Dishwasher Control: Pumps, Sensor AFEs, Heater & Door Interlock
← Back to: Smart Home & Appliances
A dishwasher control system coordinates circulation/drain pumps, turbidity/temperature/level sensing, heater power, and door interlocks under noisy mains and humid conditions.
This page shows how to prove faults with minimal measurements (current, raw sensor counts, and event logs) and then isolate whether the root cause is mechanical flow, sensing/AFE noise pickup, safety gating, or rugged power/EMC coupling.
Dishwasher Control: Featured Answer & Page Boundary
This page focuses on the dishwasher control board’s four coupled chains—pumps, sensing AFEs, heater safety, and door interlock—under a noisy mains environment where load switching can corrupt measurements or reset logic.
45–55 word extractable definition
Dishwasher control is the coordinated drive of circulation and drain pumps plus robust sensing (turbidity, temperature, water level/leak) to regulate wash cycles while enforcing heater and door-interlock safety. The design must remain accurate and fail-safe under mains noise from motor PWM and relay/triac switching that can disturb ADC readings or trigger brownouts.
How it works (3–5 engineering steps)
- Establish stable low-voltage rails from mains so MCU/AFE remain above UVLO during load transients.
- Drive loads (circ/drain pumps, heater, valves, latch) with current/temperature-aware limits.
- Acquire sensor evidence (turbidity optical AFE, NTC/RTD, level/leak) using filtering and timing windows.
- Decide control actions (wash intensity, drain completion, heater duty) using thresholds and trend checks.
- Enforce safety gates (door open, over-temp, dry-fire/leak) to disable heater and enter safe states.
If only two things can be measured…
- Probe A: LV rail at the MCU decoupling point during pump/heater switching (captures brownout/reset root causes).
- Probe B: one raw “symptom-linked” channel: pump current (drain/circ), turbidity AFE output (LED-sync), or temperature rise slope (heater proof).
These two measurements separate most failures into power integrity, load/drive, sensing AFE, or safety gating.
- Circulation/drain pumps: drive evidence, stall/air/cavitation signatures
- Turbidity/temp/level/leak AFEs: raw signals, sampling/EMI susceptibility
- Heater enable logic: dry-fire/over-temp protection, switch behavior
- Door interlock chain: latch sensing, safe-state transitions
- Whole-home gateway/Matter stack internals
- Smart plug / utility meter / HEMS architecture
- Other appliance deep-dives (washer/dryer/fridge)
- Generic certification walkthrough (IEC/UL steps)
System Block Diagram & Power/Load Domains (Dishwasher-specific)
The system diagram must do two things: (1) separate power, loads, AFEs, and safety so each later chapter has a home; (2) make the two most common failure mechanisms visible—load-to-rail brownout and load-to-AFE noise injection.
What must be present (domain separation)
- Power path: Mains entry → EMI/surge → isolated LV PSU → 3.3/5V rails
- Control path: MCU/ADC → timing windows → drive commands
- Load path: circ/drain pumps, heater, valves, latch (each with its own protection)
- Sensing path: turbidity optical AFE, NTC/RTD, level/leak inputs
- Safety path: door interlock + over-temp/dry-fire/leak gating the heater and state machine
Evidence points (where to probe, what to look for)
- LV rail waveform at the MCU decoupling point: dip magnitude and duration vs UVLO/reset thresholds.
- Ground bounce / common-mode spike during heater/relay/triac edges and motor PWM transitions.
- Correlation: align spikes/dips to ADC sample windows or fault counters (reset, watchdog, AFE saturation).
A clean rail at the PSU output does not guarantee rail integrity at the MCU/AFE; the diagram anchors probe placement at the consumer node.
Coupling paths to highlight (dishwasher-only)
- Path 1 — Load → LV rail: pump PWM or heater switching pulls the rail below tolerance → MCU reset or ADC reference shift.
- Path 2 — Load switching → ground/sense inputs: dv/dt/di/dt creates spikes that enter optical/pressure inputs → false turbidity/level events.
Later chapters will repeatedly point back to these two arrows; without them, troubleshooting devolves into generic “EMI” guesses.
Circulation Pump Drive: Control, Current Sense, Stall/Noise
The circulation pump is typically the largest continuous load and the dominant noise source. Reliable control requires a closed loop that converts motor behavior into evidence (current, speed proxies, fault flags) and separates water-path issues from electrical and mechanical faults.
Architecture chain (what matters, not textbook)
- Motor type: BLDC/PMSM (inverter PWM) or AC (triac/relay).
- Driver topology: inverter stage (PWM edges) or triac phase control (dv/dt signature).
- Sensing options: shunt (direct torque evidence), Hall (speed evidence), back-EMF (sensorless proxy).
- Outcome: stable flow/pressure without triggering over-current, over-temp, or brownout resets.
Failure taxonomy (field symptoms → root buckets)
- Very loud: cavitation/air ingestion vs bearing friction vs commutation instability.
