Cosmic Ray Air Shower Array
Leif describes a bit about what he does for the High Energy Group in the LSU Physics Department.

Figure 1a : Early testing and calibrating of phototubes: stack configuration.

I am often asked what I do at work at LSU. I find it hard to answer so this part of WCSL's Department of Physics has been created to address these concerns.

I'll begin with Figure 1a. This is a picture of the old cramped room before the Nicholson Hall annex was built. In the foreground to the right you see a PC with KmaxNT, the data acquisition (DAQ) software package, running. To the left of the monitor is a CAMAC minicrate with several hardware modules inside. Behind the grey minicrate is a small NIM bin (appears light blue because of hardware modules in it). Behind the monitor is a blue high voltage (HV) power supply distribution box. The last thing of interest is the stack of scintillators on the table to the left with phototubes pointing out to the right (with small copper base plates where the HV and signal cables are plugged in).

Figure 1b : Final arrangement of counters: array configuration.

In Figure 1b is the air shower array configuration. But more on that later on.

Now, what does all this do? It seems best to start with the phenomenon of cosmic rays.

Cosmetic Ray? What in the Hell Is That Then?

A cosmic ray is a piece of an atom (not a ray, really). There are 2 kinds of particles hitting the Earth's atmosphere: galactic cosmic rays − particles from outside the solar system solar cosmic rays − obviously, particles from the sun Figure 2 : A possible air shower. These particles are bits of atoms, such as H, He, C, O, Fe, Li, Be, and B. To reach the Earth, galactic cosmic rays must go through the solar wind as well as Earth's magnetic field. (More cosmic rays reach the poles than the equator.) The solar wind, though, is less energetic than galactic particles, and so don't penetrate the Earth's magnetic field as much. (A big question is where do really high energy galactic cosmic rays come from. Pulsars? Supernovae?) So these particles hit atoms in the atmosphere. About 85% are protons (H nucleus), 12% are alpha particles (He nucleus), and 3% are electrons and nuclei of heavier atoms. When these primary incident particles interact with the upper atmosphere, secondary particles are created: protons; neutrons; positive, negative, and neutral pions (mean lifetime of 26ns). Muons (mean lifetime of 2.2μs) and neutrinos are the decay products of charged pions; and electrons, positrons, and gamma rays are the decay products of neutral pions. (Muons decay further into electrons and neutrinos.) Muons are the largest component of charged cosmic ray particles (because they are so massive − 207× mass of the electron − and have a relatively long lifetime). Most muons originate in the upper atmosphere and lose ~2 GeV by the time they reach sea level. The mean energy for muons at sea level is ~4GeV. Figure 2 shows an example air shower starting with a proton incident on the atmosphere. It hits some atom in the air and splits into a neutron and some pions. The pions decay into muons and so forth. Further, these particles can interact with other atoms in the air and produce more particles.

Muon in My Soup

It is mostly muons that come through down from the sky and through the ceiling. I remember reading that there are 20 muons passing through your body at any one time. That's just what they do. Can't be helped unless you're under a kilometer of rock. To detect these guys, man has created scintillators. I used left-over organic plastic scintillators from some previous experiments. The way these work is the plastic is doped with some kind of atom that gets excited when a muon hits it. As it deexcites, an ultraviolet photon is produced. The trick in producing these scintillators is to have the material transparent to the frequency of photon produced. The photon bounces around and hopefully heads in the direction of the phototube (photomultipler tube, or PMT). Figure 3 : Muon exciting an atom in the scintillator. The atom deexcites, producing UV radiation. Now, it is said that the very best detectors have only a 10% photon collection efficiency. Whatever that means. So, the muon loses about 2MeV/cm in our scintillator. I guess 10% of that energy goes into the photons that make it to the PMT at the end of the light guide (as Figure 3 shows). The light guide is non-scintillating plastic with an index of refraction close to that of the scintillator.

