Figure F1 — PIR detector system block (navigation map)

What the sensor actually measures

PIR does not measure absolute body temperature. It measures change: as a warm object moves across the segmented field of view, the infrared energy entering the sensor shifts from one zone to the next. This creates a bipolar (positive/negative) AC signature at the sensor output, typically concentrated at low frequencies where slow drift and fast vibration must be rejected by design.

Why Fresnel zoning creates a bipolar waveform

• Fresnel optics (or an aperture mask) splits the scene into multiple zones.
• Crossing zone boundaries converts spatial movement into alternating increases/decreases of incident IR energy.
• The result is a small bipolar pulse train rather than a DC level, which is why band-pass shaping is fundamental.

Why dual-element (differential) sensing matters

A dual-element pyroelectric sensor effectively behaves like a differential pickup: slow common changes (room warming, gradual sunlight heating) appear similarly on both elements and are largely canceled, while a moving target produces a time-shifted response between elements and survives the subtraction. This is the first layer of false-alarm defense, before any MCU logic exists.

Main disturbance classes (and what they look like)

• Slow thermal drift: baseline wander; energy piles up near very low frequency; often correlates with temperature slope.
• Sunlight / reflections: large slow steps or long pulses when a bright spot sweeps; repeats with time-of-day patterns.
• Airflow / HVAC: low-frequency “rolling” variations; correlates with vents, fans, or heating cycles.
• Vibration / shock: higher-frequency bursts; correlates with door slam, enclosure vibration, or loose mounting.

Evidence checklist (what to measure first)

• AFE output waveform: bipolar pulse shape vs drift/step signatures.
• Band energy proxy: compare low-frequency drift vs motion band (filter output RMS in window).
• Temperature and its rate-of-change: rising drift-trigger coincidence is a strong discriminator.
• Event statistics: clustering by time-of-day or HVAC cycle points to optical/environment causes.

Figure F2 — Signal signatures: motion vs drift vs vibration

Optics and mounting do not “boost” PIR; they encode space into time. Fresnel zoning converts motion into a bipolar pulse pattern, and the same encoding can accidentally convert sun reflections, HVAC thermal plumes, or moving shadows into motion-like pulses. A stable detection system therefore starts by constraining what enters the field of view.

Fresnel lens selection (FOV, zones, range)

• FOV (horizontal/vertical): wider FOV increases coverage but also increases the chance of capturing sunlight, reflections, or warm equipment surfaces.
• Zones / segmentation density: more zones produce clearer alternating pulses for true motion, but can also make slow environmental changes look “pulsed” when they move across zones.
• Range: longer range reduces angular speed at the sensor, pushing the motion signature toward lower frequencies where drift rejection becomes critical.
• Vertical curtain vs wide: curtain optics constrain the scene (doorway/corridor) and naturally reduce cross-traffic and window reflections; wide optics require stronger environmental control and drift handling.

Pet immunity (optics + mounting constraint)

• Pet immunity is primarily achieved by limiting low-height zones and preventing near-floor thermal changes from producing strong alternating pulses.
• Mount height and downward tilt shift zone intersections; the goal is to keep “small moving warm objects” from repeatedly crossing multiple zones.

Window / filter stack (8–14 μm pass, visible/sunlight blocking)

• IR passband targets room-temperature emissions; the window should avoid leaking visible/near-IR that can create large offsets or long pulses under sunlight.
• Sunlight robustness often depends on the window + internal baffle preventing a moving hot spot from sweeping across zones.
• Condensation / contamination changes transmission and creates apparent motion when airflows move temperature gradients across the window surface.

Evidence patterns that point to optics/window (fast discrimination)

• Time-of-day clustering (morning/evening spikes) → strong hint of sun angle/reflections.
• HVAC-cycle correlation (trigger bursts align with fan/heater state) → thermal plume in FOV.
• Single-location only (one install site fails, others pass) → mounting geometry/scene content, not AFE gain.

