123 Main Street, New York, NY 10001

Submit Your BOM

Hot Water Recirculation Controller Design & Debug Guide

Hot Water Recirculation Controller Design & Debug Guide

← Back to: Smart Home & Appliances

Hot water recirculation is a local control problem: a pump, a few sensors, and an evidence-based schedule/demand strategy to cut wait time without wasting heat, adding noise, or creating false starts. A robust design proves decisions with two signals first (Treturn trend and pump current), then hardens power/EMC and fault logic so comfort improves without reboots or high daily energy.

H2-1. Definition & System Boundary — What “Hot Water Recirculation” Means (and what it doesn’t)

Hot water recirculation is a closed-loop control approach that reduces “time-to-hot” at the tap by circulating water until a temperature evidence threshold is met. The engineering cost is a three-way trade: heat loss (loop stays warm), pump energy, and noise/lifetime (speed and start/stop stress). This page focuses only on the pump + sensors + controller and excludes water-heater internals and whole-home energy platforms.

In scope (controller-level, measurable)

  • Pump control: BLDC/ECM drive, soft start, speed limiting, quiet bands.
  • Evidence chain: temperature sensing + optional flow/proxy signals to decide “hot arrived”.
  • Local logic: schedule, demand trigger, occupancy input (local), hysteresis, anti-short-cycle.
  • Device ruggedness: power rail stability and motor EMI evidence points (scope-first debug).

Out of scope (avoid cross-page overlap)

  • Water heater / tankless / boiler internal control and power topology.
  • HEMS, cloud dashboards, Matter platform stack and backend orchestration.
  • Plumbing installation tutorials, building-code certification walkthroughs.

Figure F1 The boundary diagram below separates “controller-in-scope” blocks from out-of-scope system blocks to keep design decisions and debug evidence focused.

Figure F1 — Hot Water Recirculation System Boundary Block diagram highlighting pump, sensing, and controller in scope; water heater, HEMS, and cloud out of scope. IN SCOPE: Pump + Sensors + Controller Hot Loop (Supply/Return) BLDC / ECM Pump Soft start Speed limit Quiet bands Temperature Evidence NTC / RTD → AFE → ADC Flow / Proxy Evidence Hall / Ultrasonic / dP or Pump current Controller (ULP MCU + RTC) Schedule / Demand / Local occupancy input Hysteresis + anti-short-cycle + fault actions Power Tree Motor Driver Logs OUT OF SCOPE Water Heater Tankless / Boiler internals HEMS / Platform Aggregation & orchestration Cloud / Backend Dashboards & remote policy Plumbing How-To Installation tutorials & codes Boundary
Figure F1. System boundary for hot water recirculation controller design. Blocks on the right are intentionally excluded to prevent cross-page scope creep.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F1 (System Boundary), 2026.

H2-2. Success Metrics & Constraints — Comfort, Energy, Noise, Reliability

A recirculation controller is “good” only when it can be verified with repeatable evidence. The practical definition of success is a balanced outcome: faster time-to-hot with bounded energy, acceptable noise, and predictable reliability under always-on constraints.

Primary metrics (what to measure and what counts as evidence)

Metric How to measure Evidence signal Typical range (use bands)
Time-to-hot (s) Start from trigger → until temperature evidence crosses threshold Return temp curveValve/trigger timestamp Application-dependent; define a target band per loop size
Loop stability Hold threshold with hysteresis and minimal cycling Temp rippleDuty cycleCycle count Stable within a bounded ripple band (avoid hunting)
Energy (Wh/day) Integrate pump power over day; compare modes Motor currentRun-time counter Mode-dependent; keep within an explicitly budgeted band
Noise (dBA / banded) Measure at fixed distance; capture worst-speed region Speed vs noiseCurrent ripple Define acceptable band per install environment
Reliability Track starts/day, fault retries, lockout rate Event countersFault logs Bound start/stop frequency; minimize false trips

Tip: if only two measurements are available, prioritize return temperature and pump current. Together they separate “heat arrival” from “pump doing work” and quickly expose false triggers.

Constraints that shape real designs (always-on + harsh coupling)

  • 24/7 standby: low quiescent power and deterministic wake-up behavior (no random boot modes).
  • Wet/humid environments: connector integrity and sensor drift must be expected, not treated as rare events.
  • Large ΔT / thermal inertia: temperature evidence reacts slower than user expectation; control must avoid “hunting.”
  • Low-flow uncertainty: direct flow sensing may be noisy at low rates; proxy evidence needs sanity checks.
  • Motor EMI coupling: switching edges can induce rail dips and false sensing; debug must start with scope evidence.

Design trade-offs (with actionable tuning knobs)

  • More frequent circulation vs higher heat loss: tune threshold + hysteresis + minimum-off time.
  • Higher speed vs noise/lifetime: tune soft-start slope + speed caps + avoid resonance bands.
  • More sensitive triggering vs false starts: fuse evidence (temp + flow/proxy) and raise confidence gates.

Figure F2 The metric map below shows the four success pillars and the minimal evidence signals that connect them to measurable debug points (used later in validation).

Figure F2 — Success Metrics Map Four-quadrant metrics map: comfort, energy, noise, reliability, connected to evidence signals. Success Balanced outcome Evidence-based Comfort Time-to-hot (s) Threshold stability Evidence Return temp curve Trigger timestamp Energy Wh/day budget Mode comparison Evidence Motor current Run-time counter Noise Worst-speed band Quiet operation Evidence Speed vs noise Current ripple Reliability Start/stop stress False-trip rate Evidence Event counters Fault logs
Figure F2. Success is defined by four measurable pillars. Each pillar must map to at least one hard evidence signal that can be captured during validation and field debug.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F2 (Success Metrics Map), 2026.

H2-3. Control Modes — Timer, Demand, Learning, and Hybrid (without platform creep)

A hot water recirculation controller succeeds when it turns “comfort” into repeatable local decisions: when to start, how long to run, and when to stop—all driven by evidence signals available on the device. The safest approach is to keep one shared state machine and allow different trigger sources (timer/demand/learning) to enter the same controlled sequence with identical anti-chatter rules.

Core stability rules (anti “flip-flop”): define Thigh/Tlow (temperature hysteresis), Ton_min/Toff_min (minimum on/off time), a preheat timeout, and a retry budget → lockout path. These four knobs prevent the most common failures: hunting on noisy temperature, short-cycling, and endless running after sensor faults.

Timer (Fixed windows)

Trigger: RTC window Signals: Temp + Pump current Stop: Thigh/Tlow + Timeout
  • Best for: predictable daily usage (morning/evening) with minimal sensors.
  • Key tuning: quiet hours, speed cap at night, Toff_min to avoid frequent starts.
  • Typical pitfall: runs when nobody is home → energy loss; solved by adding a local “occupancy gate.”

Demand (Event-driven)

Trigger: Button / ΔT / Flow evidence Signals: Temp + Proxy/Flow Stop: Thigh + Ton_min
  • Best for: reducing wasted runtime while keeping responsiveness.
  • Key tuning: trigger debouncing, confidence gates (temp trend + pump work evidence).
  • Typical pitfall: low-flow or noisy triggers cause false starts; solved by fusing temp rise rate and pump current.

