Why an Ice Maker Keeps Running and Makes Nothing

This page contains affiliate links. If you buy through one we may earn a commission, at no cost to you.

Before the fix, which machine is it? Almost every answer to this question is about an ice maker built into a refrigerator, so it tells you to check the water line, the fill valve, the fill tube for a plug of frozen water, the water filter, and the bin level sensor. That advice is correct and it is also completely unavailable if your machine sits on a counter, because a countertop ice maker has none of those five parts. It has a reservoir you fill by hand, a small pump, a set of chilled prongs and a compressor.

So on a countertop machine the entire received checklist is inapplicable, and the failure is nearly always thermal. The good news is that the symptom in the question, running and making nothing, is specific enough to diagnose, and the arithmetic is available from one published figure.

What 100 pounds a day actually commits the machine to

A Kismile stainless steel ice maker machine publishes up to 100 pounds of ice per 24 hours, 45 cubes a batch in about 11 to 20 minutes, adjustable thickness and production time, and a cabinet 15.28 by 14.57 by 31.42 inches weighing 59.4 pounds. The first of those figures fixes the whole thermal problem.

A hundred pounds is 45.36 kilograms. Cooling that much water from a nominal 20 degrees Celsius down to freezing takes 45.36 times 4,186 times 20, which is 3.80 megajoules. Then freezing it takes 45.36 times 334,000, which is 15.15 megajoules. The second number is four times the first, because the latent heat of fusion is 334 kilojoules per kilogram while cooling through 20 kelvin costs only 84. Four fifths of the work is the phase change, and nothing about the water's starting temperature changes that.

Add them and the machine has to remove 18.95 megajoules a day. Spread over 86,400 seconds that is 219 watts of cooling, continuously, with no pauses. That is the duty the published output commits to, and it is the number every other figure on this page comes from.

Worth noting that the record's figures are mutually consistent, which is unusual enough to say. At the fast end of the published batch range, 11 minutes gives 131 batches a day and 5,891 cubes, so each cube would weigh 7.70 grams to reach 100 pounds. At the slow end, 20 minutes gives 3,240 cubes at 14.0 grams each. Both are plausible masses for bullet ice, which means the daily output and the batch time describe the two ends of the adjustable thickness setting rather than contradicting each other. They are not, however, the same setting, so 100 pounds a day and an 11 minute batch are not a combination to expect simultaneously.

The compressor has to restart against the pressure it just built

Now the mechanism behind the symptom. A refrigeration compressor is a motor sealed inside a can with a pump, and when it stops, the high pressure it created does not disappear. It equalises slowly through the system. Ask it to restart before that has happened and the motor has to develop enough torque to compress against a pressure difference it is not designed to start under.

An induction motor at a standstill is close to a short circuit, electrically: there is no back electromotive force until it turns, so the only thing limiting the current is the winding resistance and leakage. A small hermetic compressor therefore draws four to six times its running current for the moment it takes to get going. If it is running at something like one and a half amps, the starting draw is six to nine. That is fine for a fraction of a second, which is all it normally lasts.

If the motor cannot get going, that current does not stop. It continues, and the windings have nothing to dissipate it into but themselves. Which is why every one of these machines contains a thermal cutout bonded to the compressor shell, a bimetallic disc that opens when the shell gets hot, and which then has to cool down before it will close again. Cooling a lump of metal inside an insulated can takes minutes.

So the deliberate waiting period after a power interruption, which feels like the machine being slow, is the single most protective behaviour it has. Short cycling a compressor is the one thing that genuinely destroys it, because each stalled restart is a brief overcurrent event in the windings, and the damage accumulates. The right response to a machine that is not making ice is therefore not to switch it off and on repeatedly, which is the most common instinct and the most harmful one.

Our catalogue contains exactly one page that treats this as a feature worth paying for, and it is not about ice. Our roundup of BBQ temperature controllers describes a unit with customizable compressor delay that protects connected equipment from short cycling, and another with compressor delay that prevents rapid short cycling. That is the same protection, sold explicitly, on a device whose only job is to switch something else on and off. It is the clearest statement of the principle anywhere on this site.

Why a warm room defeats it three separate times

Here is where the arithmetic earns its keep, because the ambient temperature attacks the machine from three independent directions and they compound.

First, the pressure it has to start against. A condenser can only reject heat to air that is colder than it is, so the condensing temperature always sits above the room, and the pressure in the high side is the refrigerant's saturation pressure at that temperature. Saturation pressure rises steeply with temperature. Running Clausius-Clapeyron for isobutane, which is the refrigerant in most machines of this class, between condensing temperatures of 35 and 45 degrees Celsius gives a pressure ratio of 1.266. So a room 10 kelvin warmer raises the pressure the compressor must restart against by 26.6 percent.

