A beam test for the hybrid prototype was performed before mass production of the detector modules started. The data acquisition system was based on equipment in the VME, CAMAC and NIM modular standards. The original software in the "LabVeiw" format was developed for the "Quadra 840AV" Macintosh PC to readout measurement devices, to preprocess and visualize the data and to store the information on the hard disk.
  As a first step Landau spectrum was obtained for relativistic electrons focused onto a single pad. Their analog signals provided by the readout chip were measured with analog-to-digital converter (ADC). The trigger pulse to start the ADC was discriminated from the same signal - this was so-called the "self-triggering mode" of the experiment. The low threshold voltage for the pulse discriminator insured non-biased trigger conditions. The most probable amplitude in this spectrum and the noise pedestal were estimated and the signal-to-noise ratio (SNR) for minimum ionizing particles was determined. It amounts to 13 for small pads and about 10 for large pads in rough agreement with expectation. An optimal depletion voltage for the silicon sensor and a gate voltage for the FOXFET biasing scheme were found from direct amplitude measurements in order to have the the maximum SNR.
  The minimum energy requirement in the electron spectrum introduces a gap between a signal peak and a pedestal. This allows a free threshold choice within some limits when all incoming particles can be registered. This is so-called "plateau" on the threshold scan curve which is a convolution of the Landau spectrum. The plateau width on this curve is a measure of the SNR-characteristics for the given pad while the number of counts on the plateau gives an estimation of a single detector efficiency which is very closed to 100%. The efficiency of a group of detectors, however, depends on their alignment and on their synchronization scheme. A simplified model of the silicon pad detector was assembled for the trigger efficiency studies. After fine tunings of the pulse synchronization circuitry and taking into account all possible track patterns it was tested that the total efficiency measured amounts to its mathematical expectation and preserves in a wide energy range of the incident electrons.
  A series of calibration measurements was held for 50 assembled pad modules prior to their mounting onto the BST. With 5 GeV electrons a threshold scan was made for each module and every individual pad. The SNR value was used to control the detector quality and to select efficient modules which plateau was larger than 20 mV. This allows reliable setting of the operational threshold voltage just above the noise.