run0034 stopped at 14:11

Leakage currents are increasing and once again we are not fully depleting the pad detectors at 150V.
Noticed that the strip vs energy histogram began with a well-defined band across all strips and the n-side for a given pad, but by the time we stopped event energies were spread up to a max value across all 
strips and n-side. Also saw *incredibly* long rise times and a rollover of the pre-amp pulse rather than a peak and decay.
  -> This implies loss of bias to the detector.

We increased the bias voltage to 300V on Bus 0, Mod 1, Ch 2 (6th SiPad on the inside pocket) and saw the leakage current rise in tandem with the voltage (up to 15uA) such that with a 20 MOhm bias resistor, 
the induced voltage drop is almost equal to the applied voltage.

In short, up 300V we cannot apply high enough bias to overcome the equivalent increase in leakage current.

Explanation?
This could be an effect of ions on eccentric orbits hitting something and showering the detector with electrons, leading to surface charging and the increased leakage currents.
Either we are receiving huge numbers of ions in the detector but not seeing them, or something else like above is going on.

Surface charging could/would dissipate over time and potentially bring the detectors back into a useful range. Radiation damage likely would not.

15:14 Started run #35: inside pocket at 20 mm.
Trigger mask changed. Gated only in ch8 in MSI-8 #2 (one side of photodiode)
No events hitting the detector yet.

15:37: run stopped.

No sign of any events.

After some Sergey-magic the outside pocket moves again. We slowly moved the outside detector in, we are clipping the beam at -20mm. Moved the detector to -40mm and take some data as of 17:48 (run0036.lmd)

The leakage currents of the SiPad detector jump back and forth 2 microamps (btw 25 and 4.5 microamps), the DSSD seems stable, the crystal too

https://elog.gsi.de/esr/E0203_E0018/207

At 10:50am today, we connected a remote Picoscope to the first 4 n-sides, one end of both resistive chains, and both crystal channels to directly measure the output of the preamps without the ADC clipping at 2V.

In the parked position, we see a ~1 Hz rate on the detector with small signals in the crystal and very small signals (potentially just electronics) on the n-sides. 

At 17:15, we moved the detector close to the beam (-20 mm from the beam). With each injection, the preamps output massive signals, roughly 10 V each for a 1.8 V max input. Attached is a characteristic screenshot, with more in the ZIP file.

The flash appears to be correlated with the injection timing. The signal seems to reach a cap at 10 V, which I believe from talking to Roman is the HV bias reaching its leakage current cap. After that, the RC discharge comes back into range. Roman suggested that roughly, this RC discharge is representative of the 
signal height. The 1/e constant is around 35 us, so a saturation time of 560 us implies an signal peak of e^16 higher, ie 8.9e7 V or 1.8e7 GeV or roughly 1e7 particles. There may be some other saturation effects that mean we do not get quite this many particles, especially given only 10^5 particles are stored, but 
either way, our detectors were seeing at least 10^5 particles ever injection.

Thus, we can clearly conclude that the observed leakage currents come from radiation damage and that our detectors are fried. This is definitely a hard learnt lesson that we cannot have our detectors in for the isochronous mode...