- No spin: hard stall (jam) vs undervoltage/brownout vs drive protection latch.
- Low efficiency: hydraulic restriction (filters/spray arms) vs derated drive limits.
- Intermittent stop: OTP/OCP cycling vs rail dip reset vs sensorless mis-detect at low speed.
Each symptom must be resolved using two aligned evidence layers: current waveform shape + driver status flags/counters.
Evidence points (minimum measurements with actionable discriminators)
- Stall/jam: high start peak, fast OCP trip, no convergence to steady current.
- Cavitation/air: periodic current ripple and audible modulation; mean current may drop.
- Bearing/friction: elevated baseline current with relatively stable ripple pattern.
- OCP/OTP/UVLO event ordering indicates electrical vs thermal vs supply causes.
- Derating records (reduced PWM/target speed) indicate a controlled fallback, not random failure.
- Repeated “stall detect” during low speed suggests sensorless margin/sampling-window sensitivity.
- Align current spikes with fault timestamps/counters to avoid “EMI guesswork”.
- Check whether rail dip precedes OCP or follows it (cause vs effect).
- Use a single phase/rail probe point consistently for comparable runs.
Design controls that prevent false trips (dishwasher-only coupling)
- Sampling discipline: place current/ADC sampling away from PWM edges and relay/triac transitions.
- Protection tuning: allow brief start peaks while bounding thermal rise; differentiate “jam” vs “air” by time profile.
- Noise containment: keep inverter high di/dt loops tight to reduce AFE pollution and rail disturbance.
- Graceful retry: bounded restart attempts with a clear lockout condition prevents oscillation between stall and reset.
Drain Pump & Drain Path: Anti-clog, Backflow, Dry-run
“Not draining” complaints frequently originate from the drain path (check valve, hoses, air locks) rather than the pump itself. The fastest diagnosis pairs electrical evidence (start peak vs steady current) with result evidence (water level/pressure trend).
Drain model (three states that explain most failures)
- Prime / build head: the pump must catch water and create flow without stalling.
- Steady drain: flow must persist; restrictions and check-valve behavior dominate.
- End-of-drain: controlled dry-run window + reliable “empty” detection prevents false completion.
Design controls (avoid repeated service calls)
- Start strategy: brief strong start vs soft start, tuned to avoid nuisance OCP.
- Stall criteria: time profile of current, not a single peak.
- Anti-clog actions: pulsed drain, bounded retries, optional short reverse/relief sequence.
- Backflow defense: verify check valve behavior using the level trend after stop.
Two-evidence rule (electrical + result evidence)
- Clog/jam: high peak, no steady convergence, early OCP/timeout.
- Dry-run/air: low steady current, unstable flow, “empty” mis-detect risk.
- Normal: peak → steady quickly, repeatable profile across cycles.
- Drain success: level decreases monotonically (small ripples allowed).
- No drain: level flat or oscillatory despite pump command.
- Backflow: level drops then rebounds after pump stop (check valve/path issue).
- Current indicates pump effort; level trend proves drain result.
- “Pump spins” without level change is path/airlock evidence.
- Backflow pattern points to valve/hoses, not motor replacement.
Turbidity Sensing AFE: Optical Front-End + Noise Immunity
Turbidity sensing is a dishwasher differentiator because the optical front-end must extract a small photodiode signal while pumps, relay arcs, and moisture leakage continuously inject interference. Robust designs treat turbidity as a synchronous measurement (LED pulse → correlated photodiode response) rather than a generic “slow analog input”.
Signal chain (what each block must guarantee)
- IR LED drive (pulsed): stable pulse amplitude/width sets the measurement reference.
- Photodiode + TIA: converts light to voltage; highest sensitivity to ground shift and leakage.
- ADC + reference: conversion must be time-aligned to LED pulses; reference integrity is critical.
- Synchronous sampling: sample inside the LED “ON window”, optionally subtract an “OFF window” to remove baseline drift.
Interference model (noise type → injection path → symptom)
- PWM ripple: pump edges modulate rails/ground → periodic turbidity wobble (speed-dependent).
- Relay/lock spikes: dv/dt and arcs couple into harness/ground → brief TIA saturation or ADC spikes.
- Moisture leakage drift: wet contamination creates high-R leakage at the PD/TIA input → slow offset drift and repeatability loss.
These three classes differ in time signature: periodic ripple vs impulsive spikes vs slow drift, enabling evidence-based separation.
Evidence points (minimum measurements that isolate the cause)
- LED pulse stable but PD response unstable → front-end leakage/ground reference sensitivity.
- LED pulse itself modulated at pump PWM → LED supply/return coupling, not “dirty water”.
- Relay/lock event causes brief clipping → dv/dt injection or insufficient front-end hardening.
- Noise drops when sampling shifts away from pump edges → timing/window issue.
- Noise persists across window shifts and tracks humidity → leakage drift dominating.
- Spikes align with relay actions, not LED window → transient coupling path.
- Compare ON-window vs OFF-window deltas to reject baseline drift.
