to previous bench test (MCM without detector).
I connected another MCM to the test setup in 218, and collected 1500 data packets.
This one had a kapton cable and a bad detector (inner bottom, 8.5cm cable,
broken Silicon, no bias voltage).
The data were collected in 250-event chunks, and appended to the same (ascii)
file. The events were collected at a rate of about 0.7Hz, over a period of
about 2.5 hours.
ps
- The first panel shows that the AMU cell number is evenly distributed.
- The second panel shows the raw ADC values vs channel number. The 6th
group of 32 channels is dead.
- The third panel shows channel 230 vs data packet number, that is, it is a
time history of this one channel. Note that this time there is a slow drift of
the data downward. The 'good data' cut this time is event>1000, as shown.
- Panel 4 shows, on a log scale, the raw ADC values for channel
230. The outlyers are caused by triggers that unfortunately coincide with the
periodic resets of the preamp/integrator.
- Panel 5 shows the main peak, and I defined a 'good data' cut between 380
and 410, and used it for all channels.
- Panel 6 shows what passes the 'goo data' cut. The fit shows a pedestal sigma of
0.85hannels
ps, page 2
The 8 panels above show ADC vs AMU channel, for each of the 8 groups of 32
channels. Now what we expected was a pronounced 4-cell periodicity,
like
in the real data, since we were guessing that this was caused by feedback
picked up from the MCM by the kapton cable. This is not what I see here.
Notes: The real data measurements linked above were taken on packet 2003,
which an inner middle. However, the noise levels for IM are the same as for
IB, see this
noise vs cable length plot.
Also, in the real MVD, htere are 60 MCMs radiating in synchrony, whereas we
have only one here. The inner enclosure may add to the noise too.
What is striking is the drift of the ADC on the scale of a few hours.
Note that the bare MCM settled in ~10-15 minutes, and was steady over 5 hours.
Toshi looked up the leakage current for this detector (before it broke), and
there was a 24-hour periodicity of ~15% in the current. Can the ADC drift
be due to the slow drift in leakage current?
Below are 2 examples of
leakage current vs time, with ambient temperature also recorded:
Last update 7 Nov 2001 - HvH
back