A&A 411, L153-L157 (2003)
DOI: 10.1051/0004-6361:20031259

IBIS Veto System

Background rejection, instrument dead time and zoning performance

E. M. Quadrini ¹ - A. Bazzano ¹ - A. J. Bird ² - K. Broenstad ³ - F. Di Marco ⁴ - G. La Rosa ¹ - M. Michalska ⁵ - P. Orleanski ⁵ - A. Solberg ³ - P. Ubertini ¹

1 - Istituto di Astrofisica Spaziale e Fisica Cosmica, Italy
2 - Departement of Physics and Astronomy, Southampton University, Southampton, UK
3 - Universitet I Bergen, Fysisk Institutt, Bergen, Norway
4 - VEGA IT GmbH c/o European Space Operation Centre-ESA, Darmstadt Germany
5 - Centrum Bada Kosmicznych PAN (PAS-SRC), Warszava, Poland

Received 14 July 2003 / Accepted 8 August 2003

Abstract
IBIS is the high energy imager on board the INTEGRAL satellite. The gamma-ray instruments on board will take advantage of the long uninterrupted observation made possible by the very eccentric orbit (10 000 $\rm {km}$ perigee and 152 000 $\rm {km}$ apogee). A disadvantage for orbits outside the protection of the van Allen belts is the exposure to cosmic and solar particles. Conversely, the background is quite stable throughout the 3 days orbit. In order to maximise the scientific returns and take full use of these almost 3 days continuous observations, IBIS is equipped with a light, very effective Veto System. This ensures a substantial reduction of the background due to the induced photon and hadronic component, in turn enhancing the detector sensitivity. The performance of the IBIS veto as evaluated during telescope commissioning is reviewed. In particular, the efficiency of background rejection and the resulting IBIS dead time are evaluated as well as the impact of different zoning configurations. Measured over the whole energy range, the veto system provides a background suppression effect of $\sim$50% for ISGRI and $\sim$40% for PICsIT. The definitive veto settings optimised for the operational working temperature and background conditions are described.

Key words: INTEGRAL - gamma-ray - veto - calibration

1 Introduction - The IBIS veto system

The IBIS veto system design and the results of laboratory tests have been described previously in papers Ubertini et al. (1996), Bazzano et al. (2003), Poulsen et al. (2000) and in technical documents IBIS User Manual (vol. 1), IBIS User Manual (vol. 2), IBIS FM VETO Calibrations (2001). The IBIS shielding system design derives from the challenging limited resource allocations in terms of weight and power. These led to the limitation of the active lateral shield height up to the bottom of ISGRI (the lower energy detector layer), to design the active Veto Detector Modules (VDMs) in four different shapes for shield leakage minimisation, and to use a passive shield between the mask and the detector. There are 8 lateral and 8 bottom VDMs.

	Figure 1: Rear veto assembly.
Open with DEXTER

Table 1: 511 $\rm {keV}$ line fit parameters for VDMs before and after final setting.

	Figure 2: IBIS showing lateral and rear veto.
Open with DEXTER

The bottom array and four lateral modules can be seen with their main components in Figs. 1 and 2. Each module comprises two BGO scintillation crystals optically glued along their long edge. The composite crystal is viewed by two PMTs with embedded Front End Electronics (FEA) and high voltage (HV) divider. Cross bars stiffen the two electronic units, i.e. Veto Module Electronics (VME) for control and signal conditioning and High Voltage Power Supply (HVPS). In each module the HVPS is distributed to the two PMTs whose signals, adjusted through two VME independent gain chains, are summed. The sum is delivered to the Veto Electronics Box (VEB) where all signals are discriminated, converted into strobes with adjustable length and delay, and distributed to the two detector layers ISGRI and PICsIT. The VEB also provides for module control and housekeeping (HK) data collection. Here, each HV, gain, discrimination level, strobe delay and width can be independently programmed. The strobes can also be grouped according to the proper zoning' configurations making the system very flexible and adaptable to different environmental conditions. Parameter optimisation and calibration are facilitated by collection of 256 channel VDM spectra from either a single selected VDM or from each VDM in sequence. Spectra may be acquired in coincidence or not with the On-Board Calibration Unit (OBCU) strobes.

	Figure 3: Background count distribution.
Open with DEXTER

2 Module equalisation

At the end of the first activation phase, the veto modules were equalised through the set of parameters selected during the Parameter Optimisation and calibration campaigns Bazzano et al. (2003), Poulsen et al. (2000), IBIS FM VETO Calibrations (2001).

A set of spectra were collected in coincidence with the Calibration Unit which provides a clear line of 511 $\rm {keV}$photons. Table 1 lists the parameters of the 511 $\rm {keV}$ line fits: peak position, energy resolution and standard deviation. Finally, there is the mean for all 16 modules. The energy resolution value is affected by threshold effects in the 8-bit ADC channels, enhanced by the resultant poor statistics due to the short integration time. At this stage, VDM 11 was operated with a low HV (1.0 $\rm {kV}$) set after problems that occurred during Thermal-Vacuum test in ESTEC.