- Correlate turbidity jumps with pump speed/relay timestamps.
- Use a fixed probe reference point to avoid “measurement-induced” drift.
Noise-hardening checklist (dishwasher-only priorities)
- LED return and TIA reference: prevent pump return currents from sharing the same sensitive reference path.
- Window discipline: place sampling in the center of the LED ON window; avoid relay/triac transitions.
- Leakage control: protect high-impedance input from condensation paths; prefer layouts that tolerate wet contamination.
- Spike containment: add appropriate transient control so dv/dt events do not clip the TIA output.
Temperature Sensing & Thermal Control: NTC Placement, ADC Strategy
Temperature control is not “one NTC and done”. Sensor placement changes what the reading represents, thermal lag reshapes the curve, and sampling strategy determines whether heater events appear as real temperature rise or as noise-induced artifacts.
Placement defines the physics being measured
- Tank / water path NTC: tracks wash effectiveness; slower response and flow-dependent slope.
- Heater-body NTC: best for overheat protection; responds quickly but can reflect local hotspots.
- Implication: thresholds and timing windows must be placement-specific, not copied across nodes.
ADC strategy (time alignment beats raw resolution)
- Windowing: avoid relay/triac transitions and pump PWM edges; sample in stable windows.
- Filtering: short window rejects spikes; longer window extracts slope for control decisions.
- Reference integrity: a “slow channel” can still be corrupted by ground shift and rail dips.
Evidence points (prove heating and validate sensor integrity)
- Heat OK: after heater ON, tank slope increases; after OFF, slope relaxes.
- No heat: heater ON does not change slope; indicates permission, power, or transfer failure.
- Dry-fire risk: heater-body rises rapidly while tank does not; indicates poor heat transfer.
- Open/short: counts saturate near rails and map to a clear diagnostic code.
- Intermittent contact: counts jump or dither near limits without clean saturation.
- Moisture leakage: counts drift with humidity/cleaning cycles rather than load transitions.
- Align temperature samples to heater commands and pump state for repeatable interpretation.
- Separate “spike noise” (sampling issue) from “slope change” (true thermal response).
- Use the same sensor node and reference point for comparable runs.
Thermal control rules (compact, implementation-oriented)
- Permission gating: heater enable requires valid sensor states and stable supply conditions.
- Slope-based checks: rely on rise-rate changes rather than single temperature snapshots.
- Safety priority: heater-body over-temp is fast protection; tank temp is effectiveness control.
- Event logging: store heater ON duration, fault ordering, and peak temperatures for service evidence.
Water Level / Leak Detection: Pressure/Float/Capacitive + False Trips
Level and leak false trips are a top cause of nuisance shutdowns. Robust designs separate “real water state” from pump-driven turbulence, foam, and humidity-driven leakage by using raw-signal evidence and correlation rules.
Level sensing options (native weak points)
- Pressure tube: high resolution but sensitive to bubbles, foam, tube blockage, and pump-induced pressure ripple.
- Float: simple and robust, but prone to mechanical sticking and wave slap during strong circulation.
- Capacitive: good for trends, but sensitive to condensation, surface contamination films, and common-mode injection.
Leak detection (conductive/impedance) and humidity traps
- Conductive probe: fast detection; false trips from condensation bridges and residue films.
- Impedance readout: supports drift tracking; requires baseline control to avoid “wet bias” decisions.
- Key distinction: real leaks create sustained conduction; humidity often creates slow, reversible drift.
False-trip model (time signature → likely cause)
- Ripple: periodic level wobble tracks pump speed or spray switching → turbulence/foam coupling dominates.
- Spike: abrupt threshold crossings align with valve switching or pump start/stop → hydraulic transient or coupling.
- Drift: slow baseline shift follows humidity/condensation cycles → leakage path or contamination film dominates.
Evidence points (minimum signals that isolate “real” vs “false”)
- Log raw level (pressure ADC / cap count / float state) with pump ON and speed stage.
- Pump-correlated ripple → treat as false-trip risk unless sustained out-of-range.
- Pump-weak correlation + persistent low level → more consistent with real fill/level fault.
- Slow reversible drift after hot cycles → humidity/condensation likely.
- Step change + sustained conduction → real leak or harness short more likely.
- Compare dry baseline vs post-cycle baseline to quantify drift magnitude.
- Require persistence over a time window, not a single threshold crossing.
- Apply different thresholds during pump transients vs steady circulation.
- Store “event ordering” (pump state → level anomaly → trip) for repeatable diagnosis.
Robustness tactics (changes that move the evidence)
- Correlation gating: ignore brief excursions that align with known pump/valve transients.
- Two-stage thresholds: strict threshold during steady periods; tolerant threshold during switching windows.
- Baseline management: periodic dry reference or self-calibration for capacitive/leak channels.
- Mechanical sanity: float bounce requires de-bounce logic; tube blockage requires plausibility checks.