Learning (Local pattern)

Trigger: local usage statistics Signals: Trigger timestamps Stop: Thigh/Tlow + Budget
  • Best for: homes with repeating patterns but irregular exact times.
  • Key tuning: decay/forgetting factor, minimum confidence before preheat, maximum daily energy budget.
  • Typical pitfall: pattern drift (weekends/season) → short cycling; solved by confidence gating and Toff_min.

Hybrid (Preheat + Demand)

Trigger: preheat windows + demand Signals: Temp + Pump work Stop: Hysteresis + Lockout
  • Best for: high comfort at peak hours, minimal waste during daytime.
  • Key tuning: arbitration (which trigger wins), cooldown window after a run, speed caps by time-of-day.
  • Typical pitfall: trigger conflicts → repeated starts; solved by a single shared state machine + Toff_min.

Minimal evidence set: a controller can remain robust with only return temperature and pump current. Return temperature defines “heat arrived,” while pump current proves “the pump is doing work” and helps detect dry-run/stall early.

Figure F2 The state machine below keeps all modes safe by sharing the same start/stop rules and anti-chatter guards.

Figure F2 — Control State Machine IDLE to PREHEAT to MAINTAIN with COOLDOWN and LOCKOUT, annotated with hysteresis and minimum on/off times. IDLE Sleep / Low power Wait for triggers PREHEAT Fast warm-up Ton_min enforced MAINTAIN Low duty / quiet Hold within band COOLDOWN Toff_min enforced No re-start LOCKOUT Retry budget exceeded Fault latched TRIGGERS Timer / Demand Learning / Hybrid Enter on trigger T ≥ Thigh or timeout T ≤ Tlow Toff_min Fault Retry > N Timer/Demand Start sequence Stability Knobs Thigh / Tlow (hysteresis) · Ton_min · Toff_min · Preheat timeout · Retry budget T t N
Figure F2. A single shared state machine keeps all modes stable. Different triggers may enter the sequence, but stop rules and anti-short-cycle guards remain identical.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F2 (Control State Machine), 2026.

H2-4. Temperature Sensing Chain — Where to Measure, How to Filter, How to Calibrate

Temperature is the primary “heat arrival” evidence, but it is also the most common source of false decisions when noise, thermal inertia, and motor switching interference are ignored. A robust chain needs three layers: correct measurement point, noise-aware sampling/filtering, and calibration + drift checks that keep thresholds meaningful over time.

Where to measure (control impact, not plumbing tutorial)

  • Return-side evidence is usually the most reliable stop condition because it reflects loop closure.
  • Pump-adjacent sensors may see local heating or coupling; require sanity checks (trend + pump work).
  • Single-point designs must use a second evidence signal (pump current or runtime) to avoid “fake hot.”

Two measurements that solve most ambiguity

  • Temperature ADC code: stability, ripple, and step response (hysteresis suitability).
  • Pump current waveform: proves work, reveals dry-run/stall, shows EMI-coupled artifacts.
  • Decision benefit: separates “real loop warm-up” from “local heat / noise injection.”

Sensor choice matrix (practical selection)

Option Strength Risk / coupling Calibration burden
NTC Low cost, simple divider, wide availability Non-linear; line noise + motor EMI can modulate ADC codes if routing is weak 1-point often acceptable; 2-point improves threshold accuracy
RTD Better linearity, stable over time, strong repeatability Needs excitation/current source; wiring resistance matters; requires clean analog ground 1-point may work; 2-point preferred when tight threshold bands are required

Noise-aware sampling rules: sample away from motor switching edges when possible; use a short analog RC to reduce high-frequency injection; apply robust digital filtering (median/trimmed mean) and limit physically impossible temperature slopes to catch coupling artifacts.

Calibration & drift checks (keep thresholds meaningful)

  • Boot self-check: detect open/short sensor conditions and reject out-of-range codes before enabling control.
  • 1-point vs 2-point: use 2-point if tight hysteresis bands are required or if sensor spread is large.
  • Drift monitor: track long-term offset trends; if bias grows, widen confidence gates or request service.

Figure F3 The front-end below shows the temperature evidence chain and the most common noise injection paths from the motor power stage.

Figure F3 — Temperature Evidence Front-End Temperature sensor chain from NTC/RTD through protection and filtering to ADC/MCU, with motor noise coupling arrows. NTC / RTD Sensor node Protection ESD / clamp series-R RC + Filter anti-PWM ripple median / mean ADC / MCU Thigh / Tlow sanity checks Motor Power Stage PWM edges / dv·dt ground bounce Coupling Paths sensor line injection ADC reference shift ground bounce → line injection → ADC ref shift → Capture (2 waveforms) Temperature ADC code + Pump current waveform → separates real warm-up from coupling artifacts
Figure F3. Temperature evidence must be built as a noise-aware chain. The most common failures come from motor PWM coupling into sensor wiring or ADC references.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F3 (Temperature Evidence Front-End), 2026.

H2-5. Flow/Presence Evidence — Flow Sensing Options & Proxy Signals

A recirculation controller does not need a dedicated flow meter to function, but it does need a reliable way to answer two questions: (1) Is the pump actually moving water? and (2) Is there a real hot-water intent right now? Direct flow sensing provides the strongest “loop is moving” evidence, while proxy signals can be made robust by using confidence gates that fuse pump-work evidence with temperature dynamics.

Direct flow sensing (strongest loop-closure evidence)

  • Hall turbine: low cost; sensitivity can degrade at very low flow; affected by debris and orientation.
  • Ultrasonic: wide dynamic range; higher BOM and power; careful mounting needed for repeatability.
  • dP inference: uses pressure drop across a restriction; low-flow resolution can be limited; needs stable reference.

Proxy evidence (when flow meter is unavailable)

  • Pump current / BEMF proxy: proves the motor is loaded; helps detect dry-run or stall patterns.
  • Temperature slope (dT/dt): confirms heat is propagating through the return path, not just local heating.
  • Optional acoustic/vibration: may help detect cavitation/no-flow, but must be treated as low-confidence.

Direct vs Proxy — selection matrix (cost, power, robustness, failure modes)

Method Cost / Power Strength Typical misread sources Low-flow note
Hall turbine Low / Low Direct loop motion proof Debris, orientation, magnetic noise Resolution drops near threshold
Ultrasonic Medium–High / Medium Wide range, good repeatability Mounting variance, bubbles, coupling Usually better than turbine
dP inference Low–Medium / Low Works with stable restriction Baseline drift, valve position changes Weak at very low dP
Pump current proxy Low / Low Detects stall/dry-run patterns Supply ripple, PWM artifacts Does not measure flow directly
dT/dt (temp slope) Low / Low Confirms heat propagation Thermal inertia, sensor placement Slow response at low ΔT
Acoustic/vibration Medium / Low–Medium Can indicate cavitation/no-flow Mounting noise, environmental vibration Treat as optional/low confidence

Confidence gate (proxy-only builds): require pump-work evidence (current above a minimum for Ton_min) and a return-temperature trend (dT/dt above a minimum within a window). Add a physical sanity limit on dT/dt to reject EMI-induced “fake heat.” If gates fail repeatedly, enforce Toff_min and escalate to lockout to prevent short-cycling.