Figure 3 shows the veto background count rates taken from each individual VDM. The figure shows the shape of the veto system background response, based on an irregular grid defined by the centre of each VDM BGO block and interpolated (average) values at the corners. The plot shows that the modules closest to the other instruments and the main spacecraft structures (-Z direction) have higher counts than the modules farthest away. This is due to secondary background production, the closeness to this source region located towards the centre of the spacecraft, and the greater solid angles for these VDMs with respect to the source region. The regularity shown indicates a successful calibration and setup of the veto system.

A number of tests were performed (Table 2) to compare the veto strobe counts for different ISGRI zoning configurations. For both ISGRI and PICsIT, veto all indicates that the detectors receive a logic OR signal from all lateral and bottom veto elements. Conversely, when ISGRI zoning is set to lateral, the ISGRI strobe derives from the logic OR of just the lateral VDMs. On this basis, counts from a single module can be meaningfully compared to the combination of VDM 1 and 16 (one rear, one lateral).

Table 2: Veto strobe generation tests.

The expected ratio between ISGRI and PICsIT veto strobe counters is 4.0 due to a division factor applied to the PICsIT counter. Nevertheless, the sum of bottom and lateral VDM counts are about 20% more than total veto counts coming from the "lateral.OR.bottom'' strobes. This provides a raw estimate of the cosmic particle contribution to the background. In fact, most of the particles, because of their energy, interact with two or more VDMs leading to a single strobe when the veto allconfiguration is applied. False coincidences contribute only about 4% of this difference. Corresponding multiple-site fractions on ground were 4% at Laben (of a total $\sim$9300 $\rm {c/s}$) and 2.5% at ESTEC (of a total $\sim$3700 $\rm {c/s}$).

	Figure 4: VDM temperatures after first activation.
Open with DEXTER

3 Temperature dependence

The veto first activation was performed in three main steps after an outgassing period of 15 days. First step was a check on two Modules (1 and 16) and on the two Calibration Units. The latter are the most critical being positioned on the tube wall far away from all other heat sources. Finally, all modules were checked one by one during one day test and then the whole system was activated.

Figure 4 shows that all module temperatures are between the warmer VDM4 and the colder VDM13 in a range of $\sim$8 $^{\circ}$C. CDM1 and 2 values are given for reference only. Figure 5 is a 3D representation of the spatial temperature distribution 12 hours after the activation.

	Figure 5: Veto spatial temperature distribution after 12 hours.
Open with DEXTER

The gradient between the veto modules is explained in terms of distance from different heating sources, position with respect to INTEGRAL axis, heaters or cooling straps, thermal exchange with other sub-systems, effect of sun aspect angle. It is also simple to observe the module differential heating following booster heater activation. What is important to stress is that each module has good thermal stability and the temperature difference between modules is constant along the orbit.

The thermal behaviour in different operational conditions is also important. The temperature variation during Perigee with and without Eclipse was assessed during commissioning. The veto is always off during the radiation belt passage for safety (i.e. below 60 000 km from the Earth). The not-operative temperature is as usual controlled by the INTEGRAL thermal control loop. The average variation in the Eclipse season is $\sim$1 $^{\circ}$C/hr for the VDMs. The strobe count rate is not appreciably affected by this variation (Fig. 6).

	Figure 6: Temperature evolution after perigee; with eclipse.
Open with DEXTER

When not in the Eclipse season, the VDM HVs are off for about 7 hrs during the perigee passage. In this case, with most of the Instrument on, the modules reach minimum values ranging between -12 and -2 $^{\circ}$C and the booster heaters are not activated. The plateau temperature is then recovered after the new switch on at 0.3 $^{\circ}$ $\rm {C/hr}$, while the single module temperature variations during the orbit are within 3 $^{\circ}$C. It is worth to note that the Veto System gain and strobe count rate stability comes directly from this good thermal behaviour along the whole orbit in all conditions.

4 Veto zoning and efficiency

There are two criteria on which the veto efficiency can be judged.

The first of these, reduction in count rate, i.e. detector background, is a crude measurement of the veto efficiency since it does not consider whether the reduction in count rate is at the expense of a poorer signal-to-noise ratio, or is being achieved, for example, just by increasing the detector dead time.

For the veto to induce an improvement in signal-to-noise, it must be seen to be preferentially removing background events and thus be performing at a better level than a simple grey filter. In the case of the zoning modes for ISGRI and PICsIT vetoing, it is this criterion which is appropriate. When making assessment of veto performance using in-flight data, it is crucial that dead-time corrections are made so that the improvement over a grey filter can be established. In the following analysis we refer to veto configuration and data collected during commissioning on 11th Nov. 2002.