Heater Drive & Protection: Relay/Triac/SSR, Dry-fire, Over-temp
Heater control is the safety core. Switching choice shapes surge/EMI behavior, and protection must close the loop using water-level permission, rise-rate evidence, and event ordering (trip vs reset).
Switching options (each has a distinct electrical fingerprint)
- Relay: large inrush step and arc spikes; often the strongest transient into rails/ground.
- Triac: phase/zero-cross behavior defines dv/dt and harmonic signature; timing matters.
- SSR: smoother switching but still dv/dt; leakage/thermal constraints must be respected.
Power transient chain (the reset mechanism)
- Heater ON causes mains sag during surge → LV PSU headroom shrinks.
- LV rail dips toward UVLO/BOD → MCU reset risk rises.
- Service logs should preserve event order: heater command → rail dip → reset/trip.
Dry-fire and over-temp: evidence-based closed-loop protection
- Level OK is the hard permission gate for heater enable.
- Rise-rate rule: heater-body rises fast while tank does not → transfer failure risk.
- Flow proxy: pump state/current supports “circulation present” without adding new sensors.
- Use threshold for absolute safety, slope to detect runaway, and time to reject spikes.
- Short spikes aligned with switching events indicate sampling/coupling, not real overheat.
- Sustained climb indicates true heating control failure or poor heat transfer.
- Record heater ON duration, peak temps, trip reasons, and rail minimum.
- Preserve ordering: level anomaly vs rise anomaly vs trip/reset.
- Align timestamps to relay/triac actions for repeatable diagnosis.
Evidence points (what to measure first)
- Heater ON transient: capture mains sag and LV rail dip together to confirm brownout risk.
- Over-temp triggers: correlate trip flags with rise-rate and absolute thresholds over a time window.
- Reset vs trip: determine whether the system reset occurred before protection logic could act.
Door Interlock & Safety Chain: Latch, Sensor, Functional Safety Logic
Door interlock issues frequently present as “won’t start” or “stops mid-cycle.” Reliable designs treat the latch as a full chain (actuator → position sensing → state machine → cut actions) and preserve event ordering evidence to separate mechanical rebound from sensor bounce and brownout-induced false triggers.
Hardware chain (what must be proven true)
- Latch actuator: motor/solenoid drive must reach lock position with repeatable current profile.
- Position sensing: Hall/limit signals confirm “latched,” not just “door closed.”
- Vibration + steam: humidity films and mechanical rebound can create brief “open” signatures.
- Critical nodes: actuator supply, sensor pull-up, connector/loom transition points.
Safety state machine (door open → safe cut)
- Door open must force heater OFF first (highest priority safety cut).
- Then pump STOP and valve CLOSE to prevent spraying or overflow.
- Re-entry requires stable “closed + latched verified” over a time window.
- After reset, the machine must rebuild a safe state before resuming any load.
Evidence points (minimum signals that isolate the root cause)
- Log DoorClosed and DoorLatched bits with actuator current during lock/unlock.
- Jam: current rises/plateaus high while DoorLatched never asserts.
- Rebound: DoorLatched asserts then drops briefly; current often shows a short “return” feature.
- Record sequence: input change → safety cut → resume/abort.
- Confirm heater cut occurs immediately on interlock trigger.
- Mark whether BOR/reset happened before the cut logic executed.
- If DoorLatched toggles near load switching, suspect coupling or brownout.
- If toggling aligns with vibration phases, suspect mechanical bounce/connector intermittency.
- Use persistence windows to separate a single blip from a real open event.
Fix-first playbook (fastest wins)
- Start with evidence: capture actuator current + latched status bits in one run to separate mechanical vs sensing.
- Stabilize inputs: define de-bounce windows and require “closed + latched verified” persistence before enabling loads.
- Humidity resilience: treat high-impedance inputs as leakage-prone; use baseline and robust thresholds.
- Reset awareness: if resets align with door events, hand off to H2-10 coupling/rail investigation.
Rugged Power & Load-to-AFE EMC Coupling (Dishwasher-only)
Dishwasher EMC failures are best debugged as an internal coupling map: known noise sources (pump PWM, triac dv/dt, relay arc, heater surge) inject through conducted paths, ground bounce, or capacitive coupling into victims (optical AFE, pressure/level, MCU rails). Evidence must show time alignment between load actions and glitches or brownouts.
Noise sources (signature and likely victim)
- Pump inverter/PWM: periodic ripple + common-mode spikes → AFE and pressure/level channels.
- Triac dv/dt: phase-correlated edge noise → optical AFE and MCU input lines.
- Relay arc: broadband spike bursts → long harness and high-impedance nodes.
- Heater surge: mains sag → LV rail dip → BOR/reset events.
Coupling paths (dishwasher-relevant)
- Conducted: mains sag → PSU headroom loss → LV rail dip → MCU brownout/reset.
- Ground bounce: shared return paths → AFE reference shift → ADC jumps.
- Capacitive: dv/dt + harness coupling → interlock/AFE input glitches.
- Humidity amplifier: condensation films increase leakage sensitivity on high-impedance nodes.