Photons To Electrons

The photocathode in the PMT must be thick enough to absorb photons but thin enough to prevent absorption of the photoelectrons (produced by the photon striking the photocathode). Photocathodes tend to be Cs-Sb or K-Cs-Sb alloys. The ones I used are 2cm × 2cm and are said to be ~10-30% efficient, although I have no way of really knowing. Figure 4 : An example of a phototube, showing the dynodes. About 4-5 electrons are kicked out on each dynode per incident electron. In the end, 105 - 107 electrons per incident photon are created, making a pulse that is >5 mV and ~60 ns long. This pulse gets sent along cable to NIM or CAMAC modules for further use. PMTs have to be powered since the dynodes need increasing voltages to accelerate the electrons. We have been using 2 types of tubes: those that produce many electrons per photon and those that produce less. This is to get a better idea of the number of particles for an air shower. The difference between the two is that one is higher voltage than the other. Our counters (as they are called) are scintillators with a PMT on each end, so the pair are seeing light from the same particle(s). One side is the low gain (lower voltage and so lower signal for a particle, thus able to see more particles per event before the tube saturates). The other is high gain (higher voltage and so higher signal for a particle: able to see less particles for an event before saturation). The ratio we have set right now for high versus low gain is 10:1. Arbitrarily. The tubes are powered depending on how they perform. The low gains are powered to produce 1 ADC count per particle. (I'll discuss ADC counts and calibration later on.) The high gain are powered to produce 10 counts per particle. The range of voltages used then turns out to be anywhere from 700 to 1100 V.

Determining Events With "Triggers"

Particles are always streaming down from the sky and not all of them are interesting. We are looking for air showers: those secondary cosmic rays which are produced by a high energy primary particle. There will be more particles per square meter for this type of air shower than the regular old background. These air showers tend to come down in circular cones with the axis more or less being the trajectory of the primary particle. There are 2 classes of counter used in these experiments: data counters and trigger counters. The data counters are used for data collection. The trigger counters are used to determine when an air shower occurs. You set up trigger counters in some configuration and apply some logic to the signals they produce. If the signals from the triggers meet your criteria for an air shower, you read the data from the data counters. If not, everything is ignored. Figure 5 : Stack configuration, side view. Pink=trigger. Blue=data. Figure 6 : Array configuration, top view. Pink=trigger. Blue=data. We've used two basic configurations. One is the stack (Figure 5): place all data counters on a table like pancakes. Put one trigger counter on the top, one in the middle, and one on the bottom. This configuration is ideal for calibrating the counters as they have the same particles going through them. Triggering in this configuration follows this logic: if all 3 triggers see a particle, take data. The second configuration is the array (Figure 6). The logic is the same: if all 3 triggers see something, read data. But this configuration is used for detecting larger air showers. The triggers are spaced out and each has to see some number of particles at the same time. Really, only 2 triggers are needed in either configuration. The array is now being run with only the corner triggers. Three is just a better number for reducing the number of random background noise events.

Triggers Signal Paths

Now to begin introducing the CAMAC and NIM modules. Let's follow what happens to the trigger signals and just how an event is determined to be valid. Figure 7 shows some trigger signal cables going into something called a discriminator and the output of this discriminator going into another discriminator. We set the threshold on the first discriminator to only allow signals greater than some voltage. The greater the threshold voltage, the more energetic particles have to be for them to be considered valid.

What the second discriminator does is set the number of triggers that must be hit at a time, the coincidence. This works because the first discriminator's SUM output has a voltage of the number of input channels over threshold times -50mV. Setting the second discriminator's threshold to -30mV means only 1 trigger must be good (1-fold coincidence). Setting it to -70mV means 2 triggers must be good (2-fold). And so on.

Figure 7 : Signal path for the trigger counters to produce a GATE for the ADC.

The output that we really want from the second discriminator is NIM for the GATE input on the ADC, to tell the ADC whether or not the data counters' signals should be integrated. However, the discriminator only outputs in ECL, so we have to take that ECL and put it through a level translator to convert it into a NIM signal.

Another problem is the ADC GATE should be longer than the signal produced by the second discriminator. (The level translator can't change the signal duration.) We send the level translator's NIM output through a Gate Generator. This resulting NIM signal with certain duration is then used for the ADC's GATE.

Data Signal Paths

Because the trigger signals have to go through so much hardware and the data counters get particle hits at the same time as the trigger counters, the data signals have to be delayed in some way so they can arrive at the ADC within the GATE window to be integrated. This is easy enough: just delay the data signals. Figure 8 : Signal path for the trigger counters to produce a GATE for the ADC. Figure 9 : Data signals get integrated during the ADC's GATE window. This is done with longer cable. We have some delay boxes which are just modules with lots of wire. Each data counter is getting the equivalent of 40 meters of cable before it reaches the ADC. Figure 8 shows all that goes on with the data signal cable. When the data signals arrive in the ADC input channels, the GATE has already been set and the ADC proceeds to integrate the charges (Figure 9). If a channel has no signal, only background noise gets integrated. This is known as the pedestal for that ADC channel. A normal pedestal is about 22 ADC counts, or 5.5pC.

That's about it for the hardware and what goes on with the counter signals. On the next page, I get into the software that is used to read out data and set the thresholds.

next page −>

Copyright 2002, Worldwide Center for the Study of Leif