 Fine resolution imaging (NJW) 9.30 JEMIROS and Modelling the Collimator for imaging (NL) 10.30 Refreshment break 10.50 Mosaicking (JC) 11.10 Background modelling (SM) 11.40 Imaging with and without background subtraction (NJW) 12.00 Lunch 13.00 Refined source extraction (PHA channels and energy resolution, `over-sampling'; ray tracing; photon tagging etc.; with and without background subtraction) (SL) 13.30 Vignetting problem and off axis fluxes (NL) 13.40 Vignetting problem and off axis fluxes (SL) 13.50 Vignetting problem and off axis fluxes (NJW) 14.00 Science Results (NL) 14.10 Science Results (SB) 14.20 Science Results (NJW) 14.30 Science Results (SL) 14.40 Science Results (SMN) 14.50 Science Results (JC) 15.00 Refreshment break 15.20 Science Results (PK) 15.30 Plans for OSA 5.0 (PK, CAO, NJW, SL, SM) 15.50 AI List (CAO,ALL) 16.00 Next meeting (ALL) 16.10 AOB (ALL) 16.30 End of meeting