Data acquisition electronics

Most of the DAQ electronics described in the following are modules in the NIM format, placed in several NIM bins inside a standard 19⁰⁰rack.

The positive high voltage (HV) bias of 3500 V to 4000 V that is required to oper-ate the HPGe detector is provided by an Ortec 459 Bias Supply and is applied via the preamplifier to the outer contact of the germanium crystal [Gil08]. A temper-ature monitoring circuit inside the preamplifier assures that the bias can only be applied to the crystal if it is in cold state.

The HPGe crystal is read out by a Canberra Model 2002C Spectroscopy Preampli-fier, a charge sensitive resistive feedback preampliPreampli-fier, which is mounted on the cryostat’s detector arm next to the dewar. The input stage of this device is cooled by LN₂ for noise reduction [Can04]. The preamplifier mainly consist of a low noise field effect transistor (FET) fed back by a resistor parallel to the charge inte-grating capacitor, which ensures the discharge of the latter with an approximate time constant of 50 µs. The input of the preamplifier is connected to the inner contact (-) of the crystal and therefore collects the signal induced by the holes (the electrons drift to the outer contact (+)). This results in an output pulse of positive polarity with a steep rising edge of several 100 nanoseconds duration depending on the charge collection characteristics of the crystal. The main purpose of the preamplifier is to collect the charge induced by an interaction in the crystal and provide a high impedance input for the detector and a low impedance output for the main amplifier [Gil08]. Depending on the setting of an internal jumper, the

Veto

Figure 3.9.: DAQ system with conventional read out. Multiple connections are shown by bold arrows. The multiplicity logic unit consists actually of two identical devices operated in slightly different modes. The signals are combined afterwards by a simple OR unit, which was left out for simplicity.

output amplitude is 100 mV or 500 mV per MeV deposited energy [Can04]. The latter setting is used for standard measurements with the DLB.

The main amplifier is an Ortec Model 672 Spectroscopy Amplifier that amplifies and shapes the signal. The device has several possible settings, but is usually used for standard measurements of the DLB with a shaping time of 6 µs and a triangular shaped, unipolar output pulse. The standard amplification used is 9.1, which provides an input range of approximately 0 keV to 2750 keV based on the range of natural occurring γ lines (compare 2.2.2). Furthermore, it is operated in automatic mode for pole-zero (PZ) cancellation and baseline restorer (BLR) rate. The main amplifier powers the preamplifier via a 9-pin Sub D cable. In the opposite direction the output signal of the preamplifier is transmitted by a single ended (SE) connection to the main amplifier. In addition to the output signal in the range of 0 V to 10 V, the main amplifier provides two logic signals. The pile-up rejection (PUR) signal is provided in case a pile-pile-up of two preamplifier pulses is detected. Another signal is provided during shaping (BUSY), which allows a live-time correction at the pulse analysing system [Per01].

The PMTs of the veto detector are powered via a custom made 48-channel distributor by a Heinzinger HN 2500-025 power supply. The PMT signals are discriminated by different models of LeCroy discriminators with thresholds ad-justed per channel or module (depending on the model) according to the individ-ual amplification characteristics of the PMTs. The fast (negative) NIM logic out-put signal is combined by two LeCroy 380A Multiplicity Logic Unit, what allows for the detection of an adjustable number of coincident input signals. After the final combination of the logic signals by a Borer OR 320 module, the veto output is represented by a single signal. To apply a certain rejection time per veto event

(compare discussion in Section 3.2), the signal is extended from its approximate length of 50 ns by a CERN N2255 Timer. The fast NIM logic signal of a certain length is converted by a NIM to TTL converter to create a compatible signal for the following pulse analysing system. For details see [Ned09].

Up to now the output, PUR and BUSY signals of the main amplifier are trans-mitted to an Ortec TRUMP®-PCI-8K multi channel buffer (MCB) in form of a PCI card plugged into a standard PC operated under Microsoft Windows XP. In the range of 0 V to 10 V the card increases per detected signal one of up to 8192 bins (13 bit resolution) corresponding to the height of the incoming pulse unless a PUR or signal from the veto, applied to the Gate input of the card, was detected as well. The MCB card is only able to acquire spectra for a certain adjustable time period. Unfortunately, an usually list mode called acquisition mode is not avail-able with this hardware. In this DAQ mode each signal is analysed regarding its pulse height and stored together with a timestamp to an event list.

