Abstract. A new project, JULIA , is developing new technologies for next stages DUMAND{like underwater experiments. These experiments use large area ...
Signal Processing With JULIA Christopher Henrik V. Wiebusch III.Physikalisches Institut, Rheinisch{Westfalisch Technische Hochschule Aachen, Hyskensweg, 5100 Aachen, Germany
Abstract
A new project, JULIA , is developing new technologies for next stages DUMAND{like underwater experiments. These experiments use large area lightsensors for the detection of faint C erenkov{light, produced by neutrino induced myons. During a cruise in Feburary 1991 a small testdetector was deployed in the Atlantic ocean near the Canary Islands. The technical concept of this test{experimet is discussed here. The most important aspects were tests of a new type light{sensor ("smart" photomultiplier), and an analog transmission of data to ship via single mode optical bers. First results show, that it is possible to increase experimental accuracy and reliability, and also to decrease the price of detector components.
1 Introduction | Why JULIA ? Major goals for next{stage under{water{detectors are on the one hand bigger detector areas to increase the luminosity, on the other hand smaller distances between optical sensors to include the detection of low energy neutrinos. In order to achieve this, the number of necessary optical sensors has to increase. This seems reachable, if cheaper and more simple technologies are used. For the moment the most important aim of the JULIA project is the development of new technologies for signal processing and ocean{technologies, in order to achieve a better accuracy as well as lower costs for future detectors. JULIA means: Joint Underwater Laboratory and Institute for Astroparticlephysics. The JULIA experiment is supposed to be a next{stage underwater C erenkov{detector, but opposite to current stages of DUMAND or BAIKAL the main emphasis is on medium and low energy neutrinos. The pilosophy can be stated in a small sentence: As simple as possible. Simple technology provides a high reliability as well as a low price. As we will see, it is possible, to increase accuracy as well. From a technological point of view the momentary status of JULIA is more a laboratory, than a new experiment. This talk provides information on the technical concept of a testcruise carried out in febuary 1991 in the Atlantic ocean near the Canary{islands with the german research vessel"Sonne" 1, 2].
2 A technical overview
Figure 1: Schematic overview of the test{detector The test{detector consists of three optical sensors. These were assembeled into one vertical string with spacings of two meters. Central part of each sensor is a large area ( 35cm) photomultiplier. Together with a glass{pressure housing and electronics these photomultiplier are usually called optical modules (OMs). The AC electric power was supplied with two Hi {ampli ers (100W ) via 8km Coax{ cable, which also carried the whole detector. The signals from the OMs were directly transmitted from the read{out electronics inside the OM's to the ship via three single mode optical bers (without digitization). These signals could be processed on board with conventional NIM and CAMAC electronics. The data were read out via ATARI Mega ST computers and written to disk 3]. Due to the limited size of the optical cable of 1km, larger dephths were not reachable. An overview on the testdetector can be obtained from gure 1.
3 Optical modules All OMs in dierent experiments have a glass{pressurehousing and a large area photomultiplier1 in common. Due to a small power budget and the physical require1
There are ideas for OMs consisting of some small phototubes, but none is built so far
ments the electronic is almost similar, too. The electronic parts can be divided into slow and fast devices. Whereas the slow devices ful ll control{functions, the fast electronics process the incoming data, without digitizing. In the JULIA fast circuit (see below) the analog pulses from a "smart" Philips{PMT are charge{integrated by a readout{unit called DMQT. Philips PMT
analog ;!
pulse
Readout DMQT
ECL ;!
pulse
Laser Diode
optical ;!
ber
The ECL{pulse coming out of the DMQT is feed into a circuit which drives a laser{diode. From this diode the signal is transmitted2 via a single mode optical ber to ship. Usually the central part of an optical module's slow electronics is a remote control unit (RCU) wich is essentially a microcomputer. The RCU enables the comunication to a module via a modem3. During the JULIA testcruise the OMs were only passive, what means that there was no computer inside and no comunication possible. By changing the frequency of the supplied power between 8 and 12kHz Ac (with an oszillator before the HiFi{Ampli er) it was possible to regulate the high voltage of the PMT and thus the gain of the PMT and its pulseheight. This concept was only practicable for a short cruise. Permanently detectors require a RCU to know, " what is going on deep in the ocean".
