Skip to content
Ice Maker Specialist
An owner staring at a flashing panel who has already tried the reset and wants to know which sensor latched the fault and how to test whether it's telling the truth.11 min read · Updated July 2026

Blinking Lights and Error Codes: What Each Sensor Is Actually Reporting

Every fault light on a home ice maker is a sensor reading that crossed a threshold — a thermistor resistance, a float continuity, an interrupted infrared beam, a stalled motor. A reset clears the latch; it does not change the reading that caused it, which is why the fault comes back on cycle two or three. Test the sensor with a meter before you replace anything: a 10k thermistor should read 28-34 kilo-ohms in an ice bath and 10.5-12 kilo-ohms at room temperature, and an inlet valve coil should read 200-500 ohms.

By the Ice Maker Specialist Editorial Team

A fault light is not a diagnosis. It's a sensor reading that crossed a threshold, and the control board latched it so the machine would stop instead of destroying itself. Add Water doesn't mean add water — it means the water-level input did not confirm within its timeout. Ice Full doesn't mean the bin is full — it means the infrared beam across the bin stayed blocked for longer than the control expects.

That distinction is the whole reason resets fail. Pulling the plug for ten minutes clears the latch. It does not descale the conductivity probe, unstick the float, wipe the frost off the IR lens, or change the thermistor's resistance. The board comes up clean, samples the same sensor, gets the same reading, and latches again on cycle two or three. Most machines that get replaced as dead have a $15 sensor problem and an owner who reset it eleven times.

So take a reading first. The tables below name the sensor behind each indicator pattern, the physical quantity it measures, the bench test with an expected value and units, and — the part that matters — whether the reset will hold or re-latch.

  1. What a latch is and why the reset lies

    Control boards on these machines are simple. They sample a handful of inputs at fixed points in the cycle, compare each to a window, and if an input sits outside its window for longer than a timeout, they set a flag in volatile memory and light an LED. That flag is the latch. Cutting power clears volatile memory, so the flag goes and the light goes with it.

    Nothing about the sensor changed. The board just forgot.

    This produces a signature that's genuinely diagnostic if you pay attention to it. A fault that clears and stays clear for a week was a transient — a momentary pressure drop, a single bridged batch, an ice slab that eventually melted. A fault that returns within one to three cycles is a standing condition or a dead sensor, and no amount of resetting will change that. Write down how many cycles it survives. That number narrows the suspect list more than the indicator pattern does.

  2. What you need before you test anything

    A digital multimeter is the entire toolkit for eighty percent of this. Continuity, resistance and DC volts cover every test in the main table. Everything else on this list is convenience.

    • Digital multimeter with continuity, resistance to at least 200 kilo-ohms, and DC volts — a $25 meter is fine
    • A glass of ice water, stirred, for the 32F reference point — the physics gives you a free calibration standard
    • A probe thermometer for ambient air at the intake and for inlet water temperature
    • A clamp meter or an inline plug-in power meter if you want to see harvest current draw
    • A plastic card or a strip of card stock for interrupting the bin-full infrared beam
    • A nut driver set — most fridge ice maker modules come out with a 1/4 in driver and two screws
    • A stopwatch, because cycle length is a measurement and you'll want a baseline

    Unplug the machine before any resistance or continuity test. Resistance readings taken on a powered circuit are meaningless and can damage the meter. DC voltage tests on the infrared pair are the only ones in this guide that need power applied, and those are low-voltage logic signals on the control side.