Presence / hot-water intent inputs (local signals only)

  • Button: strongest intent signal; simple debounce + minimum cooldown prevents repeated triggers.
  • PIR: best used as an enable gate (allow preheat when motion is recent), not as a continuous demand source.
  • Door contact: can bias intent near bathrooms/kitchens; treat as low-confidence and time-limited.

These inputs should only decide whether to enter the start sequence. They must not replace temperature stop rules or become a platform orchestration path.

Figure F4 Evidence fusion: direct flow (optional) + proxy signals + intent inputs, combined through confidence gates to avoid false starts.

Figure F4 — Flow & Presence Evidence Fusion Optional direct flow sensors and proxy signals feed a confidence gate that produces permit/keep running/fault hints, with local intent inputs as a separate gate. Direct Flow (optional) Hall turbine Ultrasonic dP inference Low-flow limits Proxy Evidence Pump current / load BEMF proxy dT/dt (return) Optional acoustic Sanity limits Intent Gate Button (high) PIR (enable) Door contact (low) Debounce + cooldown Confidence Gate Work: I > Imin for Ton_min Heat: dT/dt > slope_min Reject: dT/dt > phys_max Fail → Toff_min → lockout Start Permit enter PREHEAT Keep Running confirm loop motion Fault Hint dry-run / stall / noise Rule: Intent inputs only decide entry; stop rules remain temperature-based
Figure F4. Direct flow sensing is optional; robust designs fuse proxy signals with confidence gates and keep presence inputs as a limited local trigger gate.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F4 (Flow & Presence Evidence Fusion), 2026.

H2-6. BLDC Pump Control — Commutation, Speed Control, and Quiet Operation

BLDC/ECM recirculation pumps behave like small motor systems: stable comfort requires stable torque, and quiet operation requires minimizing torque ripple and control hunting. The most reliable approach is to implement a minimum viable control loop: a speed target from the state machine, a current limit for abnormal loads, fault detection for stall/dry-run, and thermal derating to protect the driver and motor.

Power-stage building blocks (what must exist)

  • 3-phase bridge: integrated driver IC or external MOSFETs with a gate driver.
  • Current sense: shunt (low-cost) or sensor; used for limiting and fault signatures.
  • Temperature sense: driver/board temperature for derating and lockout thresholds.
  • Control MCU: commutation + speed loop + protection policy (retry/lockout).

Hall 6-step (sensor-based)

Startup: strong Complexity: low Best for: stable low-speed
  • Fit: pumps with Hall sensors; reliable start under load.
  • Risk: Hall noise/glitches can cause speed ripple → acoustic noise; apply debounce and consistency checks.
  • Quiet lever: soft-start + speed ramp to avoid resonant bands.

Sensorless BEMF 6-step

Startup: sensitive Complexity: medium Best for: reduced wiring
  • Fit: when sensors are not available; lower BOM/connector count.
  • Risk: weak BEMF at low speed or with air in the loop; requires open-loop start then handover to closed-loop.
  • Quiet lever: avoid aggressive handover; enforce current limits during start.

FOC (fit-based mention)

Startup: depends Complexity: high Best for: lowest ripple
  • Fit: when low noise and efficiency matter across a wide speed range.
  • Risk: needs good current sensing and tuning; under-resourced implementations can be less stable than 6-step.
  • Quiet lever: lower torque ripple, but only if sensing and control bandwidth are adequate.

Quiet operation knobs (most common wins)

PWM freq Slew control Soft-start Speed limits
  • PWM frequency: avoid audible bands and structural resonance zones; keep margins for losses and EMC.
  • Slew control: reduce dv/dt to limit EMI and acoustic excitation (at the cost of switching loss).
  • Soft-start: ramp speed/current to avoid sudden hydraulic noise and torque steps.
  • Loop stability: cap speed-loop aggressiveness to avoid hunting near thresholds.

Minimum viable control loop checklist: speed target → current limit → stall/dry-run detection → thermal derating → retry budget + lockout. This sequence is sufficient to keep the pump stable and quiet even when the system only has temperature as the primary comfort evidence.

Fault signatures (evidence-driven)

  • Stall: current rises but speed estimate does not; action: limit, retry, then lockout.
  • Dry-run / air: current/load pattern changes without the expected return-temperature trend; action: reduce speed, cooldown, retry budget.
  • Over-temp: temperature crosses derating threshold; action: speed cap or stop, then cooldown window.

Figure F5 Power-stage block diagram: control MCU, driver/bridge, current and temperature feedback, and the protection policy loop.

Figure F5 — BLDC Pump Power Stage & Control Loop MCU drives gate driver and MOSFET bridge to a BLDC pump motor, with current sense and temperature feedback for protection and quiet operation knobs. Control MCU speed target commutation fault policy Gate Driver integrated or external slew control 3-Phase Bridge MOSFETs PWM freq soft-start BLDC Pump hydraulic load air / cavitation Current Sense limit + signatures stall / dry-run Temp Sense driver / board derating I(t) limit Tboard Quiet / Stability knobs PWM freq · slew control · soft-start ramp · speed limit bands · retry budget
Figure F5. A minimum viable loop combines speed commands with current limiting, fault signatures, and temperature derating to keep the pump stable and quiet.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F5 (BLDC Power Stage & Control Loop), 2026.

H2-7. Protection & Fault Detection — Dry-Run, Stall, Cavitation, Overheat

Protection in a recirculation controller should be evidence-driven and designed to avoid nuisance trips. The most reliable pattern is to require evidence pairs (two independent signals) within a time window, then apply a tiered action ladder: derate → retry → lockout → local alert.

Nuisance-trip avoidance rules: use a startup grace window for thermal inertia, confirm faults with N-of-M windows (multi-window confirmation), and enforce Ton_min / Toff_min to prevent short-cycling. Single-sample decisions are not robust in the presence of PWM noise and hydraulic transients.

Symptom → Evidence → Action (field-debug ready)

Symptom (what is observed) Evidence (use pairs + window) Action ladder (avoid false stop)
Dry-run / no circulation
pump sounds light, heat does not arrive
  • Low load: pump current/power lower than expected after Ton_min
  • No heat propagation: return temperature trend missing (dT/dt below slope_min in window)
  • Guard: apply startup grace window; down-weight dT/dt when ΔT is naturally small
  • Derate: reduce speed and extend observation window
  • Retry: cooldown + limited retries (with backoff)
  • Lockout: only after retry budget exceeded
  • Alert: local LED/buzzer if persistent
Stall / jam
pump stops or hums, no rotation
  • High current: current rises toward limit and persists for window
  • No speed rise: Hall edges absent or sensorless commutation fails
  • Guard: allow short high-current during soft-start; confirm persistence before lockout
  • Immediate limit: current clamp + stop if signature is strong
  • Retry: small retry budget (jam risk is high)
  • Lockout: fast escalation if repeated stall signature
  • Alert: local indication for service
Cavitation / air
raspy noise, unstable flow
  • Ripple signature: elevated current ripple / RMS in specific speed band
  • Efficiency drop: weaker return dT/dt for similar input power, or direct-flow instability (if present)
  • Guard: enable only in defined speed band; require multi-window confirmation
  • Derate: move away from noisy speed band, reduce speed
  • Retry: cooldown + re-attempt at lower ramp rate
  • Lockout: only if persistent with clear signatures
  • Alert: optional local warning (non-critical)
Overheat
driver/board temperature rises
  • Temperature: board/driver temperature crosses derate threshold
  • Duration: sustained high load or repeated starts within short period
  • Guard: hysteresis + cooldown window to prevent on/off chatter
  • Derate: cap speed or duty as temperature rises
  • Stop: if temperature crosses stop threshold
  • Cooldown: enforce minimum off-time
  • Lockout: if repeated over-temp cycles occur

Tiered action ladder (recommended default)

  • Derate first: reduce speed / limit current to keep comfort while observing evidence longer.
  • Retry second: cooldown then retry with a slower ramp; track retry budget.
  • Lockout last: only for persistent strong signatures (stall, repeated overheat) or retry budget exceeded.
  • Local alert: LED/buzzer conveys state without platform dependency.