The effect of each zoning mode can be assessed by comparing the reduction in count rate compared to the no veto case with the reduction in count rate due to extra dead time alone. Any additional reduction indicates an increase in vetoing effect that should translate to an improvement in sensitivity, provided that there is a real effect of vetoing on background and a negligible effect on the source to be observed.

Thus we can conclude that for ISGRI, the vetoing effect probably increases marginally from lateral to veto all modes, although the reduction in count rate is certainly largely due to the increased dead time from the increased number of veto strobes. In theory, the errors in the determination of the no veto rate would require further analysis of this problem. In practice, the pressure on telemetry space means that the use of veto allzoning is most appropriate since it is at worst acting as a grey filter in comparison to lateral, but in all probability is further improving the signal to noise for ISGRI events.

Table 3: ISGRI veto efficiency.

The assessment for PICsIT data is somewhat different due to the use of histograms which avoid issues of telemetry saturation, thus the noveto rate can be simply measured from telemetered science data. Therefore the rates are more securely defined for the PICsIT case.

However, the use of zoned veto signals, which act on only a fraction of the detector plane, modify the dead-time calculation somewhat, and it is simplest to determine the dead time, on a per-module basis, from the PICsIT detector module strobe counters. Again, calculating the background reduction on PICsIT data, assuming a PICsIT-VETO strobe length of 2 $\mu$s, yields:

Table 4: PICSIT veto efficiency

Again the conclusion is that the vetoing efficiency increases as more of the detector plane receives the veto strobe (i.e. as zoning is reduced), and that veto all is the optimum zoning configuration for PICsIT.

Table 5: PICsIT veto final settings.

5 Final setting

During the commissioning phase, the veto system exhibited nominal behaviour apart from two unexpected features: three times one VDM HV dropped to zero and occasional high count rates (>50 000 $\rm {c/s}$) were observed in single modules lasting from one to several hours.

These phenomena were difficult to explain due to their random occurence, and they could not be reproduced on the QM on the ground. One possible explanation was the presence of frozen water vapour or other volatiles due to CFRP degassing during the initial part of the flight. Another was HV instability induced by occasional very high energy particles. A deep investigation was performed to improve the long-term stability of the veto system. and it was decided to reduce the PMTs HV power supply to the minimum value, to increase the electronics pre-amplifier gain to 2 $\rm {keV/mV}$ to compensate for the loss of gain in the PMTs, and to set the electronic threshold to 40 $\rm {mV}$ (in Table 5). In this way the design goal of 80 $\rm {keV}$ energy threshold was maintained.

A series of VDM spectra were recorded in the final configuration, and are shown in Fig. 7. The fit parameters obtained for the 511 $\rm {keV}$ lines in those spectra are reported in Table 1.

	Figure 7: Spectra for final configuration.
Open with DEXTER

The veto system has performed stably since these settings were adopted, and the fact that there is no degradation in the measured energy resolution indicates that the veto detectors themselves are still performing well.

6 Conclusions

The effectiveness of the veto design has been confirmed during the Commissioning Phase by a background suppression of 50% for ISGRI and 40% for PICsIT, while inducing an acceptable dead time (<15% for ISGRI and 4% for PICsIT. This results is a good agreement between the recorded detector rates and early project assumptions.

The flexibility of the system was used to reach stable working conditions in the flight environment without compromise on instrument characteristics. To date, the system shows a good thermal and performance stability, with all modules working in nominal conditions throughout the orbit, including the eclipse sessions.

Acknowledgements

The authors would like to thank the Laben team for the veto engineering, S. Di Cosimo and U. Zannoni for their important support; A. Segreto and M. Gabriele for data collection facilities. The Italian part of IBIS is funded by the Italian Space Agency, ASI. The Norwegian part of IBIS is funded by the Research Council, NRC. The polish participation was founded by State Committee of Scientific Research of Poland. A. J. Bird is funded by PPARC grant GR/2002/00446.

References

 
• Bazzano A., Bird A. J., Laurent P., et al. 2003, Proc. PSD6, September 9th-13th 2002. Accepted for publication in NIM. In the text  
• IBIS Team, IBIS User Manual 5.1, vol. 1, EID-B. IN.IB.IAS.UM.010/01 In the text  
• IBIS Team, IBIS User Manual 5.1, vol. 2, Veto user manual. IN.IB.IAS.UM.010/01 In the text  
• IBIS Team, Report on IBIS FM VETO Calibrations, Dec. 2001, IN.IB.IAS.RP.026/2001 In the text  
• Poulsen J. M., Sarra P. F., Hajdas W., et al. 2000, SPIE, 4140, 293 In the text  
• Ubertini P., Di Cocco G., Lebrun F., et al. 1996, SPIE 5-7, 2806, 246 In the text