Evidence points (alignment proves coupling)
- Trigger on relay/triac/pump switching marker.
- Observe AFE/pressure outputs; prove time alignment of spikes or steps.
- If alignment is strong, focus on coupling path before sensor/algorithm changes.
- Read reset reason and brownout counters.
- Correlate with heater ON, pump start/stop, or valve switching timestamps.
- BOR-dominant resets point to rail headroom and hold-up gaps.
- Shift sampling away from known edges (triac/relay/pump PWM).
- Large noise reduction with timing changes proves a windowing problem.
- No improvement suggests leakage/drift or a hard rail disturbance.
Countermeasures (each tied to a coupling path)
- Partition returns: separate high-current load returns from analog reference; single-point connection where necessary.
- Analog guarding: protect high-impedance nodes (TIA/pressure input) with controlled RC and short sensitive loops.
- Sampling avoidance: schedule ADC windows away from known switching edges; treat “window” as a noise filter.
- Edge control: soften dv/dt where possible; reduce relay arc energy and snub high-energy nodes.
- Hold-up: ensure LV rail headroom and local decoupling at MCU/AFE pins for heater surge events.
H2-11 — IC Selection & Reference BOM Blocks
This section provides a decision framework for key dishwasher control BOM blocks (pump drive, optical turbidity AFE, heater switching, LV power). Example part numbers are representative (not endorsements). Always confirm voltage/current, isolation/creepage, EMC margin, lifecycle, and regional safety requirements.
11.1 BOM blocks covered
The selection work is split into four blocks. Each block should map to measurable evidence in earlier chapters (stall/flow evidence, glitch alignment, BOR/reset counters).
11.2 Selection Factors Table (spec dimensions + evidence hooks)
| BOM block | Selection dimension | Why it matters (dishwasher-specific) | Evidence hook (what to correlate) |
|---|---|---|---|
| Pump drive | Motor type coverage (BLDC/PMSM vs AC) | Different platforms use different pump motors; the driver class must match the motor and fault model. | Drive mode + fault bits vs symptom (“no start”, “intermittent stop”). |
| Pump drive | Current sensing path (shunt amp / Hall) | Stall, clog, cavitation, and dry-run discrimination depends on clean current signatures. | Phase/bus current waveform vs stall/cavitation signatures; stall trip timestamps. |
| Pump drive | Protection set (OCP/OTP/UVLO/lock detect) | Field failures often appear as “random stops”; protections must be explicit and logged. | Protection flag + retry count + time-to-recover. |
| Optical AFE | TIA input bias & noise (photodiode front-end) | High impedance + humidity + load switching can inject offsets and glitches into turbidity readings. | AFE baseline drift and glitch alignment to relay/triac/pump edges. |
| Optical AFE | LED pulsed drive capability | Pulsed IR improves ambient immunity and synchronous sampling; drive consistency impacts repeatability. | LED current pulse vs PD output sampling window alignment. |
| Optical AFE | ADC strategy (MCU ADC vs external SAR) | External SAR ADC can improve repeatability and timing control under noisy load switching. | Noise histogram / ENOB vs pump/heater activity windows. |
| Heater switch | Switching method (relay/triac/SSR) | Each method has a distinct dv/dt/arc/surge signature that can couple into AFE and LV rails. | Glitch occurrence rate per heater switch event; mains sag vs LV rail dip. |
| Heater switch | dv/dt immunity + snubber compatibility | Weak dv/dt immunity is a top root cause of false triggers and sensing corruption. | “Glitch aligned” to switch edge; interlock nuisance trips after heater events. |
| LV PSU | Hold-up / brownout margin | Heater or motor inrush can dip mains and collapse LV rails, causing BOR/reset loops. | BOR/reset flags vs load events; minimum rail voltage during load steps. |
| LV PSU | Surge/EFT tolerance & protection | Dishwashers face line disturbances; the PSU must survive without latent drift that corrupts sensing. | Post-surge drift in AFE baselines; reset counters after EFT bursts. |
11.3 Example MPN buckets (representative parts; not a product page)
Bucket A — BLDC/PMSM pump: 3-phase gate driver / motor driver
Use when the circulation or drain pump is a 3-phase BLDC/PMSM. Prefer devices with explicit fault reporting and robust PWM-noise tolerance.
- TI DRV8323 (3-phase smart gate driver + current shunt amplifiers)
- TI DRV8353 (higher voltage class smart gate driver)
- TI DRV8305 (3-phase gate driver family)
- ST L6234 (3-phase motor driver with integrated MOSFETs)
- ST L6230 (3-phase BLDC driver, integrated power stage)
- TI DRV10983 (sensorless BLDC motor driver for appliance pumps/fans)
fault pins/status + phase/bus current to stall/cavitation/dry-run signatures.
Bucket B — Pump current sensing: shunt amplifiers + Hall sensors
Use shunt amplifiers for waveform fidelity (stall vs cavitation). Use Hall sensors when isolation or low-loss sensing is needed.