  3. Sensor decision table: what latched, and how to test it

    Indicator patternSensor or switch that latched itPhysical quantity it readsThreshold that trips itBench test, instrument and expected readingPass bandSensor or condition at fault?Part cost bandReset verdict
    Add Water LED on with a visibly full reservoirWater-level float switch, or a two-pin conductivity probeWater column height, or conductivity between two probe pinsFloat reads open for more than 3-5 s, or probe resistance above roughly 200 kilo-ohmsMeter on continuity. Lift and drop the float by hand: it should snap between under 1 ohm and open circuit. For a probe, measure pin-to-pin with the reservoir fullFloat toggles cleanly both ways; probe reads under 100 kilo-ohms submergedSensor in most cases — scale film on the probe pins or a float stuck on mineral deposit$8-30Re-latches within 1-3 cycles until the probe is descaled or the float freed
    Add Water LED on, reservoir genuinely empty, machine plumbed to a lineInlet solenoid valve coilCoil continuity, and whether the valve physically opens under line pressureCoil open circuit, or static supply pressure below about 20 psiMeter across the coil terminals with the machine unplugged: expect 200-500 ohms. Then check static pressure at the nearest tap with a $12 gauge200-500 ohms and 20-120 psi staticSensor path if the coil is open; condition if the coil is good — kinked line, closed stop valve, clogged inlet screen$18-45 valveHolds only if water actually flowed; otherwise re-latches on the next fill attempt
    Ice Full or bin-full LED with an empty or half-empty binBin-full infrared emitter and receiver pairInfrared beam continuity across the binBeam reads blocked continuously for 8-30 s depending on modelDC volts at the receiver output while you interrupt the beam with a card. The reading must swing at least 1.5 VDC between clear and blockedClean swing both directions; emitter supply 4.5-5.5 VDCUsually condition — frost, scale haze or a fingerprint film on either lens. Wipe both windows with a dry cloth first$15-40 for the pairHolds after a lens wipe. Re-latches within 2 cycles if the emitter is actually dead
    Bin-full never trips; ice packs solid and jams the chuteSame infrared pair, receiver stuck in the beam-clear stateInfrared beam continuityNever registers blocked, so the machine never stopsSame DC volts test. Block the beam completely with a card. If the output will not move, the receiver is deadSwing of 1.5 VDC or moreSensor, straightforwardly$15-40 for the pairReset changes nothing at all — there is no latch to clear
    Freeze cycle runs long past normal and never harvestsEvaporator thermistor, or on older designs a bimetal cutout thermostatEvaporator plate or prong temperatureHarvest is triggered somewhere around 10-20F at the evaporator on most home designs10k NTC thermistor in a stirred ice bath: expect 28-34 kilo-ohms. Same thermistor at 72F room: 10.5-12 kilo-ohms. Bimetal cutout: continuity closed below its stamped temperatureResistance roughly triples going from 72F to 32FSensor if the ratio is wrong. Condition — low refrigerant charge or a fouled condenser — if the ratio is right and the plate never gets cold$8-25 thermistorReset gives you one more long cycle, then re-latches
    Harvest motor hums or clicks, no ice releasedHarvest or ejector motor and its rotation timeoutMotor current and shaft rotation within an expected windowControl aborts when it does not see rotation complete inside its windowClamp meter on the module lead during harvest: a healthy home ejector draws well under 1 A. With power off, the gear train should turn by hand with firm but not violent pressureTurns freely by hand; draws current without stallingCondition first — ice welded into the mould from a previous incomplete harvest. Motor or gear train second$25-70 motor; $60-160 for a complete fridge moduleRe-latches every cycle until the mould is thawed or the gear train replaced
    Power LED flashes a repeating count, machine otherwise deadControl board self-test across several inputs at onceMultiple sensor inputs sampled at power-upAny input outside its window during the startup sampleRead all three primary sensors cold before touching the board: thermistor resistance, float or probe continuity, valve coil resistanceAll three inside their bands aboveIf all three pass, the board becomes the suspect — and only then$60-200 boardHolds for one power cycle, then the same count repeats
    Cycles run normally, output roughly halved, no light at allNothing latched — this is a derate, not a faultAmbient air temperature and inlet water temperatureNo threshold exists; output falls continuously as either risesProbe thermometer at the air intake and at the water inlet. Weigh 24 hours of output and divide by nameplate to get the Harvest IndexHarvest Index of 0.75 or better at 85-90F ambient is normal, not brokenCondition — thermodynamics, not a defect$0There is nothing to reset
    Water sitting in the bin instead of iceEvaporator thermistor reading warm, or harvest firing earlyPlate temperature at the harvest decision pointHarvest triggers before the ice has setThermistor ratio test as above. Also confirm the condenser fan actually runs through the whole freeze stageRatio near 3x between 72F and 32F; fan running continuouslySensor most often; fan or charge second$8-25 thermistor, $20-45 fanRe-latches every cycle — the reading hasn't changed
    Machine completely dead, no LEDs, no fanInline fuse or thermal cutout in the power pathContinuity through the protective deviceOpens permanently above its rated temperature or currentContinuity across the fuse or thermal cutout with the machine unplugged: expect near 0 ohmsUnder 1 ohmCondition — something made it hot or drew too much. Find that before fitting a new one$5-20Not a latch. Nothing to reset, and replacing the fuse without finding the cause just buys you one more failure
    Nugget machine: auger stalls, motor labours then stopsAuger motor stall detection or its thermal overloadMotor current and rotation against ice compaction loadCurrent above the motor's limit, or overload opening on temperatureClamp meter on the auger motor lead during operation. Then, powered off, check whether the auger turns by hand — it should move with resistance but not be lockedTurns with resistance; motor cool to the touch after a restCondition in most cases — scale on the cylinder wall increasing the scrape load. Descale before condemning the motor$80-180 auger motorHolds for a few cycles after a rest as the overload cools, then re-latches until descaled
    Take the reading in the bench-test column before you press reset. The reset changes the latch; only the reading tells you whether the sensor or the machine is wrong.