Mechanical & environment checklist (installation constraints)

• Avoid reflective glass in the center of FOV (windows, glossy floors, mirrors) where reflections move with sun or people.
• Avoid direct view of heaters, radiator pipes, warm appliances, or sunlit walls that drift quickly.
• Avoid HVAC outlets that blow warm/cool air across the FOV or directly onto the detector window.
• Keep stable mounting: vibration or loose brackets turn mechanical shocks into high-frequency bursts that can trip thresholds.

Figure F3 — FOV zoning & installation do’s/don’ts

The pyroelectric element behaves like a very high-impedance charge source. Useful motion energy is tiny and low-frequency, so input leakage, bias current, and parasitic capacitance can reduce sensitivity or create long settling that looks like motion. This chapter treats the sensor + PCB + AFE input as one coupled system.

Equivalent model (charge source + capacitance + leakage)

• Charge source (Q): motion across zones produces a small differential charge variation.
• Sensor capacitance (Cs): sets the relationship between charge and voltage at the node.
• Leakage path (Rleak): includes sensor leakage, PCB surface leakage, humidity films, and contamination residues.

What each non-ideal produces (symptom mapping)

• Higher leakage (lower Rleak) → reduced low-frequency gain, baseline wander, “washed-out” motion pulses.
• Higher input bias current → offset drift, longer recovery after transients, threshold instability.
• Extra input capacitance (layout/ESD/clamps) → slower response, altered band-pass corner behavior.

PCB contamination & humidity (common hidden cause)

• Flux residue + humidity can create a weak conductive film, collapsing the effective Rleak and increasing low-frequency noise.
• Fingerprints / dust near the high-impedance node are enough to shift drift and settling behavior.
• Conformal coating can stabilize leakage if applied with correct keep-outs and compatible materials.

Evidence chain (what to capture)

• AFE output: baseline drift slope, noise RMS, and motion pulse amplitude distribution.
• Trigger timing: correlation with humidity events, condensation, or cleaning/rework history.
• Repro check: same unit behaves differently in a dry vs humid environment → leakage dominates.

Protection & clamp side effects (noise + recovery time)

• ESD clamping provides a discharge path, but adds parasitic capacitance and potentially leakage at the sensor node.
• Input clamps can create a “recovery window” after an ESD/EFT event where the AFE output slowly returns, causing false triggers.
• Layout rule: keep the high-impedance node short and guarded; place protection so it does not load the sensor node directly unless required.

Figure F4 — Sensor equivalent model & AFE input constraints

A PIR front-end is not a “high gain amplifier”; it is a drift-controlled, band-limited charge-to-voltage system. The chosen topology decides whether low-frequency drift is rejected early, whether 1/f noise dominates the motion band, and whether startup or ESD recovery produces false triggers.

Common AFE topologies (when to use, what it costs)

• Charge amplifier: strong match to high-impedance charge sources; drift rejection can be defined by the feedback network. Cost: leakage sensitivity and recovery time are set by the same RC.
• “TIA-style” current/charge conversion: gain is explicit and easy to budget. Cost: input bias and board leakage can collapse low-frequency response if not controlled.
• AC-coupled multi-stage band-pass: removes slow drift early and emphasizes motion energy in a narrow band. Cost: an overly aggressive high-pass can miss slow motion across zones.

Key metrics that map to field symptoms

• Input noise density & 1/f corner: dominates the motion band when drift is large; causes “phantom motion” in quiet scenes.
• Input bias current / leakage: creates baseline wander and long settling; gets worse with humidity and contamination.
• CMRR / PSRR: determines sensitivity to supply ripple and digital activity; poor PSRR turns load steps into pulses.
• Startup / overload recovery: defines the false-trigger window after power-up, ESD, or large transients.

