How to Test a Defrost Thermostat or Defrost Sensor in Ice Water
Why this matters
A frosted evaporator gets blamed on the heater because the heater is the part people know. Half the time the heater is fine and the device that tells it when to stop is the problem: a bimetal thermostat stuck open never lets the heater run, a thermistor reading the wrong value lets the board end defrost early, and a good device hanging off a loose clip ends every defrost before the bottom of the coil is clear. An ice-water cup and a meter separate those three in ten minutes, before anything is ordered. This card covers the terminating device only. Heater replacement is 'Replace a Refrigerator Defrost Heater', and the full defrost diagnosis sequence lives in 'Refrigerator Defrost System Repair Standard'.
Before the cover comes off
Power. Unplug the refrigerator, or switch off its breaker if the cord is out of reach, and prove it dead at the defrost harness with a meter checked on a known live source before and after, per NFPA 70E Article 120 (120.5 in the 2021 edition, renumbered in 2024). A refrigerator mid-defrost has line voltage on the heater circuit, and the thermostat leads sit beside it. Every resistance reading in this card is taken with the unit dead; the only live reading is the optional voltage check named below, taken at the thermostat connector.
Heat and edges. If the unit was in defrost when you arrived, the heater and its sheath can be hot enough to burn for several minutes after power is removed. Let it sit before your hand goes behind the cover. Evaporator fins cut; wear cut-resistant gloves to reach the clip.
Food and water. Move freezer contents to coolers before you start; the test takes ten minutes but a coil thaw can take an hour. Towels go on the freezer floor, because ice dropped off the coil melts onto the floor and into the drain trough you are about to inspect.
Where the device sits and why the clip matters
The terminating device clamps onto the evaporator tubing, usually near the top of the coil on the outlet side, a spot the maker chose because it clears late. The heater sits along the bottom.
evaporator, cover off, viewed from the front
outlet tube -> ====@========== <- defrost thermostat
||||||||||||||| or sensor clipped
fins and ||||||||||||||| to the tube (@)
tubing |||||||||||||||
|||||||||||||||
defrost heater ~~~~~~~~~~~~~~~ <- radiant or sheath
drain trough \_____________/ heater, bottom
|
drain hole
The device is placed where the coil clears last so that "clear here" means "clear everywhere." That only works if it reads the tube temperature, which means metal-to-metal contact under the clip. A clip that has slid off the tube, a sensor hanging by its leads, or a thermostat with its face turned away from the tube reads the air in the cavity instead. During defrost that air warms fast from the heater below, so a loose device reaches its opening temperature while the tube under it is still iced. Defrost ends early, ice survives in the rows the heater reaches last, the next cycle adds to it, and over a week or two the coil plugs and the fresh food side warms. A short defrost leaves the coil partly clear; a device that never closes leaves it frosted solid. That pattern is the first clue to which failure you have.
The two kinds of device
Bimetal defrost thermostat. A snap-disc switch in series with the heater on timer systems, or wired to the board as an input on some electronic ones. It is closed when cold and open when warm. On a timer system, the timer sends power to the heater and the thermostat is the switch that stops it; if it never closes, the heater never runs. Each thermostat has its own close and open temperatures, and some close only well below freezing, which matters for the ice-water test.
Thermistor defrost sensor. An NTC thermistor (negative temperature coefficient: resistance rises as temperature falls) read by the control board, which ends defrost when the reading says the coil is warm. There is no switch to open; it fails by reading the wrong value, opening (meter reads OL), or shorting (near zero ohms). The tech sheet for that model gives a resistance-versus-temperature chart, and there is no universal value; see related: 'NTC Thermistors and Oven RTD Sensors Explained' for how the curve behaves.
Check the wiring diagram before you assume which you have. Two wires in a small metal can on a timer unit is almost always the bimetal type; a molded sensor bead on a board-controlled unit is usually a thermistor, and the board's service mode may already show its temperature.
Testing the bimetal thermostat
- Disconnect it from the harness. Read it in circuit and you may be reading the heater in parallel. Most have a connector within a few inches of the can; on some the leads are spliced and you cut and re-splice with moisture-sealed connectors, never bare wire nuts in a freezer.
- Read it warm first. At room temperature, with the can warmed in your hand, a good thermostat reads OL (open). A reading near zero ohms at room temperature means the contacts are welded closed, and the heater would run for the whole timer period every cycle, leaving only the heater circuit's own high limit or thermal fuse to stop it. That safety stack belongs to 'Thermal Fuses, High Limits and Cycling Thermostats in Appliance Heat Circuits'; a welded-closed thermostat that has already tripped that protection is replaced together with whatever it tripped.
- Chill it in ice water. Fill a plastic cup mostly with crushed ice and top it with water. A cup that is mostly ice holds close to 32 F; one that is mostly water with a few cubes floating does not. Submerge the can, keep the connector out of the water, stir, and wait five minutes. The can has thermal mass and the snap disc lags the water.
- Read it cold. A good thermostat now reads near zero ohms (continuity). Expect a fraction of an ohm, plus your meter leads.