- TI INA240A1/A2/A3/A4 (PWM-rejection current sense amplifier)
- TI INA181 (general current sense amplifier family)
- TI INA199 (high-side current shunt monitor family)
- Allegro ACS712 (Hall-effect linear current sensor)
- Allegro ACS723 (Hall-effect current sensor family)
current waveform + trip timestamp to separate clog/stall/dry-run.
Bucket C — Turbidity optical AFE: TIA op-amps + ADC
AFE must tolerate humidity leakage and load-switching noise. Prefer synchronous sampling and a stable baseline strategy.
- TI OPA381 (photodiode TIA-friendly, low bias, wide GBW)
- TI OPA380 (photodiode transimpedance amplifier)
- TI OPA320 (low-noise, rail-to-rail op amp for TIA stages)
- TI ADS7042 (12-bit, 1-MSPS SAR ADC, SPI)
- TI ADS8860 (16-bit SAR ADC family, SPI)
- ADI AD7980 (16-bit SAR ADC, SPI, precision sampling)
sampling window that avoids relay/triac/pump edges; verify “glitch aligned” rate.
Bucket D — IR LED pulsed drive: constant-current sinks / LED drivers
Pulsed drive supports ambient immunity and synchronous sampling. The LED current pulse must be repeatable and time-aligned to ADC sampling.
- TI TLC59208 (8-channel constant-current sink, timing-friendly pulsed drive)
- TI TLC5940 (16-channel constant-current sink)
- Diodes Inc. AL8805 (constant-current LED driver, simple pulsed setups)
- Microchip MIC3289 (LED driver controller family)
LED current and align to PD output and ADC sample (one timing reference).
Bucket E — Heater switching: relay drivers, optotriac drivers, discrete triacs
Heater switching is a top disturbance source. Focus on dv/dt immunity, snubber compatibility, and predictable switching signatures.
- ULN2003A / ULN2803A (relay/solenoid driver arrays)
- onsemi MOC3063M (zero-cross optotriac driver family)
- onsemi MOC3062M (zero-cross optotriac driver family)
- onsemi MOC3163M (zero-cross optotriac driver family)
- ST BTA16-600 (triac, 600 V class)
- ST BTA24-600 (triac, higher current class)
AFE glitches per switch event and LV rail dip during heater transitions.
Bucket F — LV power: offline flyback + supervisor/watchdog
The goal is preventing reset loops under heater/motor line dips and surviving surge/EFT without latent sensing drift.
- TI UCC28740 (CV/CC flyback controller with opto feedback)
- Power Integrations LNK364PN (LinkSwitch family offline switcher IC)
- Power Integrations LNK364DN (LinkSwitch family package option)
- TI TLV803 (voltage supervisor / reset IC)
- Maxim MAX809 (supervisor/reset IC family)
- TI TPS3430 (watchdog timer family)
BOR/reset flags, minimum rail, and reset counter vs load events.
- Choosing a pump drive path without explicit diagnostics (no proof for stall vs clog vs dry-run).
- Letting turbidity sampling occur during relay/triac edges (creates “glitch aligned” false turbidity).
- Ignoring heater/motor line dips and leaving hold-up undefined (causes BOR/reset cascades).
- Leaving high-impedance AFE nodes unguarded in a humid environment (baseline drift, nuisance trips).
H2-12 — Validation & Field Debug Playbook
Validation is structured as function × environment × sensor boundary. Pass/Fail is defined with counters, time-windows, slopes, and event ordering so “nuisance stops” can be proven and eliminated.
- LV rails (3.3 V / 5 V) capture with min/hold-up.
- One load marker (heater switch edge OR pump PWM marker).
- One sensor raw channel (turbidity ADC raw OR level/pressure raw).
- Reset reason (BOR/WDT flags) + event log timestamps.
- Humidity leakage on high-impedance nodes (optical AFE, level sensing).
- Heater switching disturbances (mains sag, dv/dt spikes, relay arcs).
- Pump load steps and turbulence (sensor ripple that looks like “real change”).
12.1 Functional validation (hard-core but compact)
| Function | Stimulus | Measurements | Pass/Fail definition (concrete) |
|---|---|---|---|
| Pump start | Cold start at low line and nominal line; repeat 100 cycles. | Bus/phase current signature; fault flags; time-to-stable speed/state. | Pass: ≥99/100 starts succeed; no more than 1 auto-retry; no OCP/UVLO faults. |
| Stall / clog | Controlled blockage to force stall; repeat 20 events. | Peak current; stall trip time; logged fault code. | Pass: stall detected & heater inhibited within 300 ms; logged fault present for every event (20/20). |
| Dry-run | Low water condition; run pump sequence. | Current waveform; level/pressure trend; turbidity stability. | Pass: dry-run declared within 10 s and does not falsely pass as normal cycle (0 false negatives). |
| Drain | Normal drain + partial clog + backflow attempt. | Drain pump start current vs steady-state; level/pressure monotonic drop. | Pass: level/pressure must drop monotonically within 30 s of drain start; clog must raise a specific fault (not “unknown stop”). |
| Heater enable | Heater ON/OFF switching 200 events across line conditions. | Mains sag proxy; LV rail minimum; AFE glitch counter; BOR flags. | Pass: BOR resets = 0 across 200 events; AFE “glitch aligned” rate ≤ 1 per 10,000 switch edges. |
| Door interlock | Door close/open bounce tests; 500 cycles; vibration + humidity. | Latch actuator current; DoorLatched bit stability; event ordering log. | Pass: DoorLatched must stay stable for ≥200 ms after lock; nuisance trips ≤ 1 per 500 cycles. |
12.2 Environmental stress (humidity, low line, surge, noise injection, temperature)
- Goal: prevent baseline drift and false leak/level events.
