Electrical input metering disagrees with utility meter: where is the measurement boundary wrong?

Verify whether metering is at the AC line or at the DC link, and which loads are included (compressor inverter, pump, valves, auxiliaries). DC-link metering misses upstream PFC losses and may exclude auxiliary loads; AC-line metering includes everything, including standby and defrost-related power. Label the boundary explicitly and compare only equivalent time windows; calibrate at steady operating points.

Heat Pump & Radiant Heating Control: ΔT, Metering & Diagnostics

Q: Why is ΔT sometimes “negative” for a few minutes—sensor placement or mixing valve behavior?

Check raw Tsupply/Treturn traces and mixing valve position/command. If the ΔT sign flip aligns with valve motion or a mode transition, it is usually mixing/recirculation behavior. If it happens without valve activity and temperatures look noisy or phase-shifted, it is usually sensor contact/insulation or sampling alignment. Reduce sensor lag and time-align samples used for ΔT.

Q: Radiant loop warms slowly: check flow first or compressor capacity first—what proves it?

Start with Flow and ΔT (Tsupply−Treturn). Low/unstable flow, high pulse-loss, or poor RPM-to-Flow coherence proves a delivery problem. Stable flow with persistently tiny ΔT points to mixing dilution, sensor lag, or capacity not reaching the hydronic side. Restore circulation coherence (purge air, clear restrictions, verify valve states) before tuning capacity settings.

Q: COP estimate jumps wildly: which two signals are usually lying?

Most COP jumps come from unreliable ΔT or unreliable Flow. ΔT can lie due to phase lag, poor pipe contact, or excessive filtering; Flow can lie due to pulse loss, scaling errors, or inference drift. COP also becomes meaningless if defrost/freeze-protect windows are included without gating. Enforce COP confidence levels and repair ΔT/Flow validity before trusting the COP value.

Q: Short-cycling: is it ΔT control instability or a minimum compressor speed limit?

Compare ΔT trend to compressor speed and mixing activity. If compressor speed repeatedly clamps at its minimum while ΔT remains small, the system may be hitting a capacity floor for the current hydronic demand. If mixing position changes frequently and ΔT oscillates with strong lag/noise, instability is usually sensing/filtering and loop gain. Stabilize ΔT sensing and mixing control timing before changing compressor limits.

Q: Pump running but no heat delivered: air lock vs clogged filter—what evidence separates them?

Use Pump_RPM and Flow, and add ΔP if available. Air/cavitation often shows intermittent flow collapse with noisy ΔP spikes while RPM remains steady. A clogged strainer/filter more often shows persistently low flow with a rising ΔP trend as RPM increases. Purge air and confirm venting first; if ΔP continues to rise with stable RPM, inspect restrictions and verify pump curve compatibility.

Q: Mixing valve keeps hunting: is it sensor lag, wrong PID, or actuator deadband?

Check Tsupply response timing against mixing position changes. If the valve moves but Tsupply responds late or inconsistently, sensor lag or filtering delay dominates. If Tsupply responds promptly but overshoots and oscillates, loop gains/hold-time are likely wrong. If commands change but position evidence does not, actuator deadband, sticking, or driver faults dominate. Fix measurement delay first, then tune hold-time/hysteresis, then verify motion evidence.

Q: Compressor trips only during defrost transitions: what snapshot must be logged?

Log a trip-edge snapshot containing Vdc and Iphase in the same time window, plus compressor speed, driver fault flags, defrost state, and trip class. If Vdc sag precedes the fault, suspect margin/UVLO; if Iphase spikes with stable Vdc, suspect transient current events or driver/current-sense behavior. Enforce this snapshot at defrost entry/exit and at every trip edge.

Q: Supply temp capped for floor safety: how to prove it’s protection logic, not capacity loss?

Check the floor-safety/over-temp protection state and whether mixing control is clamped. If a protection trigger is active and mixing action is limited or pump strategy changes, the cap is intentional. If no protection state is active, the ceiling usually indicates delivery limits (Flow/ΔT) or capacity constraints. Log trigger→action mapping and exclude protected intervals when evaluating performance or COP.

Q: Flow sensor reads OK but ΔP suggests otherwise: which one to trust, how to validate fast?

Run a short Pump_RPM sweep and compare coherence, plus pulse-loss counters. If RPM changes and ΔP trends accordingly while Flow stays flat, the Flow chain is likely wrong (dropouts, scaling, installation). If Flow changes with RPM but ΔP does not respond, ΔP tapping/conditioning may be wrong or overly filtered. Establish a coherence baseline and downgrade the conflicting signal until corrected.

← Back to: Smart Home & Appliances

Heat pump radiant heating performance becomes predictable when the controller can prove delivery (Flow + ΔT), prove capacity (Compressor/Vdc/Iphase), and prove boundaries (metering + state gating) with complete logs. If any of these evidence chains is weak, COP, comfort, and fault conclusions will be misleading—so the fastest fix is to restore trustworthy sensing, stable control, and actionable snapshots.

H2-1 — Definition & boundary: Heat Pump + Radiant Hydronics

This chapter locks the scope so every later section stays evidence-based and non-overlapping: drives → ΔT/flow proof → metering/COP confidence → remote diagnostics logs.

One-sentence boundary

This page focuses on control and diagnostics for a heat pump feeding a hydronic radiant loop—compressor/valve/pump drives, supply/return ΔT and flow evidence, energy metering and COP estimation, plus remote fault logs—excluding thermostat UX, room-by-room zoning/TRVs, and cloud platform architecture.

Inside this page (must land here)

Compressor inverter drive EEV / reversing valve Circulation pump Mixing valve Supply / return ΔT Flow + pressure proof Electrical-in metering Thermal-out estimate Fault code + event log RS-485/Modbus or Ethernet

Drives: what is commanded, what feedback confirms execution, and what “trip signatures” prove a root cause.
ΔT & flow evidence: where to measure, what ranges make sense, and what failure patterns look like.
Metering: define the energy boundary (AC line vs DC link) and grade COP confidence based on sensor integrity.
Remote diagnostics: minimum snapshot bundle + counters so “remote fault” becomes actionable.

Outside this page (do not deep-dive)

Thermostat UX / automation rules (schedules, UI flows, app integration).
Room-by-room zoning, TRVs, manifold actuator design (end devices and zone logic).
Boiler/furnace flame control (combustion safety chain).
HEMS / whole-home energy orchestration (panel-level switching, solar/storage system control).
Cloud backend architecture (pipelines, dashboards, multi-tenant services).
Installation tutorials (plumbing practice, building codes, hydronic balancing procedures).

What this scope enables (reader outcomes)

Fast discrimination: separate “not enough heat” into capacity vs circulation/mixing using ΔT + flow evidence.
Trustworthy performance: compute thermal output and grade COP confidence instead of relying on misleading single sensors.
Actionable remote support: define exactly what to log so the first remote report already contains proof-quality snapshots.

Figure F1. Scope boundary map: what is inside this page vs intentionally excluded to avoid overlap.

Cite this figure Figure ID: F1 · ICNavigator

H2-2 — Reference architecture: Outdoor unit + hydraulic module + radiant loop

The goal here is not theory—it is a stable “evidence coordinate system” that every later control, metering, and diagnostics section can reference without drifting into thermostat or cloud topics.

How to read the system: four evidence paths

Power → Drive path: AC input to PFC/DC link and inverter current/voltage prove whether trips, brownouts, or ramp limits are electrical/drive-driven.
Refrigerant actuation path: compressor speed, EEV state, reversing valve state explain defrost transitions and capacity behavior without turning into a refrigeration textbook.
Hydronic delivery path: pump + mixing valve + supply/return headers, validated by ΔT/flow/pressure, prove whether heat is actually delivered to the radiant loop.
Diagnostics/log path: fault codes + snapshot bundle + counters convert “remote fault” into proof-quality evidence.

