The Time-of-Flight System on the Goddard Medium Energy Gamma-Ray Telescope

A scintillation counter time-of-flight system has been incorporated into the Goddard 50 cm by 50 cm spark chamber gamma-ray telescope. This system, utilizing constant fraction timing and particle position compensation, digitizes up to 10 ns time differences to six bit accuracy in less than 500 ns. Event selection decisions, discriminating against upward-moving particles, are made prior to spark chamber triggering. The performance of this system during a November 1978 balloon flight is discussed.


Introduction
Since any instrument for gamma-ray detection intended for use in space operates in an environment of a large number of upward-moving gamma rays, some technique mu g , be utilized to greatly reduce or eliminate tae detrimental effect of this upward flux on detection efficiency. With a picture device such as a spark chamber there is little danger of misidentifying the upward-moving particle as a downward-moving gamma ray because of the unambiguous nature of the pair-production interaction. However, any time spent by the instrument in recording and transmitting unwanted data will reduce the live-time to some extent. Spurious counts could then, significantly reduce the instrument sensitivity by contributing excessive deadtime.
:n previous instruments, an array of unidirectional Ceienkov counters was located at the bottom of the instrument as one element in the triggering telescope. The Cerenkov p unters, blackened on the upper face, provided a fair rejection of the upward flux. however, for the Medium Energy Telescope discussed here and the High Energy Gamma-Ray Telescope reported elsewhere in these transactiona l , a time-of-flight technique has been adopted. By measuring the particle time-of-flight between two scintillators and rejecting times corresponding to upward-moving ones, the sensitivity of the instrument to this type of spurious event may be drastically reduced.

Description of the Instrument
The Medium Energy Gamma-Ray Telescope utilizes digitized wire grid spark chambers of similar design to previous balloon instruments and the SAS-2 instrument.
The vertical spark chamber array consists of 20 wire grid modules, 50 cm by 50 cm inside dimension. Each module is constructed of glass-bonded-mica ceramic and wiredwith two orthogonal meta of 400 wires, one on each aide of the frame. The x-y spark position is recorded by ferrite cores located on shelves at the edges of the module. As can be seen in Figure 1, the upper 16 spark chamber modules are closely stacked. These modules are interleaved withh 25um tantalum plates for gamma-ray/ p air conversion. The four modules below the first scintillator plane are widely spaced to aid in identification of the separating electronpositron pair. These modules are separated by thin mylar light barriers only, to minimize any scattering or energy loss. The scintillators forming the timeof-flight telescopes are loc:.ted just above and below the lower Rpark chamber stack aq will be discussed in detail, Tile energy of the detected gamma ray is measure. by the energy system in the lower section of the iustl ment. Thu heart of tlhic: svetcm is a 50 cm by SO cm by 25 cm block of plastic scintillator which is viewed by eight photomultipliers. The summed signal is pulse height analyzed to six bit resolution as a coarse measure of the _-nergy of the produced pair. A Sri cm diameter plastic scintillator located below records penetrations of the energy scintillator to fla t; events where the charged particles did not deposit their total energy in the large scintillator. The upper portion of the instrument is surrounded by a 1.: In diameter dome of plastic e:cih,tillator.
This scintillator is used in anticofncidence v th the other scintillator elements to veto incident charged particle pene- A block diagram of the instrument timeof-flight electronics is shown in Figure 2. The fast signals from each photomultiplier anode are routed to a slow coincidence discriminator module and bridged to the constant fraction discriminators where the timing signal is derived. Discriminator signals r om each end of the scintillator are averaged b • a time compensation circuit, and a single timec :currence strobe isFenerated from each of the three strips. The time compensator signals act as Star t and Stop for a dual-ramp time digitizer. The digital time-offlight data are compared to a command entered threshold value for the up / down decision, and are also stored In an on-board memory. The time-of-flight data from each event meeting all the spark chamber trigger criteria are telemetered along with the "picture" data. The accumulated spectrum in the on-board memory is dumped via telemetry on a periodic basis independent of the spark chamber operation. IaL 71 O^lr10r x11

Constant Fraction Discriminating
The timing signals are derived from the photomultiplier signals by an integrated circuit constant fraction of pulse height discriminator. This implemention of a well known technique utilizes Motorola MC 1651 dual I.C. comparators as the fast switching elements. While the i.herent switching speed of this ECL device is not as great as for a charge storage diode or tunnel diode, in this application this is of minor significance; however, the stability and rugged nature of the device is of overriding importance.
In the circuit of Figure 3, the I.C. lb acts as a cross-over detector with the cross-over point adjustable by the variable resistor as shown. The input signal is attenuated to give a 0.2 fraction and applied to the non-inverting input. The unattenuated signal is delayed by a 3 na length of semi-rigid microcoax and applied to the other input. I.C. la is used as a leading edge trf g .ger to arm the AND g a te I.C. 2a. The other half of I.C. 2 is used as a unity gain amplifier to facilitate inspection of the cross-over setting. Figure 3. Schematic of the constant fraction discriminator.

Time Compensator
On this instrument the main axis dimension of the timing scintillators (50 cm) is un the same order as their separation (45 cm), so if some form of compensation were not used, the variation in the position of traversal of the scintillator by the particle would totally destroy the up/down timing resolution. By using the light signal from each end of these long scintillators it is possible to compensate for this variation. There are two techniques that are used for travel time compensation: the technique of linear addition of time-to-amplitude converter outputs 5 , and the technique of the time averager. 6 The simplest implementation for this application uses the latter technique.
The schematic shown in Figure 4 represents the method of time compensation used for this: instrument. The differential pairs Ql and Q4 receive the standardized ECL signals from the constant fraction discriminators connected to the photomultipliers on opposite ends of a scintillator strip. I.C. 3b is a gate connected to generate the ECL reference voltage of -  pulses into the 10 ns delay line. The combination of I.C. 2 and I.C. 3a form a logical OR with a threshold of -1.3v whose Inputs are connected to the delay line taps. At the time the signals from each end of the delay line overlap and add, exceeding the -1.3v reference, an output signal is generated. This time is essentially independent of the position where the particle penetrates the scintillator.