- Measure: turbidity raw baseline drift (counts/hour), leak/level nuisance counters.
- Pass: turbidity baseline drift ≤ ±3% of full-scale/day (platform-tuned), nuisance trips ≤ 1/day under steady operation.
- Goal: avoid reset loops during heater/pump events.
- Measure: LV rail minimum; BOR flags; reset counters.
- Pass: rail never drops below Vmin = 3.0 V for a 3.3 V domain for longer than 1 ms; BOR resets = 0.
- Goal: survive without latent sensing corruption.
- Measure: post-stress AFE baseline; glitch counter; reset counter.
- Pass: no permanent offset shift > 2σ (baseline noise); reset counter does not increase during or after the burst series.
- Goal: keep dry-fire / over-temp logic consistent with sensor placement.
- Measure: heater-body slope vs tank slope; trip thresholds and timestamps.
- Pass: dry-fire rule triggers when dT_heater > 1.5°C/s for 10 s while dT_tank < 0.2°C/s (example rule; platform-tune).
12.3 Sensor boundary conditions (foam, turbulence, dirt)
- Foam: verify level/pressure signals do not falsely cross safety thresholds; require “persistence window” ≥ 2 s before declaring low-water.
- Turbulence: require pump-correlated ripple rejection; turbidity/level should not trip on periodic ripple at pump switching frequencies.
- Dirt build-up: track optical baseline and response time; require recalibration only when baseline drift exceeds 5% full-scale over 24 h (example threshold).
12.4 Pass/Fail counters (the “proof set”)
| Counter / metric | How it is incremented | Acceptance target |
|---|---|---|
| BOR / reset count | Increment on BOR flag set; store timestamp + last load action (heater/pump/valve). | 0 across full validation set; any BOR requires rail capture and root-cause closure. |
| Glitch-aligned count | Increment when AFE raw shows spike aligned within ±2 ms of load edge marker. | ≤ 1 per 10,000 load edges in worst-case stress runs. |
| Nuisance interlock trips | Increment when door/level/leak interlocks trigger without confirming persistence windows. | ≤ 1 per 500 door cycles; ≤ 1/day under humidity stress steady state. |
| Stall protection timing | Measure time from stall onset signature to protection action + log entry. | Heater inhibit ≤ 300 ms; log entry present in 100% of events. |
| Drain success metric | Measure monotonic level/pressure decrease and time to reach target band. | Must reach target band within 30 s (platform-tuned); clog must produce specific fault code. |
12.5 Field debug: “First two measurements” per complaint
- Measure #1: LV rail minimum during heater/pump events + BOR flags.
- Measure #2: event log ordering (which load action occurred right before reset).
- Measure #1: turbidity raw waveform + load edge marker (glitch alignment).
- Measure #2: LED pulse current + ADC sampling window alignment.
- Measure #1: DoorLatched bit stability + latch actuator current signature.
- Measure #2: interlock trip log (heater off first? pump stop?).
Request a Quote
H2-13 — FAQs ×12 (Evidence-based, Dishwasher-only)
Each answer follows a fixed structure: Measure 2 things → Discriminate → First fix. This keeps troubleshooting repeatable and prevents scope drift.
“Not draining” but pump spins—clog, check valve, or level sensor misread?
Measure #1: drain-pump startup peak and steady current. Measure #2: raw level/pressure trend during the drain window (should drop monotonically). Discriminator: normal load-like current with no level drop points to clog/check-valve/backflow; level drops but UI still shows “full” points to sensing/foam filtering. First fix: correlate current + level trend, then adjust anti-clog strategy or level persistence windows.
Drain pump current spikes then stops—stall vs dry-run, how to prove?
Measure #1: 0–2 s current waveform shape (peak then plateau vs peak then collapse). Measure #2: concurrent level/pressure response (any real drop). Discriminator: stall typically sustains high current and sets a protection flag; dry-run often shows current collapsing with little/no level change. First fix: lock thresholds on waveform + time-to-trip, then tune retry/anti-jam reverse pulses versus dry-run declare windows.
Loud buzzing during wash—cavitation, debris, or bearing? Which waveform first?