Figure F6 Fault branches integrated into the state machine: derate/retry paths and lockout conditions with guard rails.

Figure F6 — Fault Branches Integrated into State Machine IDLE, PREHEAT, MAINTAIN, COOLDOWN, and LOCKOUT states with evidence-driven fault bubbles and guard rails for derate, retry, and lockout. IDLE sleep / RTC PREHEAT speed ramp Ton_min MAINTAIN quiet band hysteresis COOLDOWN Toff_min retry backoff LOCKOUT local alert manual clear permit Treturn ≥ Thigh end / Treturn ≤ Tlow cooldown done Dry-run I low + no dT/dt Stall I high + speed low Cavitation I ripple + weak heat Overheat Tboard high derate/retry fast lockout Guard rails startup grace · N-of-M windows · Ton_min/Toff_min · cooldown · retry budget
Figure F6. Fault detection is integrated into the control states. Evidence pairs feed derate/retry paths, while persistent high-risk signatures escalate to lockout.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F6 (Fault Branches into State Machine), 2026.

H2-8. Power Tree & Low-Power Design — Always-On Controller Done Right

A hot water recirculation controller is typically always-on: it must keep time, sample sensors, and wake predictably while surviving input dips and surges. A robust implementation separates power domains (motor supply vs always-on logic vs sensors), uses UVLO/BOR for consistent reset behavior, and budgets energy across sleep, sampling bursts, run-time, and optional radio events.

Board-level scope (what matters here)

  • Input: 12/24V DC (or adapter output) with board-level protection and UVLO.
  • Domains: motor supply, always-on 3.3V, sensor rails (switchable).
  • Consistency: brownout reset and safe restart policy prevent “resume pumping” after a dip.

Always-on power budget template (fill with measured currents)

State Typical duration/day Current range Energy/day (formula) Reduction levers
Standby sleep dominant (hours) µA–low mA E ≈ I_sleep · V_ao · t deep sleep + RTC, disable sensor rails
Sampling burst minutes total mA–tens mA E ≈ Σ(I_samp · V_ao · Δt) short window, lower rate, median/trimmed mean
Pump running event-driven controller mA + motor W E ≈ V_in · I_motor · t_run quiet bands, derate, stop on confidence
Optional radio event rare / short tens–hundreds mA (peaks) E ≈ Σ(I_radio · V_ao · Δt) batch updates, short connect window

Safe restart policy: on UVLO/BOR events, stop the motor drive first, then restore into a safe IDLE/cooldown state. Re-enter PREHEAT only after sensors re-validate temperature and intent evidence. This prevents repeated dips from creating rapid on/off cycles.

Low-power tactics (always-on friendly)

  • MCU sleep: RTC wake for schedule slots and periodic sensor sampling.
  • Intermittent sensing: power-gate sensor rails; sample in bursts, then compute stable statistics.
  • Brownout consistency: BOR thresholds chosen so firmware never runs in an undefined voltage region.
  • Write discipline: store configuration and counters with rate limiting to avoid frequent NVM writes.

Figure F7 Power tree with domains: motor supply, always-on 3.3V, and switchable sensor rails, plus UVLO/BOR and safe restart hooks.

Figure F7 — Power Tree & Domains (Always-On Controller) 12/24V input splits into motor supply, always-on 3.3V, and switchable sensor rails with UVLO/BOR, sequencing, and safe restart behavior. 12/24V Input board protection UVLO Motor Domain 3-phase driver bridge + pump inrush/peaks Always-On 3.3V low-Iq buck/LDO MCU + RTC BOR Sensor Rails switchable temp / flow / intent burst sampling Safe Restart stop drive first IDLE + cooldown re-validate rail gate BOR / reset cause UVLO event Always-on budget states sleep (RTC) · sampling burst · pump run · optional radio event Goal: predictable wake + consistent reset + no unintended motor resume
Figure F7. Separate motor, always-on, and sensor domains. Combine UVLO/BOR with safe restart sequencing for predictable behavior under dips and surges.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F7 (Power Tree & Domains), 2026.

H2-9. EMC & Ruggedness (Device-Level) — Motor EMI, ESD/Surge Evidence Points

When the pump starts, the power stage can inject noise through three dominant paths: dv/dt coupling from the switching node, ground bounce from high-current return loops, and conducted ripple on the input rail. The fastest route to a fix is to measure a small set of probe points and map the symptom to a coupling path.

Device-level boundary: this chapter focuses on board-level evidence points and first hardware fixes that stop resets and RF dropouts. It does not expand into compliance standards or full certification workflows.

Noise paths → what is seen → first fixes

  • dv/dt (SW node ringing) → radiated/capacitive coupling: RF dropouts worsen at pump PWM edges; near-field noise rises near antenna/XTAL. First fixes: gate resistor / slew control, snubber, reduce loop area around bridge, keep SW copper tight.
  • Ground bounce → reference shift: ADC codes jump, false faults, MCU brownouts correlate with pump current. First fixes: split motor return vs logic ground, single controlled join, relocate sense returns, add local decoupling at MCU/RF.
  • Conducted ripple on VIN → rail dips: MCU resets at start/stop events; 3.3V droops track VIN spikes. First fixes: TVS + π filter, common-mode choke (as needed), improve input bulk and placement, tighten UVLO/BOR behavior.

Top 6 probes (start here)

  • P1 — VIN at board entry: ripple spikes at pump start/stop; use edge-trigger on pump enable.
  • P2 — MCU 3.3V (or 1.8V) at MCU pins: dips / ringing; correlate with BOR/reset cause.
  • P3 — Driver supply rail: sag or oscillation during commutation and PWM updates.
  • P4 — SW/phase node: dv/dt level and ringing; compare “pump on” vs “pump off”.
  • P5 — Current sense + its return: ground bounce contaminating thresholds and fault logic.
  • P6 — RF near-field (antenna/XTAL zone): noise rise when pump runs; track with RSSI dropouts.

Minimal rugged hardware stack (board-level)

  • Input protection: TVS sized for the input environment; clamp path kept short to return.
  • Input filtering: π filter or CM choke where cable-borne noise dominates; place close to entry.
  • Switch-edge control: gate resistor and/or snubber to reduce ringing and radiated coupling.
  • Domain separation: motor power/return separated from always-on logic; controlled join point.
  • Sensitive rails: RF/MCU local LDO + tight decoupling; keep noisy returns away from references.

Figure F8 Device-level coupling map with noise arrows and probe points P1–P6.

