7.3.3 SAMPLING SPEED CALIBRATION
In order to convert the data of the 6 DSCs to a nanosecond time base, the
sampling frequencies of the individual DSCs have to be determined. Although VDS
is common to all, differences are occuring. The sampling speed calibration is
done by applying a pulse with a large amplitude to the TEST input of the
DSM100, which is AC-coupled to the inputs of all the DSCs by
CTest=0.3pF. The test pulse originates from a precision
clock/divider module C103 (PSI), which bases on a 100 MHz oscillator, with
a specified stability and precision of better than 10ppm. The 10 MHz
output triggers a LP103 module (PSI), the -10V output of which is supplied to
the TEST input of the DSM100. The signal with a period of 100 ns and a
pulse width of about 20 ns is differentiated at the test capacitance and
sampled by the 6 DSCs. From the difference
[Delta]b=b2-b1 of the leading edges
b1 and b2 (Fig. 7.8 bottom right) of the
resulting waveforms, the sampling frequency fs is
calculated.
The difference [Delta]b of the leading edges is determined for each
event and filled in the appropriate histogram (Fig. 7.9). The rms values are
indicated for each of the four distributions and are between 0.1% and 0.2% of
the corresponding mean values. As a result, the uncertainty, with which the
sampling speeds of the DSCs can be determined, is smaller than 0.2%. Using the
individual sampling frequencies, the smoothed data of each DSC are recalculated
on a two-nanosecond time base. At this point in the analysis, all the
parameters of interest can be extracted from the waveform data. The parameters
of each pulse, as fast integral (100 ns), the amplitude, rise-time and
fall-time (see Fig. 5.2 on page 38) as well as the leading edge of a
waveform as indicated in Fig. 7.8, is calculated.
Figure
7.8:
Single event display of the outputs of 4 DSCs of a sampling speed calibration
run. The difference [Delta]b (bottom right) between the leading edges b1 and b2
is calculated at -10 ADC counts (-10 mV).
Figure
7.9:
Difference [Delta]b of the leading edges of the 10 MHz test pulse of four
DSCs. The sampling-speed calibration run consists of 1000 events. The indicated
mean value multiplied by 10 corresponds to the sampling frequency in MHz. The
sampling frequencies vary between 739 and 770 MHz.
The sampling speed of the DSC is temperature dependent. The correlation was
measured by installing a temperature probe inside the DSM-module and performing
calibration runs at various temperatures between 20-30°C. The resulting
average temperature coefficient at 750 MHz is
(-2.7±0.1)°/oo/°C.
The assumption of a temperature stabilization of ±0.5° and a sampling
frequency of 750 MHz leads to a temperature depending timing uncertainty
of 0.13%, which is smaller than the uncertainty from the sampling speed
calibration.