How can I get stable SPAD readouts in practical conditions without cryogenics?

[Savannah_S]

I’ve been prototyping a small device that uses a single photon avalanche diode for extremely low-light sensing, but I’m hitting a wall with the signal-to-noise ratio in practical, non-lab conditions. I’m curious if anyone else has managed to get a stable readout without a full cryogenic setup.

[Emily92]

I fiddled with a little handheld setup last year. We cooled a Si SPAD a bit with a small TEC and kept the lid shut, then gated the readout to a narrow window synchronized to the pulse. Noise dropped a lot in the field, but afterpulsing and stray light kept chewing at the SNR if I ran acquisitions long enough. The practical win was keeping the electronics quiet and using tight timing; otherwise the readout would drift with temperature and supply noise.

[RyanHP]

I haven't pinned it down yet, and to be honest I keep hitting walls in the field. My guess is the bottleneck is photon flux and ambient leakage, not the electronics; when I push brighter bursts, I can read something stable, but under real dark conditions it collapses. I keep thinking maybe the problem isn't just noise but signal level.

[Ethan.M]

I swapped to a quieter bias supply and added a simple RC filter on the bias line, plus tighter optical shielding. In a few bench tests I saw the SNR jump a bit—threefold in a short run—but it only held for a minute before drifts crept back. It felt like a coping mechanism more than a solution.

[AaronZH]

Have you tried time correlated readout or time gating to separate dark counts from actual hits?