Figure F8 — EMI Coupling Map & Probe Points (P1–P6) Power stage with dv/dt, ground bounce, and conducted ripple arrows, plus probe points for VIN, rails, SW node, current sense, and RF area. 12/24V IN TVS + π filter cable entry Driver gate R / snubber slew control 3-Phase Bridge SW / phase nodes dv/dt source BLDC Pump motor load Always-On MCU + RTC 3.3V (BOR) local decoupling Sensors temp / flow / intent thresholds RF Zone antenna XTAL RSSI drop dv/dt coupling ground bounce conducted ripple P1 P2 P3 P4 P5 P6 Probe → classify → fix Resets: verify P2 dips + P1 ripple · RF dropouts: compare P4 dv/dt + P6 near-field First fixes: input filter · edge control · return-path discipline · domain isolation
Figure F8. A device-level coupling map connects symptoms to measurable probe points (P1–P6) and the most likely hardware fixes.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F8 (EMI Coupling Map & Probe Points), 2026.

H2-10. Validation Plan — Test Matrix for Comfort, Energy, Noise, and Faults

Validation should prove that the controller behaves correctly across modes, meets comfort/energy/noise targets, survives repeated edge cases (dry-run, stall, cavitation-like conditions), and remains stable under device-level ESD/EFT pre-checks. The matrix below is organized by Test item → Setup → Pass/Fail evidence.

Coverage categories

  • Functional: mode switching, hysteresis, fault ladder, brownout recovery.
  • Performance: wait time, return temperature stability, energy estimate, acoustic noise.
  • Reliability: start-cycle endurance, thermal cycling, moisture stress, repeated fault recovery.
  • EMC pre-check: device-level ESD/EFT spot checks with evidence of no resets or safe recovery.

Test Matrix — Test item → Setup → Pass/Fail evidence

Test item Setup Pass/Fail evidence
Mode switching (timer/demand/hybrid) Inject local triggers; vary schedule slots; log state transitions. State machine follows expected path; no short-cycling; transitions timestamped and repeatable.
Hysteresis & anti-short-cycle Sweep Treturn around thresholds; enforce Ton_min/Toff_min settings. No chatter near thresholds; run/stop decisions stable across multiple sweeps.
Brownout recovery Create controlled VIN dips; monitor reset cause and rails. Motor stops first; controller returns to IDLE/cooldown; no unintended “resume pumping”.
Fault ladder execution (derate → retry → lockout) Force dry-run/stall signatures; observe action timing and retry budget. Derate occurs before lockout; retry backoff honored; lockout only after budget/strong signature.
Comfort wait-time Start from cold loop; record time until Treturn reaches target. Wait-time meets target band; Treturn curve matches expected ramp profile.
Return temperature stability Run maintain mode; vary ambient and draw events; log Treturn. Treturn stays within band; no oscillation; hysteresis behaves as designed.
Energy estimate (Wh/day) Measure VIN current over day profile; integrate sleep/sample/run periods. Energy model matches measured integration within acceptable error; dominates identified states.
Acoustic noise Measure at fixed distance; sweep speeds; note “quiet bands”. Noise stays under threshold in intended operating band; no prominent tonal peaks at target speeds.
Start-cycle endurance Automate repeated start/stop cycles; track failures and temperature. Failure rate below limit; no progressive drift in current/temperature signatures.
Thermal cycling Cycle ambient/board temperature; repeat functional tests at extremes. Thresholds and protections remain stable; no unexpected resets; calibration remains valid.
Moisture stress (device-level) Operate in humid environment; observe leakage/false triggers. No false faults; insulation-related anomalies are logged and handled safely.
Repeated fault recovery Alternate dry-run/stall-like events with normal runs; track lockout triggers. Recoveries follow ladder; lockout occurs only on persistent strong signatures; manual clear works.
ESD spot check (device-level) Apply ESD to user-accessible points; monitor P2 (MCU rail) and reset cause. No uncontrolled latch-up; no unsafe motor resume; safe recovery or controlled lockout + log.
EFT/burst spot check (device-level) Inject burst on input cable; observe P1/P2/P4 simultaneously. Controller stays stable or recovers safely; no repeated reset loop; evidence captured in event log.

Evidence-first mindset: log state transitions, retry counters, and reset causes. During validation, capture at least VIN ripple (P1), MCU rail (P2), and SW node (P4) together to correlate symptoms with coupling paths.

Figure F9 Test wiring and measurement map: programmable supply, thermal/NTC simulation, motor load, and probe points.

Figure F9 — Validation Setup & Measurement Map Programmable supply feeds DUT; thermal/NTC simulation and motor load connect; oscilloscope probes at VIN, MCU rail, SW node, and current sense. Programmable Supply 12/24V events dips / surges Oscilloscope P1 VIN P2 3.3V Spectrum / Near-field RF zone scan P6 antenna/XTAL DUT Controller Board MCU + driver + sensing event log / counters fault ladder BLDC Pump Load normal / stall dry-run signature Temp / NTC Sim resistor decade or heat source Proxy Evidence current + dT/dt VIN 3-phase sensor evidence P1 P2 P4 P5 P6 Pass/Fail Evidence no uncontrolled reset safe restart (IDLE) fault ladder correct events logged
Figure F9. A practical validation wiring map ties the test matrix to measurable evidence: VIN events, rail stability, switching behavior, and RF susceptibility.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F9 (Validation Setup & Measurement Map), 2026.

H2-9. EMC & Ruggedness (Device-Level) — Motor EMI, ESD/Surge Evidence Points

When the pump starts, the power stage can inject noise through three dominant paths: dv/dt coupling from the switching node, ground bounce from high-current return loops, and conducted ripple on the input rail. The fastest route to a fix is to measure a small set of probe points and map the symptom to a coupling path.

Device-level boundary: this chapter focuses on board-level evidence points and first hardware fixes that stop resets and RF dropouts. It does not expand into compliance standards or full certification workflows.

Noise paths → what is seen → first fixes

  • dv/dt (SW node ringing) → radiated/capacitive coupling: RF dropouts worsen at pump PWM edges; near-field noise rises near antenna/XTAL. First fixes: gate resistor / slew control, snubber, reduce loop area around bridge, keep SW copper tight.
  • Ground bounce → reference shift: ADC codes jump, false faults, MCU brownouts correlate with pump current. First fixes: split motor return vs logic ground, single controlled join, relocate sense returns, add local decoupling at MCU/RF.
  • Conducted ripple on VIN → rail dips: MCU resets at start/stop events; 3.3V droops track VIN spikes. First fixes: TVS + π filter, common-mode choke (as needed), improve input bulk and placement, tighten UVLO/BOR behavior.

Top 6 probes (start here)

  • P1 — VIN at board entry: ripple spikes at pump start/stop; use edge-trigger on pump enable.
  • P2 — MCU 3.3V (or 1.8V) at MCU pins: dips / ringing; correlate with BOR/reset cause.
  • P3 — Driver supply rail: sag or oscillation during commutation and PWM updates.
  • P4 — SW/phase node: dv/dt level and ringing; compare “pump on” vs “pump off”.
  • P5 — Current sense + its return: ground bounce contaminating thresholds and fault logic.
  • P6 — RF near-field (antenna/XTAL zone): noise rise when pump runs; track with RSSI dropouts.