Measure #1: circulation pump bus/phase current envelope (modulated vs steady). Measure #2: a speed/PWM marker to align noise episodes to operating points. Discriminator: cavitation produces unstable, load-dependent modulation; debris often creates intermittent spikes/impacts; bearing wear tends to be steadier and scales with speed. First fix: separate the category by current signature, then target water-path inspection, impeller cleaning, or bearing replacement validation.
Turbidity jumps when motor PWM changes—AFE pickup or sampling-window issue?
Measure #1: LED pulse current aligned to photodiode/TIA output and ADC sample timing. Measure #2: “glitch-aligned” statistics: turbidity raw spikes within a small window around PWM/relay edges. Discriminator: strict time alignment to switching edges indicates coupling or a sampling window that lands on edges; non-aligned drift suggests leakage/baseline or optical path issues. First fix: shift sampling windows away from edges, then harden AFE reference/return paths.
Heater turns on then trips quickly—dry-fire protection or door interlock?
Measure #1: event log ordering and the exact trip reason bit (door vs dry-fire vs over-temp). Measure #2: temperature slope vs water-level permit at heater enable. Discriminator: door interlock trips show latch-state transitions and immediate heater inhibit; dry-fire trips show abnormal heater-body slope while tank response stays small, often with low/unstable level conditions. First fix: confirm the log path, then tune latch de-bounce or dry-fire slope/persistence thresholds.
“Not heating” but temperature rises slowly—sensor placement or heater power loss?
Measure #1: heater ON/OFF timestamp vs temperature curve inflection (is there a clear step response). Measure #2: heater power evidence (switch status + load current proxy). Discriminator: real power with slow response points to sensor placement/thermal lag/mixing; weak or intermittent power points to relay/triac conduction loss, wiring, or protection gating. First fix: prove power delivery first, then validate sensor placement and ADC filtering windows against heater switching.
Random reboot when heater switches—LV rail sag or EMI spike?
Measure #1: MCU VDD minimum at the decoupling point during heater transitions (depth and duration). Measure #2: reset cause flags (BOR/WDT) and counter correlation to heater events. Discriminator: a real VDD dip beyond BOR threshold indicates hold-up/brownout; stable VDD with resets indicates EMI/ground-bounce injection or reset-pin sensitivity. First fix: separate dip vs spike, then apply hold-up/edge control or reference/return path hardening accordingly.
False leak alarm in high humidity—conductive sensor drift or harness leakage?
Measure #1: leak sensor raw impedance/ADC counts over humidity ramps (baseline drift and noise). Measure #2: isolation step: disconnect or segment the harness to see whether raw returns to normal. Discriminator: raw recovers when harness is isolated implies harness/connector leakage; raw remains shifted implies board contamination, high-impedance node leakage, or threshold strategy. First fix: localize the leakage path, then implement guard/protection and persistence windows to prevent nuisance trips.
Water level oscillates during circulation—pressure-tube resonance or foam turbulence?
Measure #1: level/pressure raw vs pump PWM/speed marker coherence (locked frequency vs wandering). Measure #2: change operating conditions (detergent/foam level or pump duty) and observe amplitude response. Discriminator: a stable oscillation locked to pump dynamics suggests tube resonance or routing; oscillation that changes strongly with foam/turbulence suggests fluid effects and filtering needs. First fix: treat resonance via damping/routing/filter shaping; treat foam via persistence windows and multi-sensor permits.
Door “closed” but unit won’t start—latch current, Hall state, or state-machine gating?
Measure #1: latch actuator current signature (move, reach, settle) to detect stall or mechanical rebound. Measure #2: DoorLatched/Hall/limit raw state stability and the state-machine gating condition at start. Discriminator: good current with unstable state implies sensor/installation/harness; stable state with no start implies another gating permit (leak/level/thermal). First fix: prove actuator + sensor stability first, then trace the exact gating condition that blocks start in the event log.
Relay/triac switching causes AFE glitches—layout/grounding or dv/dt immunity?
Measure #1: “glitch-aligned” rate: AFE raw spikes within ±2 ms of relay/triac edges. Measure #2: controlled edge changes (snubber, gate resistor, zero-cross choice) and observe glitch-rate reduction. Discriminator: large reduction when edges soften indicates dv/dt-driven immunity limits; persistent glitches sensitive to probe ground/reference indicates return-path/ground-bounce. First fix: decide which mechanism dominates, then apply edge control plus sample-window avoidance, or rework AFE reference/partitioned returns.
How to design production tests for turbidity/temp/level without real dirty water?
Measure #1: turbidity chain integrity using a repeatable optical surrogate (fixed attenuator/reflector) and verify LED pulse → PD/TIA → ADC window repeatability. Measure #2: level/leak input emulation using equivalent impedance/capacitance fixtures to sweep thresholds and persistence logic. Discriminator: production tests must validate signal-chain control and diagnostics, not replicate real soil curves. First fix: define pass/fail counters (baseline drift, glitch-aligned count, sensor open/short codes) and close the loop with H2-12 evidence tree.