Blocks (and why each matters to control & diagnosis)

AC in → PFC → DC link → inverter: defines the electrical-in boundary and explains start/defrost transient failures (DC link sag, current spikes).
Compressor inverter drive: provides capacity; phase current + speed command are the quickest discriminators for overload vs control-limit behavior.
EEV + reversing valve: must be logged as state during capacity drops/defrost events; “water-side symptoms” can be caused upstream.
Hydraulic module (HX/buffer optional): acts as the coupling point between refrigerant-side capacity and water-side delivery stability.
Circulation pump + mixing valve: determines radiant supply temperature and ΔT stability; hunting often comes from sensor lag + actuator deadband.
Supply/return temps + flow + pressure: are the minimum set to prove circulation vs blockage/air-lock vs sensor error.
Remote interface (RS-485/Modbus or Ethernet): used to deliver evidence—fault class, snapshots, and counters—not a protocol-stack deep dive.

Interface evidence table (what to measure, what it proves)

Signal / Node	Sensor / Method	Typical range (field)	What it proves	Failure signature (common)	Logging rule
Supply temp Radiant supply pipe	NTC / PT1000 clamp or immersion well	Low-to-mid temp water for radiant floors (system-dependent)	Confirms delivered water temperature at the loop entry; used with return temp for ΔT and thermal output estimation.	Loose clamp: noisy steps; poor insulation: drift with ambient; slow response causes mixing valve hunting.	1–2 s
Return temp Radiant return	Same as supply	Usually below supply; gap depends on load and flow	Enables ΔT; rising return temp with low ΔT often indicates high flow or insufficient load pickup.	Thermal coupling errors make ΔT appear “too small/negative,” breaking COP estimation.	1–2 s
Flow rate	Hall/turbine flow sensor or inferred from ΔP	Varies by loop size and pump curve	Proves circulation actually occurs; required to distinguish “capacity issue” vs “delivery issue.”	Intermittent pulses: sticking rotor; flow reads OK but ΔP abnormal: sensor misread or bypass path.	1–2 s
ΔP / pressure Across filter/loop	Pressure sensor(s) or ΔP transducer	System-dependent; used as trend evidence	Separates clog/valve restriction/air-lock from normal flow; validates whether “flow” reading is believable.	High ΔP + low flow: blockage; normal ΔP + low flow: pump issue or closed valve; noisy pressure: cavitation/air.	2–5 s
Pump RPM / tach	EC motor tach output or driver estimate	Varies by pump type	Confirms actuator execution; essential when remote diagnosis claims “pump running.”	RPM present but flow missing → air-lock / closed valve; RPM unstable → supply/driver issue.	1 s
Mixing valve position	Stepper position or analog feedback	0–100% (normalized)	Explains supply temp shaping and hunting; validates whether control action matches measured ΔT trends.	Deadband/stiction: repeated small commands with no temp change; overshoot: sensor lag or wrong loop gain.	1–2 s
DC link voltage	Divider to ADC	Tracks input line and PFC state	Proves brownout/UVLO behavior; ties transient faults (start/defrost) to power integrity evidence.	Sag during defrost/start → resets / derating; spikes → OVP trips; ripple rise → capacitor aging.	Event snapshot (200–500 ms)
Inverter phase current	Shunt + amplifier or Hall sensor	Load-dependent	Separates “control-limited” from “hard overload”; supports OCP/root-cause classification.	Current spikes aligned with valve transitions; high RMS for same capacity → mechanical or refrigerant-side issue.	Event snapshot (200–500 ms)
Ambient / frost state	NTC + logic state	Climate-dependent	Explains defrost entry/exit and capacity derating triggers; prevents misdiagnosing “heat drop” as water-side only.	Frost sensor drift → excessive defrost; missing state in logs → remote diagnosis becomes guesswork.	5–10 s + state changes

Note: “Typical range” is intentionally described as field-dependent to avoid false precision across different loop sizes, pump curves, and regional design temperatures. The diagnostic value comes from trends and cross-checks (ΔT vs flow vs ΔP).

Figure F2. Reference architecture: power/drive, refrigerant actuation, hydronic delivery, and diagnostics/log evidence paths.

Cite this figure Figure ID: F2 · ICNavigator

H2-3 — Control loops that matter (and how they couple)

This chapter defines a practical control coordinate system: what each loop controls, what evidence proves it is behaving, and how defrost/freeze-protect states override setpoints so faults are not misclassified.

Loop selection rule (keep it evidence-based)

Every loop below is defined by a controlled variable, an actuator, and 2–3 proof signals. If a claimed “loop issue” cannot be proven by these signals, treat it as a measurement integrity or override-state problem first.

Key loops (what controls what)

Loop A — Capacity

Compressor speed loop (capacity control)

Controlled variable: delivered capacity proxy (speed/power trend), not abstract thermodynamic quantities.
Actuator: inverter → compressor (BLDC/PMSM).
Proof signals: speed command vs speed estimate; inverter phase current; DC link voltage/power trend.
Common failure signature: command rises but current/power does not → derating, protection limit, or measurement error.

Speed cmdPhase currentDC link

Loop B — Refrigerant-side actuation (high-level)

EEV control (state-driven stability)

Controlled variable: stable EEV state consistent with operating mode (heating/defrost transitions).
Actuator: EEV step/coil driver.
Proof signals: EEV position/step count; driver status; mode/defrost state bits.
Common failure signature: frequent state changes with weak system response → mechanical stiction, drive undervoltage, or missing logs.

EEV positionDriver statusMode state

Loop C — Water delivery

Pump control (flow/ΔT stability)

Controlled variable: flow (or flow-derived stability) and its impact on ΔT trend.
Actuator: EC pump speed/PWM.
Proof signals: pump RPM/tach; flow reading; ΔP/pressure trend (if available).
Common failure signature: RPM present but flow missing → air-lock, closed valve, or bypass path.

Pump RPMFlowΔP

Loop D — Radiant supply shaping

Mixing valve control (supply temperature shaping)

Controlled variable: radiant supply temperature (and/or ΔT target stability).
Actuator: mixing valve position (stepper or analog feedback).
Proof signals: valve position; supply temp response; sensor lag/filtered sampling window.
Common failure signature: valve hunting usually indicates sensor lag + deadband + loop gain mismatch, not only a “bad valve.”

Valve positionSupply tempFilter lag

Loop E — Overrides

Defrost / freeze-protect interactions (what gets overridden)

Controlled variable: system safety and survivability states (freeze avoidance, stable transitions).
Actuator overrides: compressor/pump/valves may be forced to preset actions during defrost or freeze-protect.
Proof signals: defrost/freeze-protect state bits; override reason; start/end timestamps; affected setpoints.
Common failure signature: “heat drop” during defrost misread as water-side fault when override states are not logged.

Defrost stateOverride reasonSetpoint rewrite

Loop coupling map (if X changes, check Y)

If mixing valve hunts (rapid position oscillation): check supply sensor response time (contact + insulation) and filter window; confirm valve deadband/stiction before changing gains.
If ΔT collapses while supply temperature looks “fine”: check flow (pump RPM vs flow) and return sensor integrity; verify ΔP trend to rule out bypass/air-lock.
If compressor command rises but heating performance does not: check inverter current/DC link evidence; then cross-check hydronic delivery (flow + ΔT) to separate capacity from delivery.
If frequent defrost coincides with heat delivery drops: check defrost state transitions and overridden setpoints; ensure snapshots include valve states and DC link during entry/exit.
If pump RPM is stable but ΔT is noisy or negative: suspect ΔT measurement integrity (placement/lag) before concluding “loop instability.”

Figure F3. Loop coupling map: commands, proof signals, and override states that must be logged to avoid misdiagnosis.

Cite this figure Figure ID: F3 · ICNavigator

H2-4 — ΔT capture: sensing chain, placement, accuracy, and sampling strategy

ΔT is the evidence currency for this topic. If ΔT is wrong, diagnostics and COP estimates become untrustworthy. This chapter treats ΔT as an integrity problem: sensor choice, placement, lag, and field validation steps.