Minimal rugged hardware stack (board-level)

  • Input protection: TVS sized for the input environment; clamp path kept short to return.
  • Input filtering: π filter or CM choke where cable-borne noise dominates; place close to entry.
  • Switch-edge control: gate resistor and/or snubber to reduce ringing and radiated coupling.
  • Domain separation: motor power/return separated from always-on logic; controlled join point.
  • Sensitive rails: RF/MCU local LDO + tight decoupling; keep noisy returns away from references.

Figure F8 Device-level coupling map with noise arrows and probe points P1–P6.

Figure F8 — EMI Coupling Map & Probe Points (P1–P6) Power stage with dv/dt, ground bounce, and conducted ripple arrows, plus probe points for VIN, rails, SW node, current sense, and RF area. 12/24V IN TVS + π filter cable entry Driver gate R / snubber slew control 3-Phase Bridge SW / phase nodes dv/dt source BLDC Pump motor load Always-On MCU + RTC 3.3V (BOR) local decoupling Sensors temp / flow / intent thresholds RF Zone antenna XTAL RSSI drop dv/dt coupling ground bounce conducted ripple P1 P2 P3 P4 P5 P6 Probe → classify → fix Resets: verify P2 dips + P1 ripple · RF dropouts: compare P4 dv/dt + P6 near-field First fixes: input filter · edge control · return-path discipline · domain isolation
Figure F8. A device-level coupling map connects symptoms to measurable probe points (P1–P6) and the most likely hardware fixes.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F8 (EMI Coupling Map & Probe Points), 2026.

H2-10. Validation Plan — Test Matrix for Comfort, Energy, Noise, and Faults

Validation should prove that the controller behaves correctly across modes, meets comfort/energy/noise targets, survives repeated edge cases (dry-run, stall, cavitation-like conditions), and remains stable under device-level ESD/EFT pre-checks. The matrix below is organized by Test item → Setup → Pass/Fail evidence.

Coverage categories

  • Functional: mode switching, hysteresis, fault ladder, brownout recovery.
  • Performance: wait time, return temperature stability, energy estimate, acoustic noise.
  • Reliability: start-cycle endurance, thermal cycling, moisture stress, repeated fault recovery.
  • EMC pre-check: device-level ESD/EFT spot checks with evidence of no resets or safe recovery.

Test Matrix — Test item → Setup → Pass/Fail evidence

Test item Setup Pass/Fail evidence
Mode switching (timer/demand/hybrid) Inject local triggers; vary schedule slots; log state transitions. State machine follows expected path; no short-cycling; transitions timestamped and repeatable.
Hysteresis & anti-short-cycle Sweep Treturn around thresholds; enforce Ton_min/Toff_min settings. No chatter near thresholds; run/stop decisions stable across multiple sweeps.
Brownout recovery Create controlled VIN dips; monitor reset cause and rails. Motor stops first; controller returns to IDLE/cooldown; no unintended “resume pumping”.
Fault ladder execution (derate → retry → lockout) Force dry-run/stall signatures; observe action timing and retry budget. Derate occurs before lockout; retry backoff honored; lockout only after budget/strong signature.
Comfort wait-time Start from cold loop; record time until Treturn reaches target. Wait-time meets target band; Treturn curve matches expected ramp profile.
Return temperature stability Run maintain mode; vary ambient and draw events; log Treturn. Treturn stays within band; no oscillation; hysteresis behaves as designed.
Energy estimate (Wh/day) Measure VIN current over day profile; integrate sleep/sample/run periods. Energy model matches measured integration within acceptable error; dominates identified states.
Acoustic noise Measure at fixed distance; sweep speeds; note “quiet bands”. Noise stays under threshold in intended operating band; no prominent tonal peaks at target speeds.
Start-cycle endurance Automate repeated start/stop cycles; track failures and temperature. Failure rate below limit; no progressive drift in current/temperature signatures.
Thermal cycling Cycle ambient/board temperature; repeat functional tests at extremes. Thresholds and protections remain stable; no unexpected resets; calibration remains valid.
Moisture stress (device-level) Operate in humid environment; observe leakage/false triggers. No false faults; insulation-related anomalies are logged and handled safely.
Repeated fault recovery Alternate dry-run/stall-like events with normal runs; track lockout triggers. Recoveries follow ladder; lockout occurs only on persistent strong signatures; manual clear works.
ESD spot check (device-level) Apply ESD to user-accessible points; monitor P2 (MCU rail) and reset cause. No uncontrolled latch-up; no unsafe motor resume; safe recovery or controlled lockout + log.
EFT/burst spot check (device-level) Inject burst on input cable; observe P1/P2/P4 simultaneously. Controller stays stable or recovers safely; no repeated reset loop; evidence captured in event log.

Evidence-first mindset: log state transitions, retry counters, and reset causes. During validation, capture at least VIN ripple (P1), MCU rail (P2), and SW node (P4) together to correlate symptoms with coupling paths.

Figure F9 Test wiring and measurement map: programmable supply, thermal/NTC simulation, motor load, and probe points.

Figure F9 — Validation Setup & Measurement Map Programmable supply feeds DUT; thermal/NTC simulation and motor load connect; oscilloscope probes at VIN, MCU rail, SW node, and current sense. Programmable Supply 12/24V events dips / surges Oscilloscope P1 VIN P2 3.3V Spectrum / Near-field RF zone scan P6 antenna/XTAL DUT Controller Board MCU + driver + sensing event log / counters fault ladder BLDC Pump Load normal / stall dry-run signature Temp / NTC Sim resistor decade or heat source Proxy Evidence current + dT/dt VIN 3-phase sensor evidence P1 P2 P4 P5 P6 Pass/Fail Evidence no uncontrolled reset safe restart (IDLE) fault ladder correct events logged
Figure F9. A practical validation wiring map ties the test matrix to measurable evidence: VIN events, rail stability, switching behavior, and RF susceptibility.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F9 (Validation Setup & Measurement Map), 2026.

H2-11. Field Debug Playbook — Symptom → Evidence → Isolate → Fix

This playbook is designed for fast field isolation using a small, repeatable evidence set: VIN ripple (P1), MCU rail dips (P2), pump current (Ipump), and return temperature slope (dT/dt). Each symptom below uses the same four-block SOP format to prevent guess-based troubleshooting.

Common evidence set (recommended)

  • P1: VIN at board entry (ripple spikes at start/stop).
  • P2: MCU 3.3V/1.8V at pins (rail dips / BOR signature).
  • Ipump: pump phase/bus current (startup peak + steady pattern).
  • Treturn: return temperature ADC code vs time (dT/dt).
  • Optional: SW node ringing (P4), current-sense return bounce (P5), RF near-field (P6).

SOP No recirculation / very short runs

First 2 measurements

  • Ipump: capture startup peak and 2–3 seconds of steady current.
  • Treturn dT/dt: log temperature ADC code for 30–120 seconds.

Discriminator

  • High Ipump + no speed build → likely stall / overcurrent protection action.
  • Normal Ipump + very weak dT/dt → likely threshold/lag window too aggressive or temperature sensing placement/filtering issue.
  • Low Ipump + no meaningful dT/dt → likely dry-run signature or air/prime not established (treat as dry-run evidence).

First fix (minimum cost)

  • Enforce Ton_min and thermal-lag window before judging “no heat movement”.
  • Raise hysteresis and add N-of-M confirmation to avoid single-sample aborts.
  • If stall evidence is present: add/verify current limit slope and retry backoff.

