For this investigation we used cosmics. TPM1 was placed with entrance window facing up. One scintillating tile was placed on top, one on the bottom, and coincidence of tile signals was used as trigger. We applied angular cuts of ± 20 degrees on reconstructed tracks to eliminate angular dependence on cluster widths and residuals.
As can be seen from the two first plots, TPM1 has some hot pads and some bad FEE cards. Plot 3, on which the lower panel is simply plot 2 with a narrower range in time bins, shows that there is some noise with frequency around 1 MHz. This noise shows up as clusters with low max ADC. Most of the noise can be eliminated in off-line analysis by increasing BrTPCClusterFinder::fMaxADCCutoff from the default 10 to 20.
However the noise gives a good contribution to the bulk of data from TPM1, and should therefore be eliminated on-line. Investigation is going on trying to find the source of this noise. Pedestal subtraction bin-by-bin is a possibility if the noise can not be sufficiently eliminated otherwise.
Except from this noise, TPM1 performs nicely. The gain is high, maybe too high for particles with high dE/dx, but this can be tuned. Resolution is on the order of 300-400 um.
TPM2 was tested using a collimated Sr source in front of the entrance window. Two scintillating tiles were placed behind the exit window, and coincidence in the tiles was used as trigger. Two tiles were used to get a better signal-to-noise ratio. However, only about 5% of the events were good source events.
Since we were dealing with MeV electrons from the source, they were subject to a lot of multiple scattering. Therefore tracking was done only in the first 3 active pad rows. However, as many of the electrons penetrated the entire chamber, we also required hits in all pad rows.
The reason for testing TPM2 with a source was that the drift velocity monitors gave contradicting results upstream and downstream, as mentioned by Ramiro. We placed the source at different measured levels, took data and projected the reconstructed tracks back to the position of the source. The first plot indicates the precision in our measurements. It shows 4 peaks, from 4 runs. The number of events in each run accounts for the different amplitudes of these peaks.
The residual plots are bad, mainly due to the nature of the source as well as the fact that we applied no angular cuts.
The test setup was the same as with TPM1.
T2 has some hot pads and some bad FEE cards. However, as can be seen from plot 2 and 3, which is just plot 2 with a narrower range in time bins, we now see signals from tracks in the in the lower regions of the rear part of T2. There does not seem to be any difference in performance between the two parts of T2. However, as plots 4 and 5 show, the gain is not that high, especially in the lower regions and for the lower anode voltages we tested. We should be careful and chose the correct anode voltage.
The residuals seem to be a little bit higher than in TPM1, on the order of 400-500 um.
......things look promising. We were able to do nice tracking over all active pad rows in TPM1 and T2, with gating grid pulsers on. The residuals show some long tails, which in part may be attributed to noise clusters. Reducing this noise and/or doing bin-by-bin pedestal subtraction will hopefully increase resolution.
The drift velocity issue from TPM2 data and drift velocity monitors is being investigated.