The G0 experiment is located at Jefferson Lab, Newport News, Virginia, USA. Its goal is to learn more about the quark substructure of protons and neutrons by measuring the flavor singlet charge form factor of the proton G.
The experiment consists in scattering electrons by protons. The particle trajectories are bent according to their momentum with a magnetic field distributed in space on 8 regions called octants. They are detected by scintillators (Bicron BC408) that select 16 levels. The French part of the collaboration had to build half of the detection, i.e. 4 octants. Each of them is composed of 16 pairs of scintillators. The of IPN Orsay had the responsibility of the mechanical design of the detector and their supports. The detector assembly was carried out in collaboration with the SDI group of LPSC at Grenoble, France.

The design studies lasted one year.

A light structure was adopted to facilitate the shipment to the United States. The support structure has to ensure a good positionning of all the scintillators (distributed over about 2 m) despite thermal fluctuation of plus of minus 10 °C around the average value. The simulations carried out with a 3-D finite element software show a variation less than 0.63 mm.

The photomultiplier tubes (Photonis XP2282B) have to be separated from the scintillators because of the magnetic field that is too strong at the scintillator level. The scintillators are connected to the photomultiplier with light guides that can be as long as 1.8 m. They must be bent on a narrow space and still allow a good light transmission. The shapes and the mounting procedure were designed with a 3-D CAD. The drawings and fabrication methods were determined from this study and transmitted to the subcontracting company in charge of cutting and bending the light guides.

Two prototypes were tested at IPN. A large scintillator associated with medium size light guides was tested to validate the effects of the bending. A small scintillator with the longest light guide allowed to verify that the number of photoelectrons was sufficient, as well as an intrinsic time resolution less than the 750 ps RMS required. Measurements were performed with a Strontium/Yttrium source and cosmic rays. The number of photoelectrons was 45 and 30 and the intrinsic time resolutions 310 and 430 ps RMS respectively.

The support assembly was carried out at IPN Orsay. The scintillator positionning was adjusted and verified with the department's tridimensional control system.

The scintillators were cut at Grenoble. They were polished and wrapped at IPN Orsay. The light guides were partly wrapped but not glued on the scintillators. The final assembly took place on the JLAB site.
The first octant was delivered at JLAB on March 2001 and the last octant on August 2001.