The MCB is read out by the Ortec Maestro®-32 MCA Emulation software, version 6.06. With this software several settings like lower and upper level discriminators of the card and the acquisition state regarding the Gate input (off/coincidence/anti-coincidence) can be set. Since the software allows the ex-ecution of batch jobs, this feature is used to overcome the weak point of the non existent list mode. With the DLB usually single spectra of 15 min or 1 h are au-tomatically acquired, saved and the measurement restarted so that it is possi-ble afterwards to check, for example, the count rate near the peaks of the radon daughter products especially at the beginning of a measurement period (due to flushing of the measurement chamber with nitrogen, radon is slowly expelled, but it can also be emanated from the sample itself) or monitor the decrease of counts in peaks of short lived nuclides produced during NA. The software also allows rudimentary calibration of the acquired data according to peak position and width, but is usually only used for data acquisition. The spectra of the single runs are stored to the hard disk of the DAQ PC and afterwards processed by the software described in Chapter 4.

According to the manual of the MCB [Per00] it should in principle be possible to acquire the correct live-time (without the in total applied rejection time, com-pare discussion in Section 3.2) by applying the Gate signal also to the BUSY input of the MCB card. As a draw back the BUSY signal provided by the main ampli-fier can not be used simultaneously and the MCB has to be switched to simple live-time correction mode, which can not handle losses caused by pulse pile-up.

Because of this, the in total applied rejection time is measured by counting the

∼50 ns lasting pulses from a standard pulser. Depending on the applied rejec-tion time per event, a pulsing frequency of 10 Hz to 20 Hz is used. One counter counts the total number of applied pulses, the other one is gated by the same veto signal also applied to the Gate input of the MCB and therefore counts only during this time. By dividing both numbers, the applied relative rejection time

can directly be calculated. This method does only determine the mean fraction of the time, since not every single veto signal is sampled and the result is only for a large number of counts equal to the real applied fraction. The determined relative rejection time is automatically subtracted by the software described in Section 4.1, unless the user does switch this off or no measured relative value is found in the corresponding data directory. This method does not account for the case where already a non negligible dead time occurs during the conversion of the signals by the HPGe detector (the total rejection time and the dead time of the MCB then overlap in parts). But this overlap does only occur in high count rate situations, for example during calibration runs. For these measurements the veto detector can be switched off (in the MCA software) anyway, since the background contribution is then negligible, so the effect described above is not of relevance for low background measurements.

To overcome the drawbacks of the used MCB and to enable an event by event data recording, the DAQ system will be replaced by an FADC based system in the near future. Therefore, the input ranges of a Struck SIS3300 12 bit 100 MHz 8 channel VME based FADC, kindly provided by the AMANDA collaboration to the COBRA experiment, were modified by the electronics workshop of the Fac-ulty of Physics to match different input ranges, which were up to now provided by the adjustable amplification of the main amplifier. It is intended to directly read out the preamplifier of the HPGe detector to get rid of the slight temperature dependence of the main amplifier gain and calculate optimal shaping parameters during the offline data analysis. The FADC is read out by a Concurrent Technolo-gies VX 511/063-23 VME Single Board Computer with an Ubuntu Linux operat-ing system. A small program that runs on the soperat-ingle board computer communi-cates with the VME bus and acts as a VME to TCP bridge. The Data-Acquisition and Control Environment (DAQCorE), developed by O. Schulz [Sch11] for the COBRA experiment using the SCALA programming language, communicates via TCP/IP with the VME bus server. DAQCorE can be executed on any plat-form, since it runs in a Java Virtual Machine (JVM). It stores the pulse shape data acquired by the FADC together with time information and corresponding settings in so called TTree structures of the file format belonging to the ROOT data anal-ysis framework. After the data acquisition the recorded pulses are processed by MAnTiCORE, an analysis tool kit based on the ROOT framework and also devel-oped by O. Schulz [Sch11]. With this piece of software it is possible to calculate the optimal pulse processing parameters and do sophisticated analyses.

In addition to the Windows based DAQ PC a workstation operated with Ubuntu Linux is available in the laboratory. It is used to manage the data storage of the spectra to the cell of the distributed network file system AFS operated by the working group. Furthermore, it allows the simple usage of the software de-scribed in Chapter 4 and will be used to read out the FADC based DAQ system.

The power supply of both DAQ PCs and the whole DAQ rack is backed up by

Energy / keV

500 1000 1500 2000 2500

Count rate / cts/(keV kg d)

10-1

1 10 102

103

104

105 Lab, no shielding

10mwe, 50mm lead castle DLB, no veto

DLB, top veto DLB, veto

Figure 3.10.: Background spectrum of the detector system, taken at different stages of completion.

an uninterruptible power supply (UPS) to protect the electronics and prevent the loss of the calibration in case of a short power interruption.