A binary readout system has been in operation in ZEUS for over one year [3]. Although the crossing frequency is lower than at LHC, key characteristics, for example threshold control and diagnostic capability, are directly applicable to ATLAS. At a threshold level of 0.78 fC (22% of the signal of a minimum ionizing particle) the occupancy does not exceed 0.1%. System operation is stable and predictable. Readouts using both the Tektronix and the AT&T ICs in conjunction with clock-driven pipelines operating at 40 MHz have been operated successfully in a beam tests at KEK and CERN, yielding results that are consistent with bench measurements [2].
Figure: Single channel threshold scan for minimum ionizing beam particles. The curve for 
m pitch is shifted upwards from the 
m data because of reduced charge sharing.
Figure: Pulse height distribution obtained by differentiating theshold scan of Fig. 3.1.
The variable threshold allows analog measurements with a binary system. To
measure gain, threshold and noise, pulses with a fixed charge are injected
through the test capacitor at the input and the hit rate is measured vs.
threshold level. The threshold setting at which the rate drops to 50% is
the pulse height and the width of the transition yields the noise (
). Threshold scans can also be
used to measure detector signals.
Figure 3.1 shows a threshold scan
taken on minimum ionizing beam particles traversing a 300 
m thick Si
detector. Differentiation yields the Landau distribution
(Figure 3.2 ), peaking at about
3.5 fC. Figure 3.3 shows the noise vs. channel number for an LBIC chip
connected to a 6 cm long strip detector with a 50 
m pitch.
Figure 3.4
shows the threshold distribution of a full-scale 6 cm 
12 cm detector module with 768 strips on a 75 
m pitch. Two of the 64
channel LBIC chips (channels 1 - 64 and 576 - 639) deviate substantially
from the mean and the last chip is inefficient, underscoring the need for
some preselection of chips prior to assembly.
Figure: Noise in fC versus channel number. For a most probable signal of 22000 el (3.5 fC) this corresponds to S/N=18.
Figure: Threshold distribution measured on a full scale detector module. The dropped threshold points are due to unbonded channels at the transition between the analog and digital ICs.
The remaining channels, however, show excellent threshold uniformity and illustrate the good threshold control inherent to bipolar transistor technology.
The following data were taken with a detector irradiated to a fluence of
about 5
10
cm
equivalent minimum ionizing particles,
i.e. well beyond type inversion. The irradiation was non-uniform, so
the effect of varying damage levels could be compared by scanning the beam
along the strips. In the highly damaged area the junction (and high-field
region) has shifted to the n-side, and charge yield is improved
relative to the p-side.
Figure: Position resolution versus track incidence angle on the n and p-side of a 
m pitch detector.
Figure 3.5 shows the position
resolution vs. track angle for a detector with a 50 
m strip pitch. The
resolution is best at about 9 degrees, due to enhanced charge sharing
between neighbor strips. The analysis uses crude interpolation, so the
optimum resolution is slightly better than the expected 50 
m/
. If only one strip responds, the track coordinate is set to the middle of
the strip. If two strips respond, the coordinate is set to the midpoint
between the two strips. Figure 3.6 shows the tracking efficiency vs.
occupancy at various track angles for a 50 
m pitch detector of 6 cm
length. At incident angles of 0 and 9 degrees the n-side occupancy is
for efficiencies approaching unity on the n-side,
whereas the p-side is markedly inferior. These data were taken with LBIC
chips; the lower noise of the CAFE chip should yield improved results.
Figure: Occupancy versus efficiency on the n-side (left) and p-side (right) for various angles of incidence on a 50 
m pitch detector.