Band-pass filtering (drift vs vibration/EMI)

• High-pass (HPF) suppresses slow thermal drift and sunlight-induced offsets. Trade-off: too high a corner frequency reduces sensitivity to slow motion.
• Low-pass (LPF) suppresses vibration, relay shock, and coupled switching noise. Trade-off: too low a corner frequency blunts motion pulses and increases threshold uncertainty.
• Design goal: maximize motion-band energy while pushing drift and fast interference out of band.

Startup & recovery (the hidden false-alarm source)

• Power-up: input node charging and feedback settling can create a burst of pseudo-pulses if the decision stage is enabled too early.
• ESD/EFT events: protection loading can cause long-tail recovery; the tail may cross thresholds multiple times.
• Measurement: record the time-to-stable baseline and event counts in the first minutes after power-up or stress.

Figure F5 — AFE functional blocks (from sensor to decision)

Temperature changes affect PIR detection through two paths: (1) the sensor/AFE path (leakage, bias, 1/f noise, settling), and (2) the scene path (sun-heated surfaces, HVAC plumes, moving reflections). Compensation must separate these causes: electronic compensation cannot fix an optical scene problem.

AGC strategies (selection logic)

• Fixed gain + temperature threshold table: predictable and easy to validate; may underperform at extremes if noise rises.
• Segmented gain steps: keeps motion pulses in a stable range across temperature; requires hysteresis to avoid gain hunting.
• Adaptive threshold (noise-aware): tracks baseline noise and drift; must clamp min/max thresholds to avoid missing real motion.

Temperature inputs and what they control

• NTC near PIR/AFE or on-die temperature to index a compensation table.
• Outputs: threshold vs temperature, (optional) gain index vs temperature, (optional) HPF corner vs temperature.
• Add hysteresis on both temperature and gain index to prevent oscillation near boundaries.

Engineering workflow (measurable, traceable)

• Step 1 — Capture: temperature sweep; collect baseline slope, motion-band RMS, and event counts.
• Step 2 — Segment: define cold/nominal/hot regions; assign gain/threshold targets for each segment.
• Step 3 — Clamp: enforce min/max threshold and gain limits; prevent over-adaptation.
• Step 4 — Hysteresis: avoid boundary hunting; require persistence before switching.
• Step 5 — Log: record temperature, gain index, threshold value, and trigger rate for field traceability.
• Step 6 — Version: store table version + CRC; enable rollback and audit in production.

Evidence chain (what to log in field)

• T (temperature) and dT/dt (rate of change)
• gain_index (or gain_dB) and thr_value (threshold) with clamp limits
• band_rms (motion-band RMS) and baseline_slope (drift indicator)
• event_count and event_rate histogram per temperature segment

Figure F6 — Segmented threshold & gain vs temperature (with hysteresis)

A practical PIR detector uses a wake-on-event pipeline: an analog threshold (or window comparator) wakes the MCU, the MCU performs a short sampling window, makes a decision, then returns to sleep. This structure reduces average current without sacrificing detection reliability in real rooms.

Typical pipeline (event → verify → sleep)

• Interrupt wake: band-limited AFE output crosses a threshold and triggers an IRQ.
• Short sampling: MCU samples ADC (or counts comparator edges) during a bounded window.
• Decision: accept only motion-like signatures (pulse count/spacing/energy) and reject slow drift.
• Re-arm: apply a brief inhibit period to avoid bursty re-triggers during recovery or vibration.

Duty cycling knobs (what each one trades off)

• Ton (sampling window): longer improves capture of pulse shape; shorter saves energy but risks partial evidence.
• Toff (blanking/off time): longer saves energy; too long increases miss probability when motion falls into the gap.
• Twake (wake latency): must be below the motion-feature time scale; excessive latency truncates the leading pulse.
• Re-arm / inhibit: reduces false bursts; too aggressive can mask back-to-back passes in corridors.