Sensor options (NTC vs PT1000/RTD) — choose by failure signatures

NTC clamp sensors: cost-effective but sensitive to contact quality, self-heating, and ADC reference stability; poor contact often appears as step noise and drift.
PT1000/RTD: better linearity but more sensitive to lead resistance and installation method; incorrect wiring or contact can bias ΔT and break COP estimation.
Selection rule: prioritize the option that yields repeatable trends under insulation and contact constraints, because trend integrity matters more than a single absolute value.

Placement pitfalls (most field failures originate here)

Stratification / non-mixed points: measuring too close to mixing junctions can read blended water rather than true supply/return.
Pipe contact errors: loose clamps create noise and “teleporting” temperature steps.
Insulation missing: ambient air affects the pipe sensor; return readings often drift more, creating a fake ΔT shrink.
Response time mismatch: slow supply sensor response introduces phase lag, causing mixing valve hunting and overshoot.

Sampling & filtering (avoid lag that destabilizes control)

Over-filtering risk: smoothing reduces noise but adds delay, which destabilizes mixing valve and pump coordination.
Recommended pattern: a slow trend channel (1–2 s) for steady-state analysis plus a fast event snapshot (200–500 ms window) on defrost start/stop and trips.
Cross-check rule: if supply temp changes but return temp shows no coherent delayed response, treat ΔT as untrusted until placement/contact is verified.

Two “first measurements” to validate ΔT quality in the field

Co-located cross-check: place a second probe or contact thermometer at the same pipe point and compare trend alignment. Large divergence indicates contact/insulation issues.
Small step response test: apply a small controlled change (pump speed or mixing valve position) and observe whether supply and return temperatures respond with a reasonable delay and direction. Lack of coherent response indicates lag/placement problems.

Deliverable checklist: ΔT sanity checks + symptoms of bad ΔT

ΔT sanity checks (proof-oriented)

ΔT should not cross zero repeatedly in stable heating; frequent sign flips suggest contact/placement or filtering artifacts.
Supply and return should move coherently with return lagging; return moving “first” often indicates ambient coupling or sensor error.
ΔT trend should correlate with flow changes; if flow increases and ΔT stays random, at least one sensor chain is untrusted.
Cross-check ΔT with ΔP/pressure trend when available to rule out bypass/air-lock misreads.

Symptoms of bad ΔT (common misdiagnoses)

False “capacity loss” diagnosis when return sensor drifts or is ambient-coupled.
False “pump insufficient” conclusion when supply sensor lag compresses ΔT.
COP estimate jumping wildly due to ΔT noise dominating thermal-out calculation.
Mixing valve “unstable control” blamed when the real cause is sensor phase lag and deadband interaction.

Figure F4. ΔT integrity: correct placement and sampling prevent false diagnoses and stabilize mixing control.

Cite this figure Figure ID: F4 · ICNavigator

H2-5 — Flow & pressure sensing: proving circulation vs guessing

“Not enough heat” must be separated into a delivery problem (circulation/valves/air-lock) versus a capacity problem (compressor/derating). This chapter treats flow and pressure as proof signals that cross-check ΔT.

Circulation proof rule (ΔT + flow + pressure must agree)

A single sensor can lie. Circulation is proven only when ΔT trend, flow evidence, and pressure/ΔP evidence form a coherent story across mode transitions and pump commands.

Flow sensing options (choose by diagnostic strength)

Option A — Hall / Turbine

Best for: low-cost trending and “flow exists / not exists” proofs.
Typical artifacts: intermittent pulse dropouts from bubbles, stiction, or debris.
Proof pairing: always compare with pump RPM/tach and ΔP trend to avoid false confidence.

Pulse continuityRPM cross-checkBubble signature

Option B — Differential Pressure (ΔP inference)

Best for: blockage evidence and resistance change detection.
Typical artifacts: pump cavitation shows as noisy ΔP and unstable ΔT.
Proof pairing: combine with valve position and pump command to separate “restriction” vs “forced bypass.”

ΔP trendRestriction proofCavitation noise

Option C — Ultrasonic

Best for: stable long-term flow trending and low-maintenance sensing.
Typical artifacts: air entrainment can reduce accuracy; treat “unstable readings” as evidence of bubbles.
Proof pairing: validate against ΔT response to small pump steps (coherent response matters).

Stable trendingAir evidenceStep response

Pressure sensing (head / blockage / air-lock)

Pump head evidence: outlet pressure and ΔP across loop correlate with commanded RPM.
Blockage evidence: ΔP rises while flow drops (or inferred flow drops).
Air-lock evidence: unstable ΔP + flow pulse dropouts + ΔT coherence loss.

Head vs RPMΔP vs flowNoise patterns

Failure signatures (what the evidence looks like)

Bubbles / air-lock: flow pulses intermittently missing, ΔP noisy, ΔT becomes erratic or flips sign during transients.
Clog / restriction: ΔP rises, flow falls, ΔT can inflate (heat stays “in the water” but delivery weak) or collapses (if circulation becomes intermittent).
Stuck valve / wrong valve state: normal pump RPM with abnormal ΔP/flow correlation; supply looks “OK” but return response is delayed or absent.
Wrong pump curve / insufficient head: pump command increases but flow barely changes; ΔP stays low; ΔT can grow (low flow) while comfort stays poor.

Discriminator table (symptom → ΔT trend → flow/pressure evidence → likely cause)

Symptom	ΔT trend	Flow evidence	Pressure / ΔP evidence	Likely cause	First confirmation
Supply temp looks OK, rooms still cold	ΔT small or unstable	Flow low or intermittent; RPM may be normal	ΔP noisy or inconsistent with RPM	Air-lock / bubbles; intermittent circulation	Check air purge behavior and pulse continuity during small pump step
Heating slow; short bursts of warmth then fades	ΔT flips or drifts	Flow pulses drop out; flow reading sporadic	Pressure signal noisy; ΔP spikes then collapses	Bubble entrainment / cavitation events	Correlate events with pump RPM and valve position changes
System trips into protection when demand rises	ΔT may collapse abruptly	Flow does not increase as commanded	ΔP rises sharply at higher RPM	Restriction/clog or partially closed valve	Inspect filter/strainer state and valve commanded state
Comfort poor; pump command higher does not help	ΔT grows (low flow)	Flow barely changes with RPM	ΔP remains low even at higher RPM	Wrong pump curve / insufficient head	Compare head evidence versus expected RPM response; check bypass paths
ΔT looks “great” but feels wrong	ΔT large but inconsistent across time	Flow reading stable but mismatched with return response	ΔP contradicts flow or shows abnormal noise	Flow sensor bias or placement artifact	Cross-check with ΔP inference or secondary flow method

Note: Diagnostic strength improves when flow and ΔP are evaluated as trends against pump commands and mode states (e.g., defrost transitions).

Figure F5. Flow and pressure proof map: cross-check flow, ΔP and ΔT to confirm circulation and isolate restriction or air-lock signatures.

Cite this figure Figure ID: F5 · ICNavigator

H2-6 — Drives & power chain (only what supports control + evidence)

Drives are included only as they impact stability, trips, and remote diagnostics. Each drive block is paired with the minimum signals required to confirm trip class and isolate whether the fault is electrical, thermal, or state-driven.

Drive evidence rule (trip class must be provable)

A trip label is actionable only when it can be confirmed by a time-aligned snapshot: DC link + current sense + driver status + temperature + state bits (mode/defrost).

Compressor inverter (evidence points that matter)

Gate driver status: driver fault/UVLO flags distinguish power-stage events from control derating.
Current sense: phase current peaks and trend confirm true OCP and identify pre-trip spikes.
DC link monitoring: sag/spike evidence supports UVLO/OVP classification and correlates with resets.
Mode coupling: defrost entry/exit often changes operating constraints; snapshots must include the state bits.

