M-unit array

The M-unit experiment

contributed by David Crawford

During 1958 The Falkiner Nuclear Research Department headed by Professor C.B.A. McCusker was offered ,most of the equipment from the Harwell experiment (Cranshaw, Galbreith and Porter(1957)). This equipment arrived early in 1957. It consisted of over one hundred detectors (in ammunition boxes) and six racks of electronics. At that time the Sydney air shower group was investigating the structure and time variations of cosmic ray air showers. It was decide to set up a closely packed array ph M-units (Martelli units; named after Giuseppe Enrico Gilberto Martelli (1923-1994)). The results were to be output on five channel punched paper tape. This paper tape was fed to the computer Silliac for analysis. Associated with the array were four penetrating shower sets with the small Geiger counter arrays. These had three trays of Geiger counters set between lead blocks. Later on, another experiment using 64 plastic scintillator counters and two cloud chambers was constructed on the edge of the M-unit array. It was run with and without lead absorbers to measure the electron structure of the air showers.

Each M-unit consisted of three Geiger counts each with an area of 16.1 square cm and were self quenched. They were used to trigger the presence of an air shower by requiring a triple coincidence in any of the 92 units. This meant that about 50% of the triggers were air showers. The output from each geiger counter in the M-unit was taken through a coaxial cable to ca cathode follower and then to a coincidence counter. The output for a triple coincidence was sent to a blocking i]oscillator and then through a balanced twin wire feeder to the central recording apparatus. The use of a blocking oscillator enables a pulse of 30 to 40 V to be sent over 30 metres through cheap twin wire flex and still have a rise time of about 1 micro-second. In addition to the coincidence output there was another output that provided a pulse if any one of the three geiger counters fired. Each M-unit had its own power supply providing +250 V, -150 V, and -900 V to -1400 V. The input to each power supply was 250V 500Hz provided by a motor alternator set that had a regulated field current to provide a constant voltage output. The higher frequency allowed the power supplies to have half wave rectification and cheap paper filtering capacitors.

The pulses from the M-units were sent to the first grid of a 2D21 thyratron (a gas filled valve that kept the anode current flowing after the grid pulse has disappeared). The second grid was tied to a master pulse line so that only those thyratrons that had a coincidence pulse remained fired after to pulse were finished. The recorded the complete results of one air-shower from all the M-units.

An air shower event could by triggered not only by the M-units but by the cloud chambers, the PSS, the 64 scintillator array or any other equipment that might be attached. The penetrating shower sets (PSS) was designed and run by Peter Poole and Ron Wand and consisted fours sets of geiger counter trays between lead absorbers. Each tray consisted of 48 geiger counters which had an area of 115 cm. The recording apparatus was designed by Chris Wallace and Michael Rathgeber and recorded the time, date, atmospheric pressure, and all the geiger counter data on two Creed Model 25 five-channel paper tape at (33 1/3 characters per second). The recording of a typical event took about 15 seconds. The paper tape was specially designed to be compatible with the computer SILLIAC.

The M-unit array started running on 8th May 1960 and by 16th May 1964 a total of 68,755 acceptable M-unit events had been recorded in a running time of 25,282 hours which is 71% of the elapsed time. During this time an average of 81.7 of the 92 M-units were operational. The four sets of GM trays started operating early in 1961.

Data processing

the experiment produced about 800 feet of paper tape per day which was processed on Silliac. Initially the programming and data processing was done by Michael Rathgeber, Peter Poole and David Crawford. Later on the earlier part of the processing was taken over by Derek Nelson who also improved some of the programs. Tony Bray took over the shower size analysis programs written by Peter Poole.

The initial part of the data processing was to transfer all the data to a magnetic tape and to perform all the maintenance checks needed to keep the experiment operational. When a tape was full (about every three to four weeks) it was analysed to obtained better statistics of the performance of the M-units. If a M-unit was faulty its response was deleted from one day prior to the discovery to one day after it was fixed. The final step of this initial processing was to remove all bad observations and compress the data so that about three years data could be kept on two magnetic tapes.