Low-power reliability essentials

• IQ budget: track sleep current of AFE + MCU + wake circuitry; leakage dominates in humid/dirty environments.
• Clocking: use RTC/low-frequency clock for stable windows; avoid drift that changes effective Ton/Toff.
• Brownout behavior: choose BOD thresholds aligned with battery ESR and cold-start droop; prevent “half-awake” decisions.
• Wake sources: whitelist only necessary sources; noisy GPIO wake-ups can destroy battery life via spurious wake storms.

Field evidence pack (minimum logs)

• wake_count and wake_reason histogram (who woke the MCU)
• active_time_ms accumulator (true duty cycle)
• reset_count and brownout_count (power integrity)
• band_rms, baseline_slope, pulse_count summary per sampling window
• alarm_count vs temperature and battery voltage segments

Figure F7 — Low-power state machine

This section covers hardware-level interfaces for wired outputs, tamper inputs, indicators, and optional wireless modules. The boundary stays at pins, rails, protection, and wake lines—no protocol-stack deep dives.

Wired I/O blocks (alarm, tamper, indicators)

• Alarm output: open-collector/open-drain preferred for system voltage flexibility; add pull-up and edge control to reduce EMI.
• Relay / high-side driver (optional): include flyback path and current-step containment; log relay events to correlate with false triggers.
• Tamper switch: debounce + ESD protection; long leads require robust clamping and defined biasing.
• LED/buzzer: limit peak current; isolate load steps from AFE rail to prevent pseudo-pulses.

Wireless module interface (UART/SPI + wake + power gating)

• Data bus: UART or SPI for module-level transport; keep lines short and protected at the connector edge.
• Wake/IRQ lines: explicit WAKE_IN/WAKE_OUT to avoid keeping the MCU active during RF activity.
• Power gating: load switch or controlled LDO enables the RF rail only when needed; prevents idle leakage and wake storms.
• RF pulse coupling: protect against burst-current droop and ground bounce by rail separation and local decoupling.

Power tree (rails, inrush, and brownout)

• Battery/DC input → regulation → separated rails: AFE, MCU, RF/IO.
• Inrush and load steps: prevent rail dips from masquerading as motion pulses; validate with brownout counters and scope captures.
• Brownout thresholds: align to battery ESR and cold behavior; avoid unstable operation in the “almost alive” region.

Port-layer protection (no stack discussion)

• ESD/EFT/surge entry is handled at the port boundary: TVS, RC, common-mode choke, and isolation as needed.
• Evidence: log reset/brownout and port fault flags; measure recovery time after stress.

Figure F8 — Power tree and I/O protection blocks

False alarms become tractable when every event is forced into a repeatable attribution workflow. Each scenario below is written with the same structure: what happens, what to measure, how to distinguish, and the first fix that stays inside the PIR module boundary.

Sun sweep / reflection

Scene: sunlight sweeps across the window or reflects from floors/walls, slowly changing the IR field inside the FOV.

Evidence: elevated baseline_slope, strong very-low-frequency energy, long event duration, correlation with time-of-day and window direction.

Discriminator: if the waveform is dominated by slow drift and lacks repeated alternating pulses across zones, it is primarily scene drift rather than human motion.

First fix: adjust zoning/遮蔽 the sun direction, tune HPF corner slightly higher, require a short second-confirm window that expects alternating pulses, and tighten thresholds when dT/dt is high.

Headlight sweep / strong transient light

Scene: car headlights sweep through an entryway or corridor-facing sensor.

Evidence: short, high-amplitude spikes with fast recovery; strong repeatability for the same direction/time; weak multi-pulse “zone walk” signature.

Discriminator: a single sharp pulse without a short sequence of zone-crossing alternations indicates optical transient rather than motion.

First fix: reduce exposure to the driveway direction using optics/shrouds, apply pulse-count requirements (minimum N pulses), and use time-gated confirmation instead of globally raising gain/threshold.

Warm airflow / HVAC

Scene: warm or cold airflow moves across the sensor, causing thermal gradients and slow IR field shifts.