Valve & actuator drives (state + execution confirmation)

EEV stepper/coil

Log: position/step count, enable, driver status (and drive current if available).
Evidence goal: distinguish mechanical stiction from undervoltage or disabled drive.

Reversing valve coil

Log: coil ON/OFF with timestamp, mode state, transition counters.
Evidence goal: correlate transitions with capacity drops and defrost cycles.

Mixing valve actuator

Log: target vs actual position (if available), supply temperature response, sensor lag marker.
Evidence goal: separate “stuck valve” from “ΔT / placement artifact” and control lag.

Pump drive (BLDC/EC)

Log: PWM/command, tach/RPM, optional power/current trend.
Evidence goal: prove execution and correlate RPM-to-flow consistency (ties back to H2-5).

Design constraints (minimal rugged note)

Surge/EFT risk: DC link spikes/sags and MCU resets can mimic “random trips” if snapshots are missing.
EMI risk: tach/flow pulses and ADC readings may glitch; use counters and plausibility checks rather than trusting a single sample.

Deliverable: trip taxonomy (OCP/OVP/UVLO/OTP) + confirmation signals

Trip class	What it looks like	Must-log signals	Fast discriminator	Typical root causes (scope-safe)
OCP	Immediate shutdown or rapid derating during load increase	Phase current peak; gate driver fault/status; timestamp; mode/defrost state	Current spike aligned to trip edge	Mechanical load anomaly, short transient, incorrect current threshold, power-stage fault
OVP	Trip coincident with voltage spike; may happen at transitions	DC link spike snapshot; driver status; line/rectified trend (if available)	DC link exceeds limit before trip	Surge event, regen transient, clamp insufficiency, measurement bias
UVLO	Reset, brownout, or forced low-power mode under demand	DC link sag; MCU reset reason; driver UVLO flags; load/current context	Voltage sag with reset reason	Weak supply, inrush, PFC limit, wiring drop, excessive load step
OTP	Gradual derating then shutdown; repeatable after warm-up	Thermal sensor(s) trend; derating state bit; fan/pump states; timestamp	Temperature crossing threshold with derating flag	Insufficient cooling, blocked airflow (unit-level), sustained overload, sensor placement error

Figure F6. Drives and power-chain evidence map: minimal signals required to classify OCP/OVP/UVLO/OTP and support remote diagnostics.

Cite this figure Figure ID: F6 · ICNavigator

H2-7 — Energy metering & COP estimation (electrical-in vs thermal-out)

Practical performance measurement requires two coherent stories: electrical input and thermal output estimate. COP is meaningful only when ΔT and flow are trustworthy and operating states (e.g., defrost) are properly labeled.

Boundary (device-side, not HEMS)

Metering is limited to equipment-side signals (AC line or DC link, ΔT, flow). Results are for engineering diagnostics and trend verification, not utility billing or whole-home energy management.

Electrical input: where to meter (AC line vs DC link)

Option A — AC line metering (system input)

What it proves: true equipment input including auxiliary loads (controls, pumps, valve drives).
Accuracy tradeoffs: PF, harmonics, phase error; low-power segments require careful sampling windows.
Best use: whole-unit trend, validation of “energy-in” across modes and transitions.

System-level inputPF / harmonic riskTrend-friendly

Option B — DC link metering (compressor-side proxy)

What it proves: inverter-side energy that correlates tightly with capacity commands and trip evidence.
Accuracy tradeoffs: excludes front-end losses and some auxiliary loads; boundary must be labeled.
Best use: capacity diagnostics, derating correlation, and trip analysis (ties to H2-6).

Capacity correlationBoundary-sensitiveTrip-friendly

Thermal output estimate: ΔT × flow × fluid heat capacity

Thermal-out estimate: heat output can be approximated from ΔT and flow with an assumed Cp for the working fluid.
Calibration note: fluid mixture and sensor scaling create systematic offsets; trend coherence matters more than a single absolute number.
Sampling note: ΔT and flow must be evaluated in the same time window; asynchronous sampling creates artificial COP spikes.
State note: defrost/freeze-protect segments must be labeled or excluded to avoid mixing incompatible regimes.

COP estimate: when it’s meaningful, when it lies

Meaningful Engineering COP trend

ΔT passes sanity checks (stable sign, coherent supply/return dynamics).
Flow is consistent with pump RPM and/or ΔP evidence (no pulse loss patterns).
Metering boundary is explicit (AC-input COP vs DC-link proxy).

Lies Common false COP patterns

Bad ΔT: poor contact/insulation or heavy filtering causes lag and oscillation artifacts.
Bad flow: bubbles, stiction, or bypass paths break coherence between RPM, ΔP and flow.
State mixing: defrost segments included without labeling corrupt averages and trends.

Deliverable: COP confidence checklist (High / Medium / Low)

Confidence level	Required conditions	Automatic downgrade triggers	How to use the COP number
High	ΔT sanity checks pass; flow correlates with pump RPM and/or ΔP; defrost state filtered or labeled; metering boundary explicit (AC or DC link).	Flow pulse loss, ΔT sign flipping, sensor error flags, or unlabeled defrost segments.	Use for performance trend, comparative commissioning checks, and remote verification of stability changes.
Medium	ΔT and flow mostly stable but partial evidence missing (e.g., no ΔP); DC link metering used with boundary label; occasional missing samples tolerated.	Increasing noise metrics, rising retry counters, or inconsistent RPM-to-flow response.	Use for trend direction only; avoid absolute claims; prioritize additional proof signals in snapshots.
Low	ΔT or flow not trustworthy (noise, drift, dropouts); defrost state not available; sampling windows not aligned; metering boundary unclear.	Frequent invalid flags; counters indicate missing/CRC errors; state mixing suspected.	Do not interpret COP; first repair measurement integrity (ΔT/flow) and logging schema (H2-8).

Figure F7. COP evidence map: choose a metering boundary (AC vs DC link), estimate thermal-out from ΔT and flow, and attach confidence levels.

Cite this figure Figure ID: F7 · ICNavigator

H2-8 — Remote diagnostics: what to log so you can actually debug

Remote diagnostics works only when logs contain time-aligned proof signals. This chapter defines an event-first logging model with snapshot bundles, counters, and communication reliability evidence—without drifting into platform architecture.

Remote diagnostics rule (event + snapshot + counters)

Every actionable fault must be supported by timestamped events, a snapshot bundle captured at the fault edge, and counters that turn intermittent issues into measurable evidence.

Event log structure (make faults self-explanatory)

Required fields: timestamp, sequence ID, fault code, trip class, operating state (heating/defrost/freeze-protect), snapshot ID.
Why sequence ID matters: reveals drops, reordering, and overwrites that otherwise look like “random missing data.”
Why state bits matter: defrost and protection overrides change setpoints; events without state are not diagnosable.

Snapshot bundle (minimum useful set)

Hydronic delivery proof

Tsupply, Treturn, ΔT
Flow
Pump RPM (tach)
Mixing valve position

Capacity & electrical proof

Compressor speed
Inverter current (Iphase)
DC link voltage (Vdc)

Environment & state proof

Pressures / ΔP (if available)
Ambient / frost state
Defrost state and transition markers

Validity flags (stop bad inference)

ΔT_valid, Flow_valid
Sensor CRC / error flags
Pulse-loss / missing-sample indicators

Counters (turn intermittent faults into statistics)

Defrost counters: defrost count, defrost duration histogram (if feasible), entry/exit timestamps.
Trip counters: count by class (OCP/OVP/UVLO/OTP), plus “last trip timestamp” per class.
Retry counters: restart attempts, compressor retry, comm reconnect.
Sensor integrity: ADC out-of-range count, CRC error count, flow pulse-loss count.

