The CLAS Cerenkov Detector

As of 28 January 1998, the last of the six CLAS Cerenkov detectors has been installed. Now, it's time for some documentation. This document came into being on 29 January 1998; as of this writing, it is still 29 January 1998. So, there isn't much here yet.

This introduction is in lieu of having that ubiquitous image of a construction worker saying "This page under construction".

Introduction

The Cerenkov detector was built by Rensselaer Polytechnic Institute (RPI), with significant technical assistance by Jefferson Laboratory. The calibration software is being written by John Price of RPI, and the reconstruction software is being written by Alexander Vlassov and Alexi Stavinsky of ITEP (Russia). Each of its 216 independent channels consists of a Philips 5" photomultipliertube (PMT) and three specially-constructed mirrors. The overriding principle of the detector is to collect a large fraction of the Cerenkov light while using the minimum amount of material, which can degrade the energy reconstruction of the calorimeters.

More will be in this section...

Construction

As part of CLAS, the Cerenkov detector is composed of six nominally identical ``sectors''. Each sector covers the angular region from approximately 8-45 degrees in theta, with the full CLAS coverage in phi. The mechanical construction of each sector consists of two sidewalls, which sit in the planes of the CLAS magnet coils, a "snout" piece at the vertex near the beamline, and a "45-degree wall" made of GreySeal, a composite material that is both strong and light.

There are three different types of mirrors in the Cerenkov detector: elliptical, hyperbolic, and cylindrical (in the order in which the Cerenkov light hits them). The elliptical and hyperbolic mirrors were constructed at RPI by a small army of undergraduate students under the supervision of an engineer. They were made by vacuum-sealing three layers of Kevlar around a foam core with epoxy, and then laminating a piece of aluminized Lexan onto the finished core. The Lexan was treated with magnesium fluoride to prevent oxidation of the aluminum.

Mirror Alignment

After building the sectors, the PMTs and mirrors must be installed. The mirrors must be aligned precisely, so that as much Cerenkov light as possible is focussed onto the PMTs. To do this, we used Alex Vlassov's simulation. As part of his work for the CLAS collaboration, he simulated the response of the detector to electrons at varying energy and angle, and (in the simulation) adjusted the positions of the mirrors to obtain the highest light collection efficiency. He then ran the program again, turning off the magnetic field, and setting the index of refraction so that the Cerenkov light is emitted precisely along the direction of the incident electron. We duplicated this in the laboratory by placing a laser at the target position, relative to the sector, and aiming it at different points along the mirrors. By comparing the reflections of the laser beam with those in the simulation, we can obtain the optimum mirror alignment.

After adjusting all of the mirrors, they are attached to a spine running down the center of the detector. This spine provides extra rigidity to the mirror array, and prevents droop in the mirrors over time.

Window attachment

After all the mirrors are in place, the windows are attached. We have recently changed the design of the windows; the new design, created by Yuri Sharabian, is discussed here.

The window material is a sandwich of 10 mil mylar surrounded on either side by 2 mil Tedlar. Tedlar is a trademark of the DuPont corporation. It is a black material that has extremely good light tightness, and is very strong. However, because the gas we are using in the Cerenkov detector is very heavy (approximately ten times the weight of air), it is not strong enough. The addition of the mylar helps a great deal. First, all the rough edges of the sector must be smoothed out so that the window will not be torn on them. Next, the window is glued on in a very conventional fashion. The curved side of the detector presents a challenge for this procedure, but the careful work on the part of the Jefferson Lab technical staff led by Tom Carstens was sufficient to attach the windows cleanly.

After both windows have been attached, the sector is tested for gas leaks. This is not a simple task, since the volume of the sector depends on the pressure differential between the inside and the outside of the sector. With a simple assumption, however, the problem becomes tractable. Begin with the ideal gas law:

pV=nRT,

where the variables are as follows:

p	Pressure within the sector
V	Volume of the sector
n	moles of gas within the sector
R	Ideal gas constant
T	Temperature within the sector

From this, we get dn/n = dp/p + dV/V - dT/T. To measure the pressure within the sector, we make two separate measurements: the first is the barometric pressure of the atmosphere (p_atm), and the second is the differential pressure between the inside and the outside of the detector (p_dif). This gives us dn/n = dp_atm/p + dp_dif/p + dV/V - dT/T. If we then make the assumption that the volume of the detector depends only on p_dif, V=V(p_dif), we get dV/V = (1/V) x (dV/dp_dif) x dp_dif
dn/n = dp_atm/p + dp_dif/p x (1 + (dV/dp_dif)(p/V)) - dT/T

dn/n = dp_atm/p + dp_dif/p x (1 + C) - dT/T,

where we assume that p and V are roughly constant. Note that for a rigid box, C=0, and for an infinitely flexible box, C=&inf;.

Nomenclature

There are at least four different numbering schemes for the PMTs in the Cerenkov detector. This is because the people who wrote the software for the readout did not communicate with the people who designed and built the detector. It would be very painful to resolve the different numbering schemes at this point, so we will instead document all of them here.

Construction scheme: Each of the sidewalls of the Cerenkov was constructed individually. The sidewalls themselves are identical; the two different sides are denoted "A" and "B" before any of the hardware is put on. Each sidewall has eighteen PMTs. Each PMT is labelled with its sector number, PMT number, and side (for example, CCS3 13B is sector 3, PMT 13B). There are two other schemes that are variations of this scheme, in which "A" and "B" become "L" and "R" (left and right) respectively, or "Lo Fi" and "Hi Fi" respectively. "Fi" is a misspelling of "phi", which denotes the azimuthal angle that the sidewall sits at.
Software scheme: In this scheme, the different phototubes are numbered so that as the number increases, the angle that the segment sits at increases monotonically. This means that each PMT is given a number from 1 to 36, and the sides alternate as the number increases.

Comparing the above schemes:

Construction scheme	Software scheme
CCS1 1A	CCS1 1
CCS1 1B	CCS1 2
...	...
CCS1 18A	CCS1 35
CCS1 18B	CCS1 36

Electronics

For a schematic of the Cerenkov electronics, click here. There are two inaccuracies in this document that I am aware of:

There are two high voltage crates; the second one is in C3-2.
There are two inputs to the UVa 132 module, corresponding to the "A" and "B" PMTs.

Work is underway to write a utility to help in diagnosing electronics problems by looking in a database to find exactly which modules are associated with which channels.

Calibration

The calibration of the Cerenkov detector primarily involves equalizing the gains of the phototubes. The main effect of this is that this action gives meaning to the hardware thresholds. This is especially important now that the Cerenkov is being used in the trigger.

To calibrate the PMT gains, we use the data acquisition system (DAQ) configured to read only the Cerenkov detector information. This speeds up DAQ immensely; while normal data acquisition is limited to approximately 400Hz, Cerenkov calibration DAQ runs at event rates as high as 3500Hz.

Last updated: 5 February 1998

Counter by Rapid Axcess