The baseline shift comes from some instable power supply inside the evaluation board which cannot be controlled to the mV level. In a real measurement, you usually get an additional baseline shift due to some environmental electromagnetic interferences, such as a 50 Hz signal. People fix this shifting baseline by always aquiring a small portion (10-20 samples) of the baseline before any signal from a particle detector. The signal is then corrected event-by-event by subtracting the baseline from each waveform. By doing that, you fix not only the 50 Hz noise, but also the shifting baseline you mention.
Hello! My name is Dmitry. I am from SiPM Lab is NRNU MEPhI (Russia, Moscow). We bought DRS4 evaluation board V5 with firmware 21305. We use 5.0.4 build 21911 2015-11-23 software version (and before that we used 5.0.3 build 21508, 2014-10-15) with Windows 7 32bit.
We observe some strange behaviour. When we save waveforms (in xml or binary data) we see that some of them have the baseline shifted of about -5 mV.
The first picture (pic1) is 1000 waveforms which were glued in one. It is clearly see that baseline quite often has the shift.
The same effect can be seen without saving (writting): rarely when we use normal or auto trigger mode (pic3), and always in single trigger mode (pic2).
The images are attached.
Do you have any idea how it can be fixed?