Communications (RS-485/Modbus or Ethernet) — reliability evidence only

RS-485 / Modbus evidence

CRC error count, frame timeout count, retry count
bus idle time anomalies, re-init count
sequence gaps (if events are polled)

Ethernet evidence

link up/down count, reconnect count
socket reset / keepalive timeout
packet drop indicator (if available)

Deliverable: minimum viable log schema (MVLS)

Event record: timestamp · sequence_id · fault_code · trip_class · mode_state · snapshot_id · comm_status_summary (optional)
Snapshot record: Tsupply · Treturn · ΔT · Flow · Pump_RPM · MixingValve_Pos · Compressor_Speed · Vdc · Iphase · Pressure/ΔP (opt) · Ambient/Frost · Defrost_State · validity_flags
Counters: defrost_count · trip_count_by_class · reboot_count · retry_count · comm_crc_err · comm_timeout · comm_retry · link_flap · flow_pulse_loss · sensor_crc_err
Trigger strategy: on_trip · on_defrost_enter/exit · on_flow_drop/pulse_loss · periodic baseline snapshot

Figure F8. Minimum viable log pipeline: event-first records with snapshot bundles, counters, and comm reliability evidence for remote debugging.

Cite this figure Figure ID: F8 · ICNavigator

H2-9 — Protection & safety logic inside THIS controller (keep it local)

Protection is actionable only when it is expressed as controller-local triggers, deterministic actions, and mandatory proof logs. This section keeps safety tied to heat pump + hydronics behavior, not certification theory.

Local safety scope (what “local” means)

Local protection reacts using available sensors (ΔT, flow, pressures/ΔP, Vdc, Iphase, ambient/frost, state bits) and local actuators (compressor inhibit, pump force, mixing valve clamp, alarm/event logging).

Freeze protection (conditions → actions)

Trigger candidates: low ambient/frost state, low Tsupply/Treturn, or persistent near-zero ΔT while flow exists.
Primary actions: force pump to circulate, clamp mixing valve to a safe position, and mark state as freeze-protect.
Escalation: if flow is absent or sensors invalid, upgrade to alarm + lockout with evidence snapshot.

Over-temp / floor safety (mixing valve + pump response)

Trigger candidates: Tsupply above floor safety limit or rapid temperature rise with stable pump command.
Primary actions: clamp mixing valve to reduce supply temperature; keep pump running to remove residual heat.
Stability note: apply hysteresis/hold time to avoid oscillation between comfort control and protection.

Dry-run / no-flow protection (prove circulation)

Trigger candidates: pump RPM present but flow ≈ 0 (or pulse-loss pattern), or ΔP signature indicates blockage/air lock.
Primary actions: stop or limit compressor demand, force pump purge window, and capture “no-flow” proof bundle.
Discriminator: differentiate “true no-flow” vs “flow sensor failure” using RPM↔ΔP↔ΔT coherence.

High-pressure / low-pressure trip evidence mapping

HP High-pressure evidence

Pressure rising trend into limit + compressor speed/current correlation.
State bits (heating/defrost) captured to avoid mislabeling transitions.
Action: compressor inhibit + fault event + retry policy with lockout timer.

LP Low-pressure evidence

Pressure falling/unstable + capacity command present + mode state captured.
Check sensor validity flags and sampling window alignment.
Action: compressor inhibit + fault event + conditional restart after stabilization.

Leak / abnormal ΔP detection (hydronic)

Abnormal ΔP: ΔP increases while flow does not increase → likely restriction/clog; ΔP noise spikes → bubbles/cavitation patterns.
Leak tendency (local evidence): pressure decay after pump off, unexpected refill behavior, or persistent mismatch between RPM and delivery evidence.
Action goal: downgrade capacity demand and preserve evidence snapshots; avoid “silent failure” without logs.

Deliverable: Protection action matrix (trigger → action → what you must log)

Trigger (local condition)	Immediate action (actuators)	Lockout / recovery rule	Must-log bundle (snapshot + counters)	Why it matters here
Freeze risk: ambient/frost low + Tsupply/Treturn below threshold	Force pump · clamp mixing valve · set Freeze-Protect state	Exit only after temp margin holds for N minutes; prevent rapid toggling	Tsupply/Treturn/ΔT · Flow · Pump_RPM · Mixing_Pos · Ambient/Frost · Mode_State · defrost_count	Prevents freezing while preserving proof signals for remote verification
Over-temp: Tsupply above safety limit or rapid rise	Clamp mixing valve down · keep pump running · reduce compressor demand	Hysteresis + minimum hold time; log exit condition	Tsupply/Treturn/ΔT · Flow · Pump_RPM · Mixing_Pos · Compressor_Speed · Vdc/Iphase	Floor safety depends on valve+flow coupling, not a single sensor
No-flow / dry-run: Pump_RPM present but Flow≈0 or pulse-loss	Inhibit compressor · purge/force pump window · raise alarm event	Retry limited; lockout escalates if repeated within time window	Flow · pulse_loss_count · Pump_RPM · ΔP/pressure (opt) · ΔT trend · trip_count_by_class	Separates delivery failure from capacity failure and prevents damage
HP trip: pressure rising into limit with correlated Iphase	Compressor inhibit · fault event · controlled restart policy	Restart only after stabilization; backoff timers; class-based counters	Pressure · Compressor_Speed · Iphase · Vdc · Mode_State · fault_code · sequence_id	Evidence mapping avoids false blame on ΔT/flow when root is capacity-side
LP trip: pressure falling/unstable with demand present	Compressor inhibit · fault event · conditional restart	Require stable pressure window; invalidate COP during LP fault	Pressure · Compressor_Speed · Vdc/Iphase · validity_flags · retry_count · fault_code	Prevents misleading COP and supports remote root-cause narrowing
Abnormal ΔP: ΔP↑ while flow not rising; or ΔP noise spikes	Reduce capacity demand · keep circulation as needed · log anomaly	Clear only when coherence returns (RPM↔flow↔ΔP) for N samples	ΔP/pressure · Flow · Pump_RPM · Tsupply/Treturn · comm_crc_err (if relevant)	Hydronic restrictions and air patterns are diagnosable only with ΔP evidence

Figure F9. Protection map: local triggers feed safety logic that commands compressor/pump/valve actions and enforces must-log evidence bundles.

Cite this figure Figure ID: F9 · ICNavigator

H2-10 — IC/function selection map (with example part categories)

The selection map is organized by function blocks and the evidence/control needs of this controller. It provides procurement-ready requirements without turning into a brand catalog.

Selection principles (tied to this page’s evidence chain)

Diagnosability first: time-aligned snapshots (ΔT/flow/pressure + Vdc/Iphase + state bits) must be feasible.
Stability first: sensing latency/noise influences mixing valve control and COP confidence.
Recoverability first: log memory write strategy determines whether remote debugging is possible after trips/resets.

Deliverable: compact selection table (category-level MPN types)