Second, the cooling it can produce. The best possible performance of any refrigeration cycle is the cold temperature divided by the temperature difference it is pumping across. With an evaporator at minus 10 and a condenser at 35, that ceiling is 263 over 45, which is 5.84. Move the condenser to 45 and it is 263 over 55, which is 4.78. The ratio is 0.818, so 18.2 percent of the available cooling has gone, and a real machine with real compressor losses does worse than that.

Third, the load itself. A warm room means warm water in the reservoir. Water arriving at 30 degrees instead of 20 adds 45.36 times 4,186 times 10, which is 1.90 megajoules a day, or 22 watts on top of the 219. Ten percent more to remove.

Three penalties, none of which causes the others. The machine is starting against a higher pressure, with less capacity, against a larger load. The worklist framing for this page said a warm room defeats these machines twice over. It is three times, and the third one is the easiest to forget because it does not feel like it belongs to the refrigeration system at all.

Which is exactly how it runs and makes nothing

Put the capacity loss together with how these machines actually form ice and the symptom falls out. Our own guide to home ice makers, how they work and the frozen storage myth describes the process correctly: They circulate water over chilled metal prongs or a plate, freezing thin layers in minutes rather than hours. The cubes then drop into a storage bin. And it gets the other half right too, which most coverage does not: The storage bin is insulated, not refrigerated, so ice left sitting gradually melts. That melt water is usually recycled to make fresh ice.

That is the loop that produces the complaint. The machine freezes a layer on the prongs for a fixed cycle time, then briefly warms them to release it. If the cooling available during the freeze cycle has dropped, the layer that forms is thinner. Thin enough, and it does not survive the harvest: it releases and melts on the way down, or in the basket within a minute, and the water returns to the reservoir to be frozen again. The machine has completed a textbook cycle. The pump ran, the compressor ran, the harvest ran, and the basket is empty.

So the machine sounds perfectly healthy because it is perfectly healthy. Nothing has failed. It is running a cycle that is no longer long enough for the conditions it is in, and the adjustable production time its own record advertises is precisely the control that fixes it: a longer freeze cycle builds a thicker layer. Our Kismile stainless steel ice maker review covers that control, and our comparison of countertop versus refrigerator ice makers is the page to read on why the two machine types fail in completely different ways, which is where this page started.

The melt-back figure is worth putting a number on so it is not overstated. For an insulated bin of around a quarter of a square metre at a plausible one and a half watts per square metre per kelvin, a 20 degree room drives 7.5 watts in, which melts about 4.3 pounds a day. At 32 degrees it is 6.8 pounds. That is real but it is not 100 pounds, so melt alone does not explain an empty basket. The capacity collapse during the freeze cycle does.

The honest limit, and the free technique

The limit here is the most complete of any cluster in this batch. The record publishes an output, a batch count, a batch time, dimensions and a weight. It publishes no wattage, no current, no voltage, no refrigerant, no coefficient of performance, no ambient temperature that the 100 pound figure was measured at, and no restart delay. The coefficient of performance of 2 used above is an assumption and is labelled as one, and the ambient sensitivity is the mechanism rather than a specification, because the specification does not exist. There is also only one ice maker in our catalogue, so nothing here can be cross checked against a second product the way the heater and cable arguments were.

Which makes the ambient temperature the one thing you can control, and that is where the free technique is. The condenser has to dump 329 watts into the room: the 219 watts it removed from the water, plus roughly 110 watts of the compressor's own work, since all of that ends up as heat too. Three hundred and twenty nine watts is three good incandescent lamps, and it has to leave through the grille at the back.

Work out what that needs. To carry 329 watts away while letting the air rise only 3 kelvin takes 0.109 kilograms of air per second, which is about 193 cubic feet per minute. Allow a 5 kelvin rise and it is 116, and a 10 kelvin rise, which is already enough to trigger all three penalties above, still needs 58. So the machine needs a real and continuous flow of room air past its condenser, and a cabinet, an alcove or a gap of two inches behind it cannot provide one. Pulling the machine out from the wall is free and is the single largest intervention available in this subject, because it is the only one that attacks the condensing temperature that all three penalties are driven by.

Three more that cost nothing. Fill it with cold water rather than whatever is in the tap, which directly removes the third penalty. Give the basket a chance by taking ice out and putting it in a freezer if you need a stock, since the bin was never refrigerated and the guide above is right that expecting it to store is the commonest disappointment. And if it has stopped, leave it alone for five minutes rather than cycling the switch, because the thermal cutout needs to cool and the high side needs to equalise, and repeated restart attempts are the one thing capable of turning a machine that was working into a machine that is broken.

For the general case of a motor that stops for thermal reasons rather than for the reason printed on the box, our page on why a 180 PSI inflator stops before 60 PSI works through the same family of failure on a compressor you can hold in one hand.