Skip to Main Content
Two resistor network multiplexing circuits for a 12 × 4 array of SiPMs were constructed and tested. Both circuits encode the position and energy information from 48 SiPM pixels in only 4 analog channels. The two circuits differ in that one buffers each SiPM output with a non-inverting voltage-feedback operational amplifier before multiplexing, whereas the second one connects the output of the SiPMs directly to a charge division resistor network. The energy and timing resolution were measured with a 4 × 4 array of LYSO scintillator crystals with size matched to the SiPM pixel size. The measurement was done in 3 steps to cover all 12 × 4 SiPMs. Both circuits gave an energy resolution of 14%. The single sided timing resolution for the buffered output circuit was 2.86 ns, using a 350-650 keV energy window. In comparison, the timing for the circuit with direct connections between SiPMs and the resistor network was 3.54 ns, using the same energy window. Based on these results, the predicted coincidence timing resolutions are 4.0 ns and 5.0 ns, respectively. The coupling of the SiPM capacitance with the resistor network results in different signal shaping time constants for different SiPMs in the passive resistor network, causing a delay in trigger time for the inner SiPM signals. On the other hand, the circuit with buffer amplifiers does not suffer from this effect, and the pulse shape is more uniform across the SiPMs. We also demonstrate, using a 10 × 10 array of 1.5 mm LYSO crystals, that the inclusion of multiple SiPMs in both circuits reduces the detector's ability to resolve crystals in the flood histograms. The amount of noise increases with number of SiPMs in the multiplexing circuit.