Time Digitizer
The signals from the time compensators, which represent an average time-of-occurrence in the upper and lower scintillation counter strips, are crans^titLed via twisted pair cable to the time digitizer shown In Figure 5. The differential signals originating from the three upperaid three lower scintillatirs are converted to single-ended by ECL line receivers 2nd logically OR'd to be applied to the Start and Stop flip-flops (I.L• la,b). Prior to the occurrence of an event the conversion capacitor C is maintained at the reference voltage V R by the differential amplifier-regulator made up of Q5, Q6, and Q7. On arrival of a Start pulse, the regulator is released and the pull-down current source Q8 begins to decrease the voltage across the capacitor. A Stop signal terminates the run-down h-turning off Q8 and the run-up begins from currerr source Q9. The ratio of run-down to runup current is the stretching factor of the circuit. The current sources are biased by zener diodes operated at near-zero temperature coefficient. The diode-connected transistors compensate the Vbe variation of the current source transistors and should be in good thermal contact for best results. The capacitor voltage is sensed by the circuits of I.C. 4, (used here as a comparator) and a time gate is generated. It is initiated by the Stop pulse and lasts until the voltage across the capacitor returns to VR. This time gate is applied to the delay line gated oscillator (I.C. 6) which generates a 175 Mliz train of pulses, digitizing the run-up time of the conversion capacitor. This pulse train is routed to a high speed ECL scaler in the digital section where this digitized time-of-flight is processed.
Supervisory logic functions, such as requiring valid Start before Stop, res.atting the system whenever the time ditference overranges, and holding when the digital data system is hu:.y, are included on the digitizer board and shown In the schematic.

ri i^ital Data System
This section wil' deal with the time-of-flight digital subsystem and ,tow it interacts with the overall experiment event handling ulv,itronics. The on-board event handling system consists of a number of subsystems which perform the various tasks of time-of-flight measurement, pulse height analysis, command decoding and coircidence determination. The time-of-flight sul.=vstem encodes the particle timeof-flight, generates an enable signal if a preset threshold :ondition is met, and accumulates tine time spectra in on-hoard memorv. The pulse height analyzer digitizes the signal amplitude frow tht' energy scintillator, generates ati enanle signal if the Ivilse height exceeds the preset threshold, and accumulates the pulse height spectra in on-board memory. The command decoding and coincidence electronics f orm the telescope coincidence logic, sucject to enabling si,;nals from the time-of-flight, pulse height analyzer, anticoincidence, and ground commands. The instrument may be commanded Into various sparK chamber trigger modes; which ma y logicall y include or !xclude the time-of-flight result, the pulse height value, the penetration counter flag, or the anticoincidence. The on-board storage condition, may be commanded independently. For instance, the spark chamber may be triggered on neutral events only, while a histogram of charged particle data is being accumulated.
the entire event handling system in under the control of an AMD2911 microprogram sequencer. The control program is contained in a tie by 32 bit bipolar PROM. The controller handles transfer of data and control signals between the various subsystems, the control of on-board histo;ram memory, and the  A simplified block dlagraut of the event handling system a+ It relates to the time-of-tllght is seen in Figure b. Me high fr,quencv hurst from the time digitizer gated oscillator is counted b y a Six bit ECL scaler. The translated time-ot-flight data is parallel-loaded into tale output 1'4 Sllitt register by the End-of-Conversion (Elk') pulse from the digitizer. The EOC pulse is also used to notifv the coincidence subsystem that a TOF i'vt• tlt (TOFF) has occurred. The six bits of time-of-flight data .to compared to the value held in the comm.tnd register. A value less than this preset threshold :,.; g as it (TOF0 enable to be sent to coincidence. The accumlt'ation of the time-ot-flight histogram is controlled by the control/avquencer. The occurrence of it (R1W) flag from coincidence causes the controller to increment the address determined by the value tit output shitt register. The histogram memory is organized as .1 b+ channel lsv 04K count array. The R1W cycle time is about 3usec. For events where a spark chamber trigger is generated the RIW sequence is held until the noise from th,` spark chamber has subside,l.

Laboratory 'rest
The constant ft'actlon discrimin:hors wore tcgted for function and time response versus input :amplitude characteristics prior to heinp, lntccratod lilt, , the instrument.
In order that tilt-Input pul gt• "haracter-tst1c be representative of the actual application, a photomliltiplier, seintillat,r, and Small 141., ,t-source wort, used as tit, , test -nitse yctlerator. A laboratory time-to-ampliludc convertor and multichannel analvzer wore used to measure tilt , performance.
Typienlly. Cie cir"lllt8 had loss than 50 I'S of tinewall, for an Input amplitude varied from -100 n1C to -1 volts, a rank representative of the instrument signals.
The timing char y "terI,tI -s of this Los trumcnt call he lief luonc. •d by electronic Itroadoning its well as the sire and 11.1Crment of tilt' gcintill.ltors.
In order to determine till relative c,,ntrihutit"IS to the timing uncertointy .1 mensurt,nk`nt was made of the inherent resolution ,,t tilt , Instrtimont electronic system.