Evidence: strong correlation with fan cycles; elevated dT/dt; low-frequency dominant energy; repeated events with consistent period.

Discriminator: if triggers align with fan timing and show slow-drift signatures, the dominant cause is thermal flow rather than a person crossing zones.

First fix: prioritize installation changes (avoid vents and direct heat paths), increase drift rejection with HPF, and tighten thresholds dynamically when dT/dt rises.

Curtain / plant motion in the FOV

Scene: curtains or plants move due to wind or HVAC, modulating the IR pattern.

Evidence: periodic triggers matching the motion frequency; lower amplitude but long persistence; correlation with wind or airflow conditions.

Discriminator: strong periodicity with small-amplitude repeats suggests scene modulation instead of multi-zone traversal.

First fix: reduce zoning density toward that region, apply minimum motion-energy criteria, and add rate limiting for repeated periodic triggers to prevent alarm storms.

Pet immunity (low-height movers)

Scene: pets move within lower zones and near-floor paths, often close to the sensor.

Evidence: triggers cluster by low-height trajectories; shorter bursts; patterns that over-weight lower zones; frequent near-field events.

Discriminator: if events localize to the lower region and lack strong multi-zone alternations at human height, treat it as optics zoning rather than threshold tuning.

First fix: use pet-immune zoning (suppress lower zones), require stronger multi-zone consistency for alarms, and keep temperature compensation focused on drift—not on masking pets.

Insects / near-lens disturbances

Scene: insects crawl on or near the lens, creating localized IR modulation with abnormal proximity effects.

Evidence: unusually frequent triggers; near-field signatures; distorted waveforms that do not match zone-walk sequences; time-of-night clustering.

Discriminator: high repetition without zone-crossing alternation indicates near-field interference rather than room motion.

First fix: add mechanical barriers (mesh/shields), enforce “alternation” criteria in a second-confirm window, and rate-limit repetitive events with logging for maintenance actions.

Cross-class checks: power injection and mechanical disturbance

• Power injection: triggers correlate with RF TX bursts, relay switching, LED current steps, or brownout counters. First fix is rail separation, local decoupling, edge control, and wake-source filtering.
• Mechanical disturbance: vibration changes the FOV or mounting alignment. Evidence includes vibration correlation and repeatable patterns during door slams; first fix is mounting stiffness and isolation.

Figure F9 — False alarm attribution map

Validation must prove two independent properties: motion detection performance and false alarm probability. The plan below is structured as a repeatable SOP: each test specifies setup variables, pass criteria, and the minimum log fields needed for attribution.

SOP matrix (Test / Setup / Pass criteria / Logs)

Figure F10 — Test bench overview (target path + recorded items)

This playbook turns field issues into a repeatable workflow. Each symptom uses the same four steps: Symptom → Evidence → Isolate → Fix. The goal is to converge quickly without sacrificing detection margin.

Minimal tools (what is enough on-site)

• DMM: battery/rail voltage, static current, voltage drop under load.
• Simple scope or sampler: AFE output / comparator output / rail dip timing.
• Temperature readout: NTC or on-board temperature sensor.
• Isolation props: cardboard遮挡 for FOV masking, tape/shroud, small fan, small heat source.

Evidence checklist (minimum fields to log)

Symptom 1 — False alarms are frequent at night

Symptom 2 — Missed detection (should trigger but does not)

Symptom 3 — Random triggers in the first minutes after power-up

Symptom 4 — Wireless reporting drops (power-burst coupling)

Recommended additions (common in the field)

• Door slam / vibration triggers: isolate by tapping/mount flex test; fix with mounting stiffness, lens shroud, and mechanical isolation.
• Relay/LED step triggers: correlate with rail_dip and wake_reason; fix with edge control, rail split, and improved return paths.
• Humidity/rain increases false alarms: simulate leakage; fix with input cleanliness, coating strategy, and leakage-aware thresholding.

Figure F11 — Field triage decision tree (fast attribution)