Function block	Key specs (what to require)	Why it matters here	Example MPN types (category-level)
MCU / Control SoC	motor-control PWM resources; multi-channel ADC; fast capture window for events; safety IO (fault pins, watchdog); comm interfaces (RS-485 UART, Ethernet MAC/PHY support).	Enables MVLS snapshots and protection actions without timing drift; supports state bits and counters at fault edges.	Motor-control MCU class · Industrial MCU class · MCU + external ADC architecture
Gate driver (inverter)	robust fault reporting (OC/UVLO/desat class); compatible with power stage; fast fault response; clear fault pin/status readability.	Trip taxonomy requires proof (driver fault + current + Vdc) to distinguish real power-stage faults from control derates.	3-phase gate driver family · Half-bridge driver family · Intelligent power module driver class
Current sense path	shunt/Hall options; bandwidth for trip capture; dynamic range; noise immunity; optional isolation where needed.	OCP classification and “spike vs sustained” evidence depends on capture fidelity around trip edges.	Shunt amplifier family · Hall current sensor IC class · Isolated current sense amplifier class
DC link monitoring	stable divider/reference; ADC range and sampling strategy; ripple/undervoltage event capture capability.	UVLO/OVP evidence and correlation with compressor demand is only credible with Vdc snapshots.	High-voltage ADC front-end class · Precision reference family · Supervisor/monitor IC class
Energy metering (optional AC input)	real power + PF handling; harmonic tolerance; line-voltage/current sensing interfaces; calibration support.	Enables AC-boundary COP and system-level energy-in trend; avoids ambiguous “DC-only” energy claims.	Single/3-phase energy metering IC family · Isolated metering AFE class
Temp sensor interface (NTC/RTD)	low-noise excitation/reference; wiring error detect; linearization support; sampling/filter knobs for loop stability.	ΔT quality controls COP validity and mixing valve stability; bad ΔT creates false diagnostics and oscillations.	RTD interface IC family · Precision ADC + front-end class · NTC AFE class
Pressure / ΔP conditioning	ratiometric conditioning; protection against transients; noise filtering without excessive lag; diagnostics flags.	Abnormal ΔP/leak evidence depends on stable trends; laggy filters hide bubbles/clog signatures.	Instrumentation amplifier class · Sensor AFE class · Pressure interface ADC family
EEV stepper / coil drive	step control + stall/oc detect (if available); position tracking; fault reporting; thermal robustness.	Remote debugging needs “command vs position vs effect” evidence to separate actuator failure from sensing errors.	Stepper motor driver family · Coil driver / smart switch class
Mixing valve actuator drive	H-bridge/stepper/servo interface (by actuator type); position feedback capture; fault detect.	Over-temp protection and supply shaping rely on deterministic valve response and measurable position evidence.	H-bridge driver family · Stepper/actuator driver class · Position feedback interface class
Pump drive interface	PWM/analog command support; tach capture; optional current/power trend; EMC-hardened input filtering.	No-flow discrimination requires RPM evidence and coherence with flow/ΔP; without tach the diagnosis collapses.	BLDC/EC motor controller class · Tach capture timer + driver interface class
Nonvolatile memory for logs	endurance; power-fail robustness; ring-buffer strategy; write amplification control; fast commit for events.	MVLS needs reliable preservation of “trip edge” snapshots; without robust NVM, remote debugging fails after resets.	FRAM family · SPI NOR flash class · EEPROM class (limited endurance) · Wear-leveling strategy
RS-485 / Ethernet physical layer (optional)	robust transceiver; ESD protection capability; error counters visibility; link status reporting.	Remote diagnostics requires reliability evidence (CRC/timeouts/reconnect) to separate comm faults from system faults.	RS-485 transceiver family · Ethernet PHY class · Isolated transceiver class

Note: Example MPN types are intentionally category-level. If concrete part numbers are needed, request “List concrete MPNs by function block”.

Figure F10. Function selection map: sensing AFEs feed the MCU for control + snapshots; logs persist to NVM; actuator drives execute protection actions and stable control.

Cite this figure Figure ID: F10 · ICNavigator

H2-11 — Validation & commissioning test plan (bench → loop → field)

A repeatable commissioning plan must prove control integrity, hydronic delivery, metering credibility, and remote diagnosability using pass/fail signals.

Staging model

Bench: rails + sensing + actuators Loop: ΔT/flow/mixing stability Field: transitions + robustness + logs Always: fault injection → log completeness

Each stage uses a checklist and acceptance metrics tied to controller-visible signals (Tsupply/Treturn/ΔT, Flow, Pump_RPM, Mixing_Pos, Compressor_Speed, Vdc, Iphase, State bits, Fault codes, Counters).

Bench bring-up (prove rails, sensing, and actuation)

B1Power rails & reset integrity

Measure: Vdc + low-voltage rails (3.3V/5V as applicable), reset reason, reboot counter.
Pass: no reset loops; rail dips do not coincide with false trips; reboot_count stays stable.
Fail signals: Vdc sag events without load change; repeated reset_reason patterns; log gaps.

B2Inverter current sense calibration

Measure: Iphase offset/gain at known points; trip-edge capture window; driver fault pin status.
Pass: stable offset; gain error within target; trip-edge snapshots include Vdc + Iphase + fault pin.
Fail signals: drifting offset; clipped peaks; inconsistent Iphase vs commanded compressor speed.

B3Valve actuation tests (EEV / reversing / mixing)

Measure: command vs position/step count (if available), coil drive status, response time trend.
Pass: position changes are observable; faults are latched with event snapshots.
Fail signals: command issued but no movement evidence; frequent coil faults without current proof.

B4Pump interface validation

Measure: Pump_RPM tach capture; Flow response; ΔP/pressure trend (if available).
Pass: RPM ↑ leads to Flow ↑ and ΔP trend coherence; pulse-loss counter remains low.
Fail signals: RPM present but Flow≈0; ΔP noise spikes; tach dropout under EMI events.

Loop tests (prove ΔT/flow/mixing stability and transitions)

ΔT step response: apply mixing valve step or capacity step → validate Tsupply/Treturn alignment, ΔT lag, and noise floor.
Flow sweep: sweep Pump_RPM across range → verify Flow and ΔP coherence; identify air lock/clog patterns.
Mixing stability: hold supply target → evaluate overshoot, oscillation, valve activity rate, and settle time.
Freeze/defrost transitions: enter/exit transitions → ensure actions match protection matrix and logs include mode/state bits.

Metering tests (protect COP credibility)

Known operating points: compare electrical-in trend vs thermal-out trend at several stable points (avoid chasing absolute COP only).
Offset checks: verify ΔT offset and Flow scaling; confirm “COP confidence” downgrades when sensors become unreliable.
Sanity cases: introduce pulse loss or ΔT lag → confirm COP is flagged low-confidence rather than reported as a clean number.

Remote diagnostics tests (fault injection → log completeness verification)

Inject no-flow: force Flow≈0 with RPM present → event must include Flow, Pump_RPM, ΔP (if), ΔT, counters.
Inject over-temp: exceed Tsupply threshold → event must include Tsupply/Treturn, Mixing_Pos, mode/state, action taken.
Inject DC link sag: reduce Vdc margin → event must include Vdc, Iphase, compressor speed, trip class UVLO/OVP.
Inject comm errors: raise CRC/timeouts → logs must prove comm failure vs missing data generation (sequence gaps).

Deliverable: staged checklist + acceptance metrics (pass/fail signals)

Stage	Step (what to do)	Stimulus	Measure (proof signals)	Pass/Fail metric	If fail, first suspect
Bench	Rail & reset integrity	Power cycle + load step	Vdc + LV rails + reset_reason + reboot_count	Pass: no reset loops; rail dips do not align with false trips	Supervisor thresholds / wiring / ground reference / sampling window
Bench	Current sense calibration	Known current points + trip-edge capture	Iphase offset/gain + Vdc + driver fault pin	Pass: stable offset; peaks captured without clipping; consistent correlation	Shunt amp bandwidth / ADC range / anti-alias filter / layout coupling
Loop	ΔT step response	Mixing valve step or capacity step	Tsupply/Treturn time series + ΔT lag/noise	Pass: bounded lag; no sign flips; noise within target	Sensor placement/insulation / filtering too heavy / sampling misalignment
Loop	Flow sweep coherence	Pump_RPM sweep	Flow + Pump_RPM + ΔP trend + pulse_loss_count	Pass: RPM↔Flow↔ΔP coherence; low pulse loss	Air lock / clog / wrong pump curve / flow sensor dropout
Field	Freeze/defrost transitions	Transition entry/exit scenarios	Mode/Defrost/Frost bits + actions + snapshots	Pass: correct action sequence; logs complete and ordered (sequence_id)	State machine gaps / missing log triggers / hysteresis too small
Any	Fault injection log completeness	No-flow / over-temp / Vdc sag / comm CRC	Fault code + trip class + snapshot bundle + counters	Pass: evidence fields present; counters increment; sequence continuity	Schema missing fields / snapshot trigger missing / buffer overwrite policy

Concrete MPN examples (controller-relevant blocks used by this test plan)

The following part numbers are reference examples for selection and validation; equivalents are acceptable when key specs match. Choices depend on voltage class, isolation needs, and the targeted compressor/pump power level.