MPN examples (hardware levers)

TI INA240A1TI INA181A1Vishay WSL2512 0.01Ω TI DRV10983TI DRV8323RSST STSPIN233 Murata BLM31PG600SN1LWürth 74438357006

Prevent

  • Log: run attempts/day, abort reason, retry counters; flag repeated short-run patterns.
  • Lockout only after persistent evidence (avoid “single spike” lockouts).

SOP Very noisy / resonance at certain speeds

First 2 measurements

  • Ipump ripple: compare RMS/peak-to-peak at “quiet speed” vs “noisy speed”.
  • SW ringing (P4) or driver PWM edge behavior: capture at the same operating points.

Discriminator

  • Noisy band + Ipump ripple rises sharply → likely resonance / cavitation-like operating region; prefer speed avoidance.
  • Noisy across many speeds + large SW ringing → edge too aggressive; likely EMI/mechanical excitation from dv/dt.

First fix (minimum cost)

  • Create “quiet bands”: avoid narrow speed regions that consistently spike ripple/noise.
  • Add soft-start slope and limit acceleration (reduces excitation).
  • If SW ringing dominates: increase gate resistor or add RC snubber near bridge.

MPN examples (edge & drive)

TI DRV8323RSTI DRV8301ST L6234 Nexperia PSMN2R8-30YLCInfineon BSC009NE2LS5I Vishay CRCW1206 10ΩTDK C3216X7R1H104K

Prevent

  • Maintain a speed-to-noise map in validation; keep it stable across temperature and supply variation.
  • Do not treat “noise-only” as a fault unless paired with current/temperature evidence.

SOP Reboots when the pump starts

First 2 measurements

  • P2: MCU 3.3V/1.8V dip magnitude and duration (look for BOR/POR signature).
  • P1: VIN ripple spikes synchronized with pump enable.

Discriminator

  • P2 dips first → power tree / decoupling / UVLO/BOR issue (conducted event).
  • P2 stable but RF drops / logic glitches → dv/dt coupling / ground bounce / return-path contamination.

First fix (minimum cost)

  • Harden always-on rail: add local bulk + low-ESR decoupling at MCU/RF; confirm BOR threshold behavior.
  • Strengthen input entry: TVS + π filter near connector; shorten clamp/return path.
  • Reduce source noise: gate resistor / snubber and keep SW copper compact; separate power return from logic return.

MPN examples (PI & input)

TI LM5163ATI TPS54202MPS MP1584EN TI TPS7A2033Microchip MCP1700-3302E Littelfuse SMBJ33ADiodes Inc. SMBJ33A Würth 744232101Murata BLM41PG600SN1L Panasonic EEE-FK1V470P

Prevent

  • Record reset cause and rail min/max during pump start events (field log).
  • Use controlled join for grounds and keep current-sense return away from MCU reference.

SOP Temperature reading jumps / false triggers

First 2 measurements

  • Treturn ADC raw vs filtered: capture with pump off and pump on.
  • Ipump: check if temperature spikes are time-aligned with commutation/PWM edges.

Discriminator

  • Spikes only when pump runs → EMI injection / sampling phase problem / ground reference shift.
  • Spikes even when pump is off → sensor wiring/contact, divider/reference instability, moisture leakage paths.

First fix (minimum cost)

  • Move ADC sampling away from switching edges; add digital median/outlier reject.
  • Add/verify RC at sensor input and ensure sensor return is routed to a quiet reference point.
  • Increase hysteresis and require multi-window confirmation before mode switching.

MPN examples (sensing chain)

Murata NCP18XH103F03RBVishay NTCLE100E3103JB0 Vishay PTS0805 1kTI OPA333AIDBVR Yageo RC0603FR-0710KLTDK C1608X7R1H104K

Prevent

  • Self-check on boot: detect open/short sensor; log drift rate and flag abnormal patterns.
  • Keep sensor harness and ADC reference separated from motor power loops.

SOP Energy use unexpectedly high

First 2 measurements

  • Run-time counters: minutes/day in PREHEAT/MAINTAIN and number of starts/day.
  • Trigger attribution: count timer vs demand vs learning decisions.

Discriminator

  • Too many starts/day → thresholds too sensitive or anti-short-cycle missing; likely false “need heat” triggers.
  • Long maintain time → temperature band too tight; hysteresis too small; learning schedule too aggressive.

First fix (minimum cost)

  • Increase Toff_min, widen temperature band, and require confirmed dT/dt before entering MAINTAIN.
  • Cap daily run-time and add a local “energy guard” rule to prompt retuning.
  • If always-on current is high: audit sleep state and sensor duty cycle (budget by states).

MPN examples (always-on efficiency)

ST STM32L072KZST STM32L052K8NXP MKL03Z32 TI TPS62743TI TPS7A0233 Microchip MCP7940NABLIC S-1335A33

Prevent

  • Persist daily counters and expose a simple field report: starts/day, run-min/day, abort reasons/day.
  • Keep learning local and bounded (no platform dependency); clamp schedule aggressiveness.

Figure F6 Decision tree for fast isolation using “Pump spins → Current normal → dT/dt present”.

Figure F6 — Field Debug Decision Tree A three-step decision tree: Pump spins, current normal, temperature rise slope observed, with branches to likely root causes and first fixes. 3-Step Field Isolation Tree (Evidence-First) Step 1 Pump spins? Step 2 Current normal? Step 3 dT/dt present? NO HIGH / LOW NO No spin → isolate power/drive Check P2 rail dips + enable Then driver + wiring Current high → stall/overload Limit + retry backoff Verify speed build Current low → dry-run signature Use dT/dt + time window Derate → retry No dT/dt → sensing/threshold Increase lag window Check EMI injection dT/dt OK → tune comfort/energy Widen band + anti-short-cycle Validate noise bands Probe tags: P1 VIN ripple P2 MCU rail dips I pump current T dT/dt
Figure F6. A minimal evidence tree accelerates isolation: confirm spin, classify current signature, then validate heat movement (dT/dt).
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F6 (Field Debug Decision Tree), 2026.

Request a Quote

Accepted Formats

pdf, csv, xls, xlsx, zip

Attachment

Drag & drop files here or use the button below.