    Read the last column as a diagnosis in its own right. Every row where the verdict is re-latches within N cycles is telling you that the underlying reading is unchanged, and that repeated resets are burning your time. The only rows where a reset is a legitimate repair are the ones where you fixed the physical condition first — wiped the lens, freed the float, thawed the mould — and the reset is just clearing the flag afterwards.

  4. Brand crosswalk: what the panel calls it, what to meter

    Brand or model familyWater-side fault appears asBin-side fault appears asThermal or harvest fault appears asSensor to meter first
    GE Opal (countertop nugget)Add Water indicator, usually steadyIce Full or bin indicator litCombination of indicators, or auger simply stops extrudingReservoir float first, then the auger motor current
    GE Profile (undercounter and refrigerator lines)Fill valve does not energise; no water at the mouldBin thermostat or optical board keeps the module parkedModule cycles without releasing, or never reaches harvest temperatureInlet valve coil resistance, then the module thermistor
    Frigidaire (EFIC countertop series)Add Water LED, often with the machine halted mid-fillIce Full LED, sometimes with both LEDs on togetherBoth LEDs flashing together is the usual control-side patternConductivity probe pins in the reservoir — these scale fast
    Whirlpool (refrigerator ice modules)No fill at all; the module parks after a failed fill attemptOptical emitter/receiver board on the freezer wall holds the module offMould heater or ejector motor fails to complete a harvestOptical board output voltage, then the mould thermistor
    Samsung (refrigerator ice modules)Ice maker indicator on the door panel; no fillIce-off state persists with an empty bucketIce sheeting or freezing solid in the tray before harvestModule thermistor, then the fill tube for a frozen plug
    Kenmore (Whirlpool- or LG-built refrigerators)Follows whichever manufacturer built it — check the module tagSame, optical or mechanical bin sensing by buildSame as its builder's module designIdentify the module maker from the tag first, then meter to that pattern
    Igloo (countertop portable)Add Water LEDIce Full LEDBoth LEDs on with the compressor idleFloat switch continuity in the reservoir well
    Euhomy (countertop and undercounter)Add Water LED, sometimes flashing rather than steadyIce Full LEDPower LED flashing a repeating countReservoir float, then the infrared pair, then the thermistor
    Silonn (countertop portable)Add Water LEDIce Full LEDBoth LEDs together, compressor offFloat switch, then the infrared pair across the bin
    Whynter (portable and undercounter)Add Water indicator; on plumbed models the valve fails to openIce Full indicatorCompressor protection delay holding the unit off for several minutesInlet valve coil on plumbed units; float on reservoir units
    Insignia (countertop and undercounter)Add Water LEDIce Full LEDAlternating LEDs with no compressor runFloat or probe, then the condenser fan and thermistor
    KBice (self-dispensing nugget)Add Water LED with the pump running dryBin sensing holds the auger offAuger labours and stops; motor gets hotAuger motor current draw, and the cylinder wall for scale
    Manitowoc (home-installed undercounter)Service or fault LED on the control board rather than a front panelBin switch or thermostat holds the machine in standbySafe-mode lockout after several consecutive failed harvestsBoard LED sequence first, then evaporator thermistor and bin switch continuity
    Scotsman (home-installed undercounter)Status LED sequence on the control board; water sensing on the reservoirBin thermostat or bin switch keeps the unit parkedBoard reports a long-freeze or failed-harvest condition and locks outEvaporator thermistor resistance, then bin thermostat continuity
    Use this only to find the sensor. Once you know which one latched, go back to the decision table and take the actual reading — the panel wording never tells you whether the sensor is honest.