- If it still reads OL, do not condemn it yet. Some defrost thermostats are rated to close below 32 F, and ice water never gets them there. Put the thermostat in the freezer compartment of a working freezer, or pack it in ice with a handful of salt, for ten minutes and read again. If it closes there and the rating on the tech sheet or the can agrees, it is good. If it stays OL at a temperature colder than its rated close point, it is bad.
Optional live check, at the connector only. On a timer system you can confirm the thermostat in place: restore power, advance the timer into defrost, and read voltage across the two thermostat terminals at the connector with insulated probes. A closed thermostat reads near 0 V; an open one reads close to line voltage, because the heater on the other side is completing the circuit to your meter. Keep both hands clear of the heater and fins while powered, and kill power again before touching anything.
Testing the thermistor sensor
- Disconnect it at the board or at the evaporator connector, whichever the tech sheet marks as the test point. Reading at the board tests the harness too; reading at the evaporator isolates the sensor.
- Find the chart for this model. The tech sheet is usually behind the kick plate or taped inside the back panel; if you have not found it, 'How to Find and Read the Tech Sheet Before the First Reading' covers where to look.
- Chill in ice water the same way, five minutes, connector dry.
- Compare with the chart value at 32 F. Most charts give a tolerance; use it. If yours does not, a reading a few percent off is bath and meter noise, while a reading roughly double or half the chart value, OL, or near zero ohms is a failed sensor.
- Watch it move. Lift the bead out and hold it in your hand. On an NTC sensor the resistance should fall steadily as it warms. A value that jumps, drops out to OL, or freezes in place while you flex the leads points to the lead or the crimp, not the bead.
A worked case: the thermostat was good
A side-by-side, freezer warm, fresh food at 46 F. The evaporator cover comes off and the pattern is the clue: the bottom rows nearest the heater are clear and the upper rows are packed with ice, worst at the outlet end. A thermostat that never closed would leave the whole coil frosted; this one is clearing some and stopping.
The thermostat is a bimetal can on a timer system. It is found hanging below the outlet tube, its spring clip slipped off, resting against the fins of a lower row, close to the heater and among the first to clear.
- Unplugged, proved dead at the harness, thermostat disconnected at its connector.
- Room temperature, can warmed in hand: OL. Correct.
- Ice-water cup, mostly crushed ice, five minutes: 0.3 ohms. Correct.
- Heater read across its own connector cold: within the tech sheet range for this model. The sibling card 'Estimating Heater Resistance From the Rating Plate Wattage' owns the sanity check on that figure.
Nothing on the meter is bad. The fault was physical: hanging free, the thermostat sat in the heater's warmest zone and opened while the outlet tube above it was still iced, so every defrost ran short and the upper rows gained ice each cycle. Ordering a new thermostat would have cured it only because the tech fitting the new one would have clipped it on properly, and the next tech would have trusted a part that was never at fault.
The fix: thaw the coil fully (a steamer or a pan of hot water with the doors shut, never a heat gun near the plastic liner or a sharp tool near the tubing), clean the outlet tube where the clip sits, reseat the clip so the thermostat face is flat on bare tube, and check that the drain hole is clear by pouring a cup of warm water into the trough.
What would have changed the call. If the coil had been frosted solid top to bottom, the cause is a defrost that never starts: a thermostat stuck open, a heater open, or a timer or board not calling. If the can had read OL in ice water and closed in the freezer, the thermostat is simply a low-close design and is fine. If the clip had been tight and the thermostat good, the short defrost points back to the timer or board ending defrost early, which is the defrost system standard's next branch.
The finish test
- Reassemble and restore power, and let it cool first. A coil you just thawed is above freezing, so a bimetal thermostat on it is open and a forced defrost draws nothing. Let the unit run in cooling until the coil is frosting and well below freezing, often half an hour or more, then put it into defrost: advance the timer to the defrost click, or start a forced defrost from the board's service mode per the tech sheet.
- Clamp the heater lead at a reachable point in the harness, hands clear of the heater and fins, and confirm current flows. The current will be close to what the heater's rated watts divided by line voltage predicts.
- Watch the coil clear. With the cover off in a warm room, you will see frost melting from the heater upward. Defrost should end on its own once the area under the thermostat or sensor is clear, the current dropping to zero.
- Confirm the clip is still seated before the cover goes back on, and that water from the melt is reaching the drain pan below, not pooling under the crisper.
- Recheck the next day if you can, or ask the customer to report fresh food temperature in 48 hours. One good defrost proves the parts; it takes several cycles to prove a short defrost will not rebuild ice.
Write on the ticket: device type, room-temperature reading, ice-water reading, clip condition as found, and the defrost current seen at the finish test.
References
- See related: Refrigerator Defrost System Repair Standard; Replace a Refrigerator Defrost Heater
- See related: NTC Thermistors and Oven RTD Sensors Explained; Thermal Fuses, High Limits and Cycling Thermostats in Appliance Heat Circuits
- See related: Estimating Heater Resistance From the Rating Plate Wattage; How to Find and Read the Tech Sheet Before the First Reading
- NFPA 70E, Standard for Electrical Safety in the Workplace, Article 120, current edition (verifying absence of voltage)
- Manufacturer tech sheet for the model: defrost device type, close and open temperatures, and thermistor resistance chart