Chapter 4: A MIP Timing Detector (MTD) for CMS at HL-LHC
4.3 Studies to optimize BTL design parameters
4.3.2 Time resolution measurements at the Fermilab Test Beam
Figure 4.15: The test beam setup at Fermilab. From left to right, the scintillator is used for trigger, the silicon tracker is defines the MIP impact position in the x-y plane, and the MCP-PMT is used to provide a reference time. The LYSO+SiPMs test setups, one for the 1-bar and the other for the 3-bar array, are positioned along the beamline. The beam direction is the z-axis.
Figure 4.16: Fermilab April 2019 test beam. (Left) The experimental setup at FTBF.
(Right) The box and mechanical structure capable of rotating the 3-bar assembly.
Figure 4.17: Sensors used at the Fermilab April 2019 test beam. (Top) Naked and wrapped individual crystal bars with the two FBK SiPMs (left) and single wrapped bar glued to SiPMs (right). (Bottom) Three-bar assembly in a crystal holder with the screws used to adjust the alignment of the crystals to the SiPMs (left); HPK SiPMs soldered onto a readout board (middle); single bar assembly (right).
The crystal bars of LYSO:Ce used in this test were manufactured by Crystal Photonics Inc. (CPI) in three different geometries of 3×t×57 mm3, where the thickness, t, varies between 2, 3 and 4 mm. Two different SiPM types were tested, the first type belongs to a set of S12572-015 SiPMs from Hamamatsu (HPK) with an active area of 3×3 mm2, while the second set was provided by Fondazione Bruno Kessler (FBK), with an active area of 5×5 mm2 based on the NUV-HD-ThinEpi technology. The HPK and FBK SiPMs provide gains of 1.8×105and 2.5×105, respectively, and the PDE weighted by the emission spectra of LYSO:Ce is similar for the two SiPMs and reaches about 36% for 6V over-voltage. The crystal bars were wrapped in Teflon (∼100 𝜇m thick) to improve light collection. Crystal bars coupled to HPK SiPMs were assembled in a system with a three-bar holder, with only one thickness t = 3mm, where the bars are placed parallel to each other and can be rotated with respect to the beam direction. Bars read out by FBK SiPMs were set up with a one bar holder, with the bar thickness varying as t = 2, 3 and 4 mm. The crystals and SiPMs can be seen in Fig.4.17.
Customized electronic boards were used to apply the bias and perform the readout of the SiPMs, and the full pulse shapes were digitized with a sampling frequency of about 5.12 GSample/s. In the 3-bar setup, each SiPM signal was amplified with a Gali74+ low noise amplifier, filtered with a 500 MHz low-pass filter and then split into two paths. One of the two signals was further amplified by about 44 dB
using two Gali52+ amplifiers in cascade to provide a saturated waveform with a steep rising edge, used for extracting the time of arrival using low threshold leading edge discrimination, while the other unsaturated signal was used to measure the deposited energy. For the single bar readout, each SiPM signal was amplified with Gali74+ amplifier and filtered by a 500 MHz low-pass filter. The signal was split and one output was read out directly to measure the signal amplitude while the second output was further amplified by a second stage Hamamatsu C5594 amplifier with a gain of 36 dB and used to measure the MIP arrival time. A CAEN V1742 digitizer [105] hosted in a VME crate was used for the readout of all the waveforms:
two readout channels for each SiPM under test and one for the MCP-PMT used as time reference. The digitizer was triggered by TTL-level signals originating from the trigger counter.
The time resolution of the setup can then be defined in a few different ways. We define the following quantities based on the the times of arrival measured at the two SiPMs at the ends (using the high gain channel), 𝑡
left and𝑡
right , and the time measured by the MCP-PMT,𝑡
MCP:
• Δ𝑡
bar=𝑡
average−𝑡
MCP= (𝑡
left+𝑡
right)/2−𝑡
MCP
• 𝑡
diff =𝑡
left−𝑡
right
• Δ𝑡
left=𝑡
left−𝑡
MCP
• Δ𝑡
right =𝑡
right−𝑡
MCP.
Since we use a fixed threshold at the leading edge (of the high gain channel) to measure the time of arrival, the𝑡
left (𝑡
right)depends on the amplitude of the signal.
This variation (∼ few hundreds of ps) in the time of arrival needs to be corrected to achieve the optimal time resolution, and the correction is derived by studying the dependence of𝑡
SiPM−𝑡
MCPon the amplitude of the SiPM signal, and is known as "amplitude-walk correction". Position based corrections are also derived by measuring the dependence of𝑡𝑙 𝑒 𝑓 𝑡+𝑡𝑟 𝑖𝑔 ℎ𝑡 on the impinging position of the MIP on the bar, to achieve a uniformity in the time resolution across the bar irrespective of the MIP impact position. A Gaussian fit is then performed to the distributions of Δ𝑡
bar,Δ𝑡
diff,Δ𝑡
leftandΔ𝑡
right, and the time resolution of the bar is then obtained as:
• 𝜎𝑡
average =
√︃
𝜎2
Δ𝑡bar−𝜎2
𝑡MCP
Figure 4.18: Time resolution measurements from the Fermilab April 2019 test beam. (Top) Time resolution of the left and right SiPMs, their average, and half of the time difference as a function of the MIP impact point for a 3×3×57 mm3 LYSO:Ce bar coupled to HPK SiPMs (left) and for a 3×4×57 mm3 LYSO:Ce bar coupled to FBK SiPMs (right). (Bottom) Global and local time resolution for a 3×3×57 mm3 LYSO:Ce bar coupled to HPK SiPMs (left) and for a 3×4×57 mm3 LYSO:Ce bar coupled to FBK SiPMs (right).
• 𝜎𝑡
average = 1
2
𝜎𝑡
diff = 1
2
√︃
𝜎2 𝑡left+𝜎2
𝑡right
• 𝜎𝑡
left =
√︃
𝜎2
Δ𝑡left −𝜎2 𝑡MCP
• 𝜎𝑡
right =√︃
𝜎2 Δ𝑡
right
−𝜎2 𝑡MCP .
The final results are shown in Figure4.18. A very spatially uniform time resolution can be achieved throughout the bar for both types of SiPM, as well for different crystal dimensions. In addition, a global and local resolution is computed using the beamspot information, as shown in Figure4.18(bottom). In all the measurements the target time resolution of 30 ps was achieved, demonstrating the feasibility of the BTL design.