    Panel wording varies by firmware revision within the same model name, and it changes between production runs without the model number changing. That's exactly why a code-to-meaning list goes stale and a resistance value does not. A 10k thermistor reads 10k at 25C on every machine ever built, whichever badge is on the door.

  5. The thermistor test, done properly

    This one test resolves more ice maker faults than any other, and most people do it badly. Two things go wrong: they test at an unknown temperature, and they compare against a number from the wrong sensor family.

    Fix the temperature problem with physics. A glass of ice and water, stirred, sits at 32F within a fraction of a degree as long as ice is still present. That's a free calibration standard sitting in the machine you're repairing. Put the thermistor bead in it, wait ninety seconds, read the resistance.

    Fix the family problem by measuring twice. Rather than trusting a single absolute number, read the thermistor at room temperature and again in the ice bath, then look at the ratio. A 10k NTC with a typical beta value climbs to roughly three times its room-temperature resistance at 32F — around 11 kilo-ohms at 72F becoming 28-34 kilo-ohms at 32F. The ratio holds even if the sensor is a 5k or 50k part, because beta is what sets the curve shape.

    A sensor that barely moves between the two readings is dead. A sensor that reads open on both is disconnected or broken internally. A sensor that reads near zero on both is shorted, usually where the harness chafes on a bracket. And a sensor that passes both readings cleanly has just told you the fault is somewhere else — which is the most valuable result the test produces, because it stops you buying the wrong part.

  6. When the sensor is honest and the machine is the problem

    Roughly a third of fault lights are correct. The sensor read what it read, and something real is wrong.

    Low refrigerant charge shows up as a thermistor that passes its ratio test while the evaporator never gets cold enough to trigger harvest. You get long cycles, then a timeout latch. There's no home fix for this; it's a sealed-system job and, on any countertop machine, it costs more than the machine.

    A fouled condenser produces the same picture more cheaply. Dust and kitchen grease on the fins raise head pressure, the compressor works harder for less cooling, cycles lengthen, and eventually a timeout fires. Vacuum the fins, run the machine, take the Harvest Index again. A machine that goes from 0.55 to 0.85 after a condenser clean was never faulty.

    Inlet pressure below about 20 psi means the valve opens and not enough water arrives inside the fill timeout. The coil passes its 200-500 ohm test, the valve clicks, and the machine still latches an Add Water fault. Check a stop valve that's been half-closed since the last plumbing job before you order anything.

    And ice welded into the mould from an interrupted harvest is the most common honest fault of all. The harvest motor genuinely cannot turn, the stall detection is genuinely correct, and the repair is a hair dryer on low and ten minutes of patience.

  7. The rule this whole page comes down to

    Measure, then reset. Never the other way round.

    If you reset first, you destroy the evidence — you don't know what the sensor was reading when it latched, and you have to wait for the fault to return before you can measure anything. If you measure first, the reset becomes a test rather than a hope: fix the condition, clear the latch, and count cycles. Three clean cycles means you fixed it. One clean cycle means you didn't.

    Under $50 of parts and a $25 meter resolves most of what gets thrown away in this category.

More guides