4 The "smart" Phototube The ("smart") 15' Photomultiplier XP2600 by Philips 4] consists of the combination of an electro-optical preampli er with a conventional small phototube (see gure 2). Photoelectrons are accelerated with high voltage (25kV ) to a scintillator placed in the center of the glass{bulb. This scintillator is read out by a small conventional 11{stage phototube. This principle is sketched in gure 3. The result of the high accelleration voltage is a high gain in the rst stage. As a typical value, 30 photoelectrons (PE) in the small phototube are converted for each PE from the cathode. Due to the good statistics in the rst stage, the tube provides an excelent time and energy{resolution on the low photoelectron{level. The energy{resolution on the 1PE{ level is 50% FWHM. A good seperation between one, two or more PE is possible and can be obtained from gure 4. This gure shows a typical distribution of the integrated charge. The tube is illuminated with a LED of constant intensity. A t with equidistant, poisson distributed gaussiansp ts the distribution well. The time{jitter lies below 5ns for 1PE and decreases as 1= n with n, the number of PE. Due to a collection eciency of 100% and the high acceleration voltage, the tube behaves uniform over the whole cathode. The spherical cathode{geometry is not the best shape to optimize transit{time{dierences. The transit{time dierence between 0 and 90 is about 4ns for single point ilumination. 2 3
Without digitisation of the time{information. Signals are superimposed on the powerline.
Figure 2: Photo of the "smart" tube It is proven4, that this value can be reduced to less than 1ns. As a consequence of the decay time of the scintillator ( 45ns) this PMT has long pulses ( 150ns). It should be possible to reduce the pulselength as well as the time{jitter with the choice of a faster scintillator 5].
5 "Smart" readout | DMQT Requirements for a readout circuit tting to the "smart" Photomultiplier are: 1. It should be self{triggering because there is no external trigger in the ocean). 2. In respect of the good PMT{energy resolution it should work lineary to re ect this advantage. Due to the scintillator decay the pulses are not smooth, so linearity can be achieved only by integrating the pulse's charge. This integration should be fast to avoid a long dead{time. 3. This circuit must have a low time{jitter. 4. The outputs should be norm{pulses to enable easy further processing. The readout circuit, DMQT, measures the integrated charge with good linearity and time accuracy 6]. The integration of the pulse{charge takes place parrallel to its collection. As result the time of conversion is proportional to the integrated charge. Contrary to conventional charge{integrating{circuits with a xed integration{window the DMQT 4
The russian tube: Quasar (see this proceedings) and calculations 5].
Kathode 35cm |
e;
;!
scintillator
25kV optical preampli er {z
;!
conventionell fast photomultiplier
}
Figure 3: Schematic diagram of the "smart" tube
Figure 4: Distribution of integrated charge and a t with gaussians
has a low deadtime for small signals5 of about 130ns for one PE. Each additional PE increases this number by 80ns. The width of the ECL{outpulse gives the integrated charge and the leading edge the time{information of the phototubepulse. This time{information is obtained by a low threshhold ( 10mV ). A coincidence of a second higher threshhold ( 100mV ) with the 5ns delayed low threshhold provides a safe trigger and suppresses noise of the small photomultiplier.
6 Optical Data{transmission The major feature of the JULIA experiment is the analog transmission of DMQT signals to ship. Each module has its own ber, so trigger and any other data{processing electronic can be build up with conventional electronic, on board without restrictions on power{ consumption and size.
a) 1m ber-line b) 4000m ber-line Figure 5: Timing accuracy of the optical transmission line. The timing{accuracy for the transmission of a short signal (50ns) can be seen in gure 6. For a 1m optical ber no dispersion can be seen within the measuring accuracy of 100ps (a). Even for 4km only a minimal dispersion is measured ( 100ps) (b). If the jitter of the DMQT is taken into account, one can see in table 6, that the transmission{line has nearly no contribution to the accuracy of the time{signal. Since the damping of optical bers is dominated by the damping in connectors and splices, it should be possible to transfer data over 40km ber with nearly the same accuracy. This might be an important advantage for future underwater experiments as well as any other experiment, where a transmission of data from a detection zone to a laboratory 5
Keep in mind that most ocean noise is about 1PE
short NIM{signal DMQT 1 PE DMQT 5 PE 0m 0:21ns 0:42ns 1m < 0:1ns 0:27ns 0:45ns 4000m 0:1ns 0:31ns 0:45ns Table 1: Time accuracy for the transmission of dierent signals is desirable. Compared to DUMAND II , where the timing accuracy is limited by the digitisation in the ocean to 2ns, this method is both cheaper and more accurate.