H2-12. FAQs (Accordion ×12) — Evidence-First, No Scope Creep

Rules

  • Each answer starts with the first 2 checks, then a discriminator, then a first fix.
  • Each answer ends with a strict chapter backlink: “→ Back to H2-x”.
  • No plumbing tutorial, no water-heater internals, no cloud/platform deep dive.
Q1 Timer is ON but hot water is still slow — check temperature point or pump speed first?
First 2 checks: log Treturn dT/dt for 30–120 s and capture Ipump (startup + steady). Discriminator: normal current/speed but weak dT/dt usually means sensor placement/lag window/hysteresis is too strict; low speed or early limiting points to the drive loop. First fix: add a thermal-lag window before judging “no heat movement,” and confirm target speed + current limit ramp.
MPN examples: TI DRV10983TI DRV8323RSTI INA240A1
→ Back to H2-4 / H2-6 / H2-10
Q2 Wi-Fi/Thread drops as soon as the pump starts — which two waveforms first?
First 2 checks: probe P2 MCU rail dips (3.3 V/1.8 V at pins) and P1 VIN ripple at board entry with the same trigger. Discriminator: if P2 dips first, it is power integrity (UVLO/BOR/decoupling). If rails stay solid but links drop, suspect dv/dt coupling and ground bounce. First fix: harden always-on rails and reduce source noise (gate R/snubber + compact SW loop).
MPN examples: TI LM5163ATI TPS7A2033Littelfuse SMBJ33AMurata BLM41PG600SN1L
→ Back to H2-9 / H2-8
Q3 False starts at night — overly sensitive threshold or drifting occupancy input?
First 2 checks: attribute triggers (timer vs demand vs occupancy) and plot Treturn noise/drift over the same period. Discriminator: if occupancy events dominate, fix input stability (debounce/pull/brownout behavior). If temperature triggers dominate, the sensor chain is likely noisy or sampled at a bad phase. First fix: add N-of-M confirmation + minimum OFF time, then tighten input debounce and filter outliers.
→ Back to H2-3 / H2-5
Q4 Pump is very hot but there is no effect — dry-run or stall, and how to prove it?
First 2 checks: capture Ipump signature and confirm dT/dt (heat movement) during the run. Discriminator: high current without speed build points to stall/overload; low current with no dT/dt matches dry-run/air/prime loss; strong ripple + noise can indicate a cavitation-like operating region. First fix: derate speed, enforce retry backoff, and lock out only after persistent evidence.
→ Back to H2-7
Q5 Hot water arrives but cools quickly — hysteresis tuning or recirculation strategy?
First 2 checks: review MAINTAIN duty cycle (minutes/day) and the temperature band (upper/lower thresholds). Discriminator: rapid on/off cycling means hysteresis is too tight or Toff_min is missing; short “warm then cool” windows often mean a strategy biased to brief preheat instead of bounded maintain. First fix: widen the band, add anti-short-cycle timing, and cap daily maintain time to reduce heat loss.
→ Back to H2-3 / H2-2
Q6 Noise happens only at one speed band — PWM setting or mechanical resonance first?
First 2 checks: compare Ipump ripple at “quiet speed” vs “noisy speed,” and capture P4 SW ringing at the same points. Discriminator: a narrow noisy band with ripple spike is usually resonance/cavitation-like behavior → avoid that band. Broad noise with large SW ringing indicates edges are too aggressive. First fix: define “quiet bands,” soften acceleration, then add gate R/snubber only if SW ringing evidence is dominant.
MPN examples: Vishay CRCW1206 10ΩTDK C3216X7R1H104KTI DRV8323RS
→ Back to H2-6 / H2-10
Q7 Temperature display jitters — ADC sampling problem or motor EMI injection?
First 2 checks: log Treturn raw vs filtered with pump OFF and ON, and check if spikes align with commutation/PWM events in Ipump. Discriminator: spikes only during pump operation imply injection via ground reference or sampling phase; spikes even when OFF imply sensor wiring/reference instability or moisture leakage paths. First fix: move sampling away from switching edges, add RC + median/outlier reject, and route sensor return to a quiet reference node.
MPN examples: Murata NCP18XH103F03RBTI OPA333AIDBVRTDK C1608X7R1H104K
→ Back to H2-4 / H2-9
Q8 Power consumption is high — check recirculation duty cycle or pump efficiency point?
First 2 checks: read run minutes/day and starts/day from logs, then correlate with trigger source (timer/demand/learning). Discriminator: too many starts imply sensitive thresholds or missing Toff_min; long maintain time implies overly tight band. If current is high for the same heat outcome, the pump may be off its efficient speed point. First fix: widen the band, cap daily run time, and add a local energy guard rule; then limit speed and verify dT/dt per Wh.
→ Back to H2-2 / H2-6
Q9 Frequent reboots — UVLO first or ground bounce first?
First 2 checks: probe P2 rail dips and capture reset cause (BOR/POR/watchdog if available). Discriminator: if P2 dips coincide with pump start/stop, prioritize UVLO/decoupling and input filtering. If rails look stable but logic glitches appear, prioritize ground bounce, SW loop coupling, and current-sense return contamination. First fix: strengthen always-on rails, shorten clamp/return paths, and separate power return from logic reference; reduce dv/dt with gate control when SW ringing is large.
MPN examples: TI TPS54202Würth 744232101Littelfuse SMBJ33A
→ Back to H2-8 / H2-9
Q10 Button press does nothing — debounce issue or wake domain issue?
First 2 checks: verify GPIO interrupt/wake counter increments and confirm the always-on rail stays within spec during sleep-to-wake. Discriminator: no wake event implies input biasing/debounce/pull configuration issues; wake occurs but no action implies the controller is locked out (cooldown/lockout) or policy gates are blocking the command. First fix: implement hardware-robust debounce and ensure the wake path lives on the always-on domain; add a clear lockout indicator (LED/beep) when commands are ignored.
→ Back to H2-3 / H2-8
Q11 Flow reading is not trustworthy — low-flow resolution or bubbles/noise?
First 2 checks: compare flow signal stability with Ipump and dT/dt correlation, then inspect noise synchrony with commutation. Discriminator: quantized steps at low flow indicate sensor resolution limits; jitter aligned with motor events indicates EMI/ground reference issues; intermittent spikes often indicate bubbles or unstable operating conditions. First fix: low-pass + synchronous sampling avoidance, add plausibility gates (flow must agree with dT/dt), and fall back to proxy evidence when flow confidence is low.
→ Back to H2-5
Q12 Worse behavior in winter — temperature-delta policy or supply sag first?
First 2 checks: capture minimum VIN during pump start and compare dT/dt under the same policy across seasons. Discriminator: VIN sag/failed starts point to UVLO margin and input impedance; stable power but slow comfort indicates winter thermal inertia is outside the tuned lag window/band. First fix: add a winter tuning profile (wider band + longer lag window) while keeping Toff_min, and harden the input/always-on rails to prevent cold-start dips.
→ Back to H2-2 / H2-8

Figure F10 FAQ-to-chapter map (each Q maps back to evidence chapters).

Figure F10 — FAQ to Chapter Map Mapping of 12 FAQ items to H2 chapters (H2-2 through H2-11) using connection lines. FAQ → Chapter Map (Evidence Backlinks) FAQs Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11 Q12 Chapters H2-2 Metrics & Constraints H2-3 Control Modes H2-4 Temp Sensing Chain H2-5 Flow/Presence Evidence H2-6 BLDC Pump Control H2-7 Protection & Faults H2-8 Power Tree & Low Power H2-9 EMC & Ruggedness H2-10 Validation Plan H2-11 Field Debug Playbook Lines show required evidence backlink(s).
Figure F10. Each FAQ is constrained to evidence chapters only (no platform creep), keeping the page mechanically verifiable.
Cite this figure — Suggested citation: ICNavigator, “Hot Water Recirculation,” Fig. F10 (FAQ → Chapter Map), 2026.
icnavigator

ICNavigator helps engineers find verified ICs, alternatives, and fast quotes.

Explore
Categories
  • Power Management
  • MCU
  • Automotive ICs
  • Sensors
  • Interface & Transceivers
  • Analog Front-End
Get in Touch
  • Email: hello@icnavigator.com
  • WeChat/WhatsApp: +86-xxxx

© 2025 ICNavigator. All rights reserved.