Last week I was working together with “my” summer students Jeanette and Silvia at the DESY test beam. We looked into the electronic noise of the “old” ZEUS strip telescope.
This is a relative old telescope build something like 10 years ago to characterize the ZEUS MVD silicon strip modules (MVD stands for Micro Vertex Detector and was an upgrade silicon strip detector of the ZEUS tracking system). Since then many different groups used the telescope as a tool to define the tracks of the test beam particles at the DESY test beam. But as you can imagine, the system is getting old after 10 years of operation. And recently the performance is not as it used to be. Mainly the electronic noise increased quite a bit. Electronic noise is a random signal characteristic of all electronic circuits and a constant offset underlying the real signal. Therefore the signal to noise ratio is a crucial performance parameter of our detectors. If the noise increases too much, the signal disappears in this carpet of noise and we have a hard time to track the particles. There are of course books on how to design and set up the readout system to reduce the noise to a minimum, but somehow noise is most of the time tricky. This I experienced many times before.
When Jeanette, Silvia and I were investigating the noise of the telescope last week we checked if the grounding was done according to the book (no loops, connections from on ground to all component etc.). Everything seemed to be ok, but somehow we still had a high noise level. So we started to change the system, just to see what happens. Whenever we expected the system to improve, it got worse and vise versa. In the end we found a configuration with really good noise behavior, but we absolutely did not understand why it improved.
So this was a really nice demonstration for the summer students, that an exact science like physics can also be something not at all exact, but rather like voodoo…