Block	Why it is needed in H2-11/H2-12	Concrete MPN examples (non-exhaustive)
Motor-control MCU	PWM/timers + ADC capture windows + fault pins + logging orchestration	ST STM32G4 series · TI C2000 F2837x series · Microchip SAM E70 series
3-phase gate driver	Inverter trip proof (fault pins) and deterministic shut-down behavior	TI DRV8323 · Infineon 6EDL04I06NT · onsemi NCV7725 (driver class reference)
Shunt current sense amplifier	Accurate Iphase evidence near trip edges; avoids false OCP claims	TI INA240 · ADI AD8418A · TI INA181
Hall current sensor IC/module	Galvanic isolation option and robust current trending for diagnostics	Allegro ACS758 · Allegro ACS70331 · LEM HO series (module family)
Energy metering (AC input)	Electrical-in credibility for COP confidence and commissioning reports	ADI ADE9153A · TI AMC3302 (isolated modulator class) · Renesas RAA211x (metering class reference)
RTD/temperature interface	ΔT trust: stable excitation + linearization + diagnostics flags	ADI MAX31865 · TI ADS124S08 (precision ADC) · ADI ADT7310 (temp sensor option)
Pressure / ΔP analog front-end	Hydronic restriction/air patterns: ΔP trend evidence without excessive lag	TI INA826 (in-amp class) · ADI AD8421 · TI OPA333 (low-drift op-amp)
EEV / stepper driver	Command→movement evidence; fault reporting for actuator isolation	TI DRV8846 · ST L6470 · Allegro A4988 (stepper class reference)
Solenoid / coil driver	Reversing valve / solenoid actuation with open/short detection options	TI DRV103 · Infineon BTS500xx (smart high-side family) · ST VNQ/VN5E families
RS-485 transceiver	Remote diagnostics reliability evidence (CRC/timeouts) depends on robust PHY	TI THVD1450 · Analog Devices LTC2862 · Maxim/ADI MAX13487E
Ethernet PHY	Link up/down evidence and stable field connectivity	Microchip LAN8720A · TI DP83825I · Microchip KSZ8081
Supervisor / watchdog	Proves resets and prevents silent lock-ups that erase evidence	TI TPS3823 · Maxim/ADI MAX706 · Microchip MCP131
Log memory (NVM)	Preserves trip-edge snapshots for field root-cause	Fujitsu MB85RS64V (FRAM) · Cypress/Infineon S25FL (SPI NOR family) · Microchip 25AA1024 (EEPROM class)

Figure F11. Staged validation pipeline with pass/fail metrics and remote-log completeness checks.

Cite this figure Figure ID: F11 · ICNavigator

H2-12 — Field debug playbook (symptom → evidence → isolate → first fix)

Fast field value comes from forcing each symptom into an evidence chain: first 2 measurements → discriminator → first fix. Each item also lists the minimal must-log fields required for remote reproduction.

Rules for evidence-driven debugging

Start with delivery evidence: Flow + ΔT outrank comfort impressions.
Separate delivery vs capacity: Pump_RPM/ΔP coherence distinguishes circulation faults from compressor-side faults.
Demand logs before conclusions: if snapshots are missing, treat it as an observability fault and fix logging first.

S1 “Supply temp looks OK, rooms still cold”

First 2 measurements: Flow + ΔT (Tsupply − Treturn).

Discriminator: Flow present but ΔT very small → mixing over-dilution / too much flow / ΔT lag; Flow weak or pulse-loss → delivery failure (air lock, clog, closed valve).

First fix: verify RPM↔Flow↔ΔP coherence; purge air / confirm valve open state; reduce filter lag on temp chain if ΔT shows unrealistic phase shift.

Must-log: Tsupply/Treturn/ΔT, Flow, Pump_RPM, ΔP/pressure(if), Mixing_Pos, validity_flags.

S2 “Short-cycling (frequent on/off)”

First 2 measurements: ΔT trend + compressor speed (or run-time window).

Discriminator: ΔT near zero with high flow → capacity not transferring to slab (mixing/flow mismatch); ΔT oscillates with large lag → temp sensing/filtering causing control instability.

First fix: run ΔT sanity checks (sign stability, lag window); tune mixing control hysteresis/hold time; adjust pump command to keep ΔT in a stable band.

Must-log: Tsupply/Treturn/ΔT, Flow, Pump_RPM, Mixing_Pos, Compressor_Speed, mode_state.

S3 “Compressor trips on cold days”

First 2 measurements: Vdc sag + Iphase peak (trip-edge window).

Discriminator: trips coincide with defrost transitions/state changes → transient management; Vdc sag without state transition → power margin/UVLO thresholds; Iphase spikes with stable Vdc → current-sense/driver behavior.

First fix: enforce trip-edge snapshot capture (Vdc, Iphase, driver fault); adjust ramp limits during transitions; validate UVLO/OCP thresholds against real waveforms.

Must-log: Vdc, Iphase, Compressor_Speed, fault_code, trip_class, mode/defrost state, sequence_id, trip counters.

S4 “Frequent defrost / heat drops”

First 2 measurements: frost/defrost state evidence + valve position (EEV/Mixing where available).

Discriminator: defrost triggered without matching frost evidence → sensor/threshold issue; defrost triggered with evidence but delivery collapses → hydronic side not buffered (flow/mixing response).

First fix: log defrost entry/exit snapshots; verify transitions do not override pump/mixing into no-delivery; adjust state gating so COP and comfort are not computed across mixed modes.

Must-log: frost_state, defrost_state, Tsupply/Treturn/ΔT, Flow, Pump_RPM, valve positions, counters.

S5 “COP looks great/bad but feels wrong”

First 2 measurements: COP confidence inputs: (ΔT validity + Flow validity) plus state bits (defrost/freeze-protect).

Discriminator: ΔT or Flow invalid / pulse-loss present → COP is not trustworthy; state mixing (defrost windows included) → COP polluted.

First fix: enforce COP confidence gating (High/Medium/Low); fix ΔT placement/insulation and flow measurement coherence before interpreting COP.

Must-log: ΔT, Flow, validity_flags, mode/defrost state, electrical-in boundary marker (AC vs DC link if used).

S6 “Remote faults but no actionable data”

First 2 measurements: snapshot bundle completeness + sequence_id continuity.

Discriminator: missing fields with continuous sequence → schema gap; sequence gaps + CRC/timeouts rising → communication loss; continuous comm but no new events → event trigger missing.

First fix: implement minimum viable log schema: fault_code + trip_class + snapshot_id + counters; add triggers at trip edges and state transitions; protect ring buffer from overwrite of critical events.

Must-log: fault_code, trip_class, snapshot fields, sequence_id, comm_crc_err/timeout/retry counters.

S7 “Supply temperature oscillates; valve ‘hunts’”

First 2 measurements: Tsupply trend + Mixing_Pos activity rate.

Discriminator: Tsupply oscillation with slow ΔT sensors → measurement lag; oscillation with stable sensors → loop gain/hysteresis/hold-time mismatch.

First fix: reduce excessive filtering delay; add minimum hold time on valve moves; validate ΔT sampling alignment to prevent control chasing stale data.

Must-log: Tsupply/Treturn/ΔT, Mixing_Pos, Flow, Pump_RPM, validity flags.

S8 “Pump runs, but heat delivery collapses intermittently”

First 2 measurements: Pump_RPM + Flow (plus ΔP if available).

Discriminator: RPM steady but Flow drops with ΔP spikes → air/cavitation; RPM steady but Flow drops with ΔP rising trend → clog/filter restriction; Flow signal drops while ΔP/RPM coherent → flow sensor dropout.

First fix: purge air and verify filter/strainer; use coherence checks to isolate sensor failure; add pulse-loss counter thresholds to trigger “no-flow” protection with evidence snapshots.

