Access Official Technical PDF

Share Summary:LinkedInEmail

In a silicon photomultiplier (SiPM) array, adding pixels can expand an imaging system’s sensing area, but it also makes signal extraction more challenging. Directly combining fast outputs increases the capacitive load and brings more sensor noise into the shared channel. Signal-driven multiplexing (SDM) addresses this trade-off by using Schottky diode pairs to route an active pixel’s fast signal to a common readout while isolating other pixels. For engineers designing photodetector arrays, the approach offers a way to simplify readout without simply tying all fast outputs together.

Why direct SiPM output summing can be a problem

SiPMs provide a fast, capacitively coupled output in addition to their standard output. The fast output produces a short pulse proportional in amplitude to the photocurrent signal at the standard output, making it useful when the optical event is pulsed and timing matters. In an array, however, summing these fast outputs directly connects the capacitances of multiple pixels, along with circuit parasitics, to the readout.

As the connected capacitance grows, the fast signal can have a slower rise time, lower pulse amplitude and longer pulse width. These changes can impair timing measurements. Noise is another concern: combining channels can add dark noise from multiple pixels, raising the noise level at the shared output and reducing the signal-to-noise ratio (SNR). Smaller signals may consequently be harder to distinguish.

These effects make multiplexing more than a wiring problem. The readout topology and printed circuit board (PCB) layout influence the signal seen by downstream electronics, so array size, timing requirements and parasitic capacitance need to be considered together.

How signal-driven multiplexing works

SDM places a Schottky diode network between each SiPM fast output and a common readout node. When a pixel responds to incoming photons, its fast-output pulse changes the state of its associated diode pair: one diode conducts and transfers the pulse to the common node, while the other turns off. The signal is then coupled from the common node to the output.

The pulse at the common node also reverse-biases the readout path associated with other pixels. That action helps isolate their fast outputs during the active pixel’s readout, suppressing contributions from inactive channels. In effect, the signal itself selects the path to the shared output.

This arrangement reduces the capacitance presented to the common readout compared with directly summing all SiPM fast-output capacitances. The Schottky network still contributes capacitance, but the design goal is to keep that contribution lower than the load created by direct summing. The method is intended to preserve individual-pixel pulse characteristics and limit noise and timing degradation as pixels share a channel.

The role of Schottky bias

The diodes can operate without a separate bias, but applying a Schottky bias can establish a standing current through the network. In the tested configuration, this made the diodes respond faster and more sensitively to a SiPM breakdown, which can benefit rise time and timing measurements, particularly when the signal is weak.

Bias selection involves a trade-off. Increasing the bias changes the effective resistance in the diode network and can reduce fast-pulse amplitude. At the same time, faster response can improve timing. The useful operating point therefore depends on the circuit and array rather than following a single setting for every design.

In evaluation boards using 3 mm B-Series SiPMs, the best measured Schottky bias differed by array size: −12 V for 16 pixels, −20 V for 32 pixels and −30 V for 64 pixels. A −20 V setting also produced consistent results on the 64-pixel board. These are results for the evaluated configuration, not universal design values; bias and diode current should be assessed in the intended circuit.

What the timing evaluation shows

The evaluation compared arrays containing 16, 32 and 64 pixels using coincidence resolving time (CRT) measurements. CRT describes the timing window used to determine whether detected events are coincident; lower values are important in applications such as time-of-flight positron emission tomography (PET).

The best reported CRT was 304 ps for the 16-pixel board, 371 ps for the 32-pixel board and 473 ps for the 64-pixel board. The results indicate that the tested SDM design maintained sub-400 ps timing for arrays up to 32 pixels. Timing became less favorable as pixel count increased, with the evaluation attributing much of the deterioration to greater parasitics associated with increased board area. The results establish a design reference for the tested setup, not a guarantee for other layouts, sensors or applications.

PCB design considerations for SDM

At fast signal edge rates, interconnect and component placement can affect the outcome as much as the readout concept. The evaluation placed each Schottky diode pair directly beneath its associated SiPM to keep routing short. It also used a symmetrical layout and minimized parasitic elements on fast signal traces.

For a practical implementation, design reviews and validation should account for:

  • Short fast-signal paths: Place diode pairs close to their corresponding sensors and use high-speed PCB layout practices.
  • Array symmetry: Keep routing and geometry consistent where possible; the evaluation found layout and parasitics important to CRT.
  • Schottky bias decoupling: Decouple the bias line and use the series resistance to limit diode current and isolate the output from the bias supply.
  • Added circuitry: Amplifiers or simultaneous standard-output readout may be useful in a custom design, but additional components and routing can introduce parasitics that affect timing.
  • Component validation: Alternative Schottky diodes or resistor values should be evaluated rather than assumed to reproduce the reference circuit’s behavior.

Optical coupling and alignment also matter when validating a detector assembly through CRT measurements. Repeatable test conditions help distinguish circuit effects from changes in optical alignment.

Practical implications for array designers

Signal-driven multiplexing provides a structured alternative to direct summing when a SiPM array needs a shared fast-output channel. Its value depends on maintaining low parasitic capacitance, choosing an appropriate Schottky bias and validating timing across pixel locations and array sizes. The evaluation supports sub-400 ps timing for the tested 16- and 32-pixel designs, while also showing that scaling to more pixels can worsen timing.

For system integrators and technical decision-makers, the central lesson is to treat the diode network, PCB layout and bias conditions as one readout system. Prototype testing is essential: optimize the bias for the specific array, verify performance across the board, and assess the effects of any added amplification or output connections before committing to a final design.

Access Official Technical PDF

About The Author