7 Signal processing on board The incomming data were processed on board in a conventional way (see gure 6).
Figure 6: Signal processing on board The optical receivers for each module are realised as Nim{modules. Dierent kinds of coincidence conditions allowed the reduction of data. Another kind of trigger, an amplitudetrigger, gives the opportunity to set sharp threshholds in energy (number of photoelectrons) 7, 5]. Since the "smart" tube has a very good energy{resolution and the length of the DMQT{signals is proportional to the number of PE, only those signals are accepted which are longer than a certain PE{value. Table 2 shows a table of rejection as an example for the eciency of an amplitude trigger. Depending on the setting of the amplitude cut, it can be easily calculated, how many percent of the gaussian{distributed6 1PE, 2PE, 3PE . . . signals are cut out. Due to the good energy{resolution it is easily 6
Due to the high gain in the rst stage of the "smart" tube, each PE-distribution is gaussian.
recognized, that with an amplitude trigger nearly all ocean noise (1PE) can be cut, without loosing signals of higher PE. After processing with a certain trigger condition, the following values were measured with CAMAC: 1. Energy- and time-information of each module 2. Single and coincidence countrates (frequently readout). The CAMAC{crate itself was read out by an Atari Mega ST Computer with a special CAMAC Interface 8]. The maximum readout{rate was 100Hz. cut 1.2 PE 1.3 PE 1.4 PE 1.5 PE 1.6 PE
1 PE 80,4 % 90,0 % 95,6 % 98,4 % 99,5 %
2 PE 0,8 % 1,7 % 3,5 % 6,5 % 11,3 %
3 PE 0,0 % 0,0 % 0,0 % 0,0 % 0,0 %
Table 2: Percentage rejection of equidistant gaussian PE{contributions depending on the cut{threshhold ( E=E = 55% FWHM)
8 Conclusion and outlook The technical concept during a testcruise as part of a new project, JULIA , was described. The conclusion are: accuracy improvements are possible, a reduction of costs in nearly all elements of next stage detectors, especially in case of data transmission and the C erenkov{ lightsensors can be achieved Some future investigations should be carried out. There are two feasibilities. 1. Since the ship "Sonne" will be equiped with a new optical ber cable, tests in dephts down to 5km can be done. 2. A german research platform in the Northern Sea could serve as a prototype for a permanent ocean laboratory. The dephth is only 40m and the water is not at al clear as ocean-water, but this platform with its good infrastructure can be used for many real{condition tests under nearly laboratory conditions. Future investigation topis are: 1. Multiplexing of signals from many OMs to one optical ber. 2. Long term in situ tests, and calibration of OMs. 3. Background studies, especially on K 40 and bioluminiscence 4. Tests of detector{geometries, and triggers, with the emphasis on the seperation of low energy neutrinos from the background. 5. With a small detector (about 20 OMs) investigations on Muon{, Gamma{, and Neutrino{physics are possible. 6. By placing air{shower{detectors on the sides of the plattform and a C erenkov{detector at the sea{bottom, Extensive{air{shower{physics is possible.
References 1] DUMAND Collaboration. DUMAND II | Proposal. HDC-2-88, Hawaii DUMAND Center , August 1988. 2] P.C.Bosetti. Ergebnisse der JULIA Durchfuhrbarkeitsstudie. Internal report, April 1991. 3] Thomas Mikolajski. Signalwandlung und U bertragung im JULIA Experiment. Internal report. Institut fur Hochenergiephysik, Berlin{Zeuthen, Mai 1991. 4] van Aller et al. A "smart" 35cm Diameter Photomultiplier. Helvetia Physica Acta, 59:1119 ., 1986. 5] Christopher Wiebusch. Zum Nachweis schwacher Lichtquellen im Ozean mit Hilfe eines neuartigen gro" achigen Photomultipliers. PITHA 91/20 masters thesis, RWTH Aachen, Oktober 1991. 6] F.Bei"el und V.Commichau. A fast charge to time converter V04. Internal report HD04, III.Phys.Inst., Sommerfeldstr., 5100 Aachen, February 1991. 7] Christoph Ley. Berechnungen uber den Nachweis hochenergetischer Neutrinos mit dem DUMAND Detektor. Masters thesis at the Rheinisch{Westfalische Technische Hochschule Aachen, Mai 1990. 8] F.Bei"el, C.Camps, V.Commichau. Atari databox CAMAC-Coupler V01B Internal report HD04, III.Phys.Inst., Sommerfeldstr., 5100 Aachen, February 1991.