Must-log: Flow, Pump_RPM, ΔP/pressure(if), pulse_loss_count, ΔT trend, event timestamps.

Figure F12. Field decision map: symptoms route into delivery/capacity/sensor/observability branches using minimal proof signals.

Cite this figure Figure ID: F12 · ICNavigator

Request a Quote

Name

Company

Part Number(s) / BOM

Quantity & Target Lead Time

Alternates Allowed

Temperature Grade

Package / Footprint

Compliance

Budget Window

Lot Size / Qty

Message

Attachment

Accepted Formats

pdf, csv, xls, xlsx, zip

Attachment

Drag & drop files here or use the button below.

H2-13 — FAQs (evidence-anchored, controller-local)

Each answer lands back on measurable signals: ΔT, Flow/ΔP, drives (Vdc/Iphase), metering boundary, and log completeness. No thermostat UX, no zoning/TRVs, no cloud platform design.

FAQ 01 Why is ΔT sometimes “negative” for a few minutes—sensor placement or mixing valve behavior?

First 2 measurements: raw Tsupply/Treturn traces + Mixing_Pos (or mixing command). Discriminator: if ΔT sign flips coincide with mixing moves or mode transitions, it is usually mixing/recirculation behavior; if sign flips occur with no valve activity and temps look noisy or phase-shifted, it is usually sensor contact/insulation or sampling alignment. First fix: reduce temperature chain lag, improve pipe coupling, and time-align samples used for ΔT.

Read more: H2-4 ΔT capture H2-3 Control loops

FAQ 02 Radiant loop warms slowly: check flow first or compressor capacity first—what proves it?

First 2 measurements: Flow + ΔT (Tsupply−Treturn). Discriminator: low/unstable Flow, high pulse-loss, or incoherent RPM↔Flow trends prove a delivery problem; stable Flow with persistently tiny ΔT points to mixing dilution, sensor lag, or capacity not reaching the hydronic side. First fix: restore circulation coherence (purge air, check restrictions, verify valve open state) before tuning compressor or mixing targets.

Read more: H2-5 Flow & pressure H2-3 Control loops

FAQ 03 COP estimate jumps wildly: which two signals are usually lying?

First 2 measurements: ΔT validity (noise/lag) + Flow validity (pulse_loss_count / coherence). Discriminator: COP is unreliable when either ΔT is phase-lagged/unstable or Flow is inferred incorrectly or dropping pulses; COP also becomes meaningless if defrost/freeze-protect windows are included without gating. First fix: enforce COP confidence levels (High/Medium/Low) and exclude mixed-mode intervals; repair ΔT placement and Flow coherence before interpreting COP numbers.

Read more: H2-7 Metering & COP H2-4 ΔT capture

FAQ 04 Short-cycling: is it ΔT control instability or a minimum compressor speed limit?

First 2 measurements: ΔT trend + Compressor_Speed (or run-time window). Discriminator: if compressor speed repeatedly clamps at its minimum while ΔT remains small, the system may be hitting a capacity floor for the current hydronic demand; if Mixing_Pos activity is high and ΔT oscillates with strong lag/noise, instability is usually driven by sensing/filtering and loop gains. First fix: stabilize ΔT sensing and mixing control hold-time before changing compressor limits.

Read more: H2-3 Control loops H2-4 ΔT capture H2-6 Drives & trips

FAQ 05 Pump running but no heat delivered: air lock vs clogged filter—what evidence separates them?

First 2 measurements: Pump_RPM + Flow (add ΔP if available). Discriminator: air/cavitation often shows intermittent Flow collapse with noisy ΔP spikes while RPM is stable; a clogged strainer/filter more often shows persistently low Flow with a rising ΔP trend as RPM increases. First fix: purge air and verify venting first; if ΔP continues to rise with stable RPM, inspect restrictions and confirm the pump curve matches the loop.

Heat Pump & Radiant Heating Control: ΔT, Metering & Diagnostics

Heat Pump & Radiant Heating Control: ΔT, Metering & Diagnostics

H2-1 — Definition & boundary: Heat Pump + Radiant Hydronics

One-sentence boundary

What this scope enables (reader outcomes)

H2-2 — Reference architecture: Outdoor unit + hydraulic module + radiant loop

How to read the system: four evidence paths

Blocks (and why each matters to control & diagnosis)

Interface evidence table (what to measure, what it proves)

H2-3 — Control loops that matter (and how they couple)

Loop selection rule (keep it evidence-based)

Key loops (what controls what)

Loop coupling map (if X changes, check Y)

H2-4 — ΔT capture: sensing chain, placement, accuracy, and sampling strategy

Sensor options (NTC vs PT1000/RTD) — choose by failure signatures

Placement pitfalls (most field failures originate here)

Sampling & filtering (avoid lag that destabilizes control)

Two “first measurements” to validate ΔT quality in the field

Deliverable checklist: ΔT sanity checks + symptoms of bad ΔT

H2-5 — Flow & pressure sensing: proving circulation vs guessing

Circulation proof rule (ΔT + flow + pressure must agree)

Flow sensing options (choose by diagnostic strength)

Failure signatures (what the evidence looks like)

Discriminator table (symptom → ΔT trend → flow/pressure evidence → likely cause)

H2-6 — Drives & power chain (only what supports control + evidence)

Drive evidence rule (trip class must be provable)

Compressor inverter (evidence points that matter)

Valve & actuator drives (state + execution confirmation)

Design constraints (minimal rugged note)

Deliverable: trip taxonomy (OCP/OVP/UVLO/OTP) + confirmation signals

H2-7 — Energy metering & COP estimation (electrical-in vs thermal-out)

Boundary (device-side, not HEMS)

Electrical input: where to meter (AC line vs DC link)

Thermal output estimate: ΔT × flow × fluid heat capacity

COP estimate: when it’s meaningful, when it lies

Deliverable: COP confidence checklist (High / Medium / Low)

H2-8 — Remote diagnostics: what to log so you can actually debug

Remote diagnostics rule (event + snapshot + counters)

Event log structure (make faults self-explanatory)

Snapshot bundle (minimum useful set)

Counters (turn intermittent faults into statistics)

Communications (RS-485/Modbus or Ethernet) — reliability evidence only

Deliverable: minimum viable log schema (MVLS)

H2-9 — Protection & safety logic inside THIS controller (keep it local)

Local safety scope (what “local” means)

Freeze protection (conditions → actions)

Over-temp / floor safety (mixing valve + pump response)

Dry-run / no-flow protection (prove circulation)

High-pressure / low-pressure trip evidence mapping

Leak / abnormal ΔP detection (hydronic)

Deliverable: Protection action matrix (trigger → action → what you must log)

H2-10 — IC/function selection map (with example part categories)

Selection principles (tied to this page’s evidence chain)

Deliverable: compact selection table (category-level MPN types)

H2-11 — Validation & commissioning test plan (bench → loop → field)

Staging model

Bench bring-up (prove rails, sensing, and actuation)

Loop tests (prove ΔT/flow/mixing stability and transitions)

Metering tests (protect COP credibility)

Remote diagnostics tests (fault injection → log completeness verification)

Deliverable: staged checklist + acceptance metrics (pass/fail signals)

Concrete MPN examples (controller-relevant blocks used by this test plan)

H2-12 — Field debug playbook (symptom → evidence → isolate → first fix)

Rules for evidence-driven debugging

S1 “Supply temp looks OK, rooms still cold”

S2 “Short-cycling (frequent on/off)”

S3 “Compressor trips on cold days”

S4 “Frequent defrost / heat drops”

S5 “COP looks great/bad but feels wrong”

S6 “Remote faults but no actionable data”

S7 “Supply temperature oscillates; valve ‘hunts’”

S8 “Pump runs, but heat delivery collapses intermittently”

Recommended topics you might also need

Request a Quote

Accepted Formats

Attachment

H2-13 — FAQs (evidence-anchored, controller-local)

Explore

Categories

Get in Touch