Skip to content
Ice Maker Specialist
Ice maker owners deciding whether a filter is worth fitting, and which media class actually addresses the problem they can see.8 min read · Updated July 2026

Ice Maker Water Filters by Hardness and TDS: Which Media Stops Scale, Which Only Fixes Taste

Carbon block fixes taste and chlorine and does nothing for scale; polyphosphate stops scale and does nothing for taste; only reverse osmosis meaningfully lowers total dissolved solids. If your water is under about 3.5 gpg the scale case for filtering is weak, and if you're seeing pink film no filter on the market will help because that's airborne biofilm. Roughly 120 gallons of water makes 1,000 lb of ice, which is how you price a cartridge honestly.

By the Ice Maker Specialist Editorial Team

There are two separate problems and people keep buying the filter for the wrong one. Taste problems — chlorine, staleness, a chemical note — are carbon's job. Scale problems — white crust on the evaporator, lengthening cycles, ice that won't release — are a hardness job, which carbon does absolutely nothing about. A carbon block cartridge fitted to solve scale is money set on fire.

So start with a number, not a product. Grab a $12 TDS meter and a $10 hardness strip pack. TDS in ppm tells you roughly how much dissolved material is in the water; hardness in grains per gallon tells you how much of it is calcium and magnesium, which is the part that deposits on cold metal.

One honest warning that saves people a lot of money: below about 3.5 gpg, the scale case for filtering an ice maker is thin. Descale on schedule and skip the cartridge.

  1. Match the symptom to the media

    What you're seeingWhat your water readsMedia that actually fixes itMicron ratingReplace everyCost per 1,000 lb of iceMedia that won't help
    Cloudy or white-cored cubesAny TDS above roughly 50 ppm, plus dissolved airReverse osmosis, and only in a directional-freeze machineMembrane, effectively sub-0.001 µmMembrane 24–36 months; prefilters 6–12 months$18–$35Carbon block alone, polyphosphate, sediment
    Chlorine or stale off-tasteChlorine detectable by smell; TDS irrelevantCarbon block0.5–5 µm6 months, or 200–1,500 gal by cartridge rating$3–$27Sediment, polyphosphate, RO (overkill)
    White crust on the evaporator prongsHardness above 7 gpg / 120 ppmPolyphosphate scale inhibitor, or RONot a filtration mechanism — it's a sequestrant feeder6 months$6–$15Carbon block, sediment
    Pink or orange film in the binAny water at allNone — this is airborne Serratia biofilm, not a water faultn/an/a$0Every filter on this table, without exception
    Cycles lengthening, ice sticking at harvestHardness above 7 gpg with visible depositPolyphosphate plus a shortened descale intervaln/a6 months$6–$15Carbon block — it changes nothing about this symptom
    Grit or sediment settling in the binWell supply, or TDS above 250 ppm after plumbing workSpun polypropylene sediment prefilter5 µm6–12 months$2–$6Carbon alone — it clogs and you lose flow
    Metallic taste after new pipeworkTDS spike against your usual baselineCarbon block, after flushing the line hard0.5–5 µm6 months$3–$27RO — expensive answer to a temporary problem
    Find your visible symptom in column one and buy only the media in column three. If your symptom is the pink-film row, close the shopping tab and go clean the machine instead.
  2. What each media class does and doesn't do

    MediaRemoves or addressesDoes not removePressure dropRequiresTypical cartridge lifeRight for
    Spun polypropylene sedimentGrit, rust flakes, pipe scale particles down to 5 µmAnything dissolved — hardness, chlorine, TDSLow until loaded, then steepNothing special6–12 monthsWell supply, or the first stage of any multi-stage setup
    Carbon blockChlorine, chloramine on some blocks, taste and odour compounds, some particulateHardness, TDS, dissolved mineralsModerate — 5–15 psi through a loaded cartridgeAdequate inlet pressure6 months or 200–1,500 galTaste complaints. This is what a refrigerator cartridge mostly is
    Polyphosphate sequestrantKeeps calcium and magnesium in solution so they don't deposit on cold surfacesNothing — it doesn't remove hardness, it holds it in suspensionLowContact time; sized to flow rate6 monthsHard-water owners who want longer descale intervals
    Reverse osmosis90–98% of dissolved solids, including hardnessDissolved gases entirely; won't clarify ice in a non-directional machineVery high — this is its defining constraint40 psi minimum at the membrane; produces reject waterMembrane 24–36 months, prefilters 6–12Extremely hard water, or clear-ice machines where TDS matters
    Softened water as an inlet sourceExchanges calcium and magnesium for sodium, so no scale depositsTDS — softened water often tests higher, not lowerNegligibleAn existing softenern/aScale prevention only, and with a real caveat below
    UV sterilisation on the inletLive organisms in the supply lineScale, taste, TDS, and anything already inside the machineNonePower, clear water upstreamLamp annuallyWell supply with a microbiological result. Rare in this application
    Read the second and third columns together before buying. If the thing you want gone appears in the third column, that media is the wrong purchase no matter how good the reviews are.

    The softener row needs its caveat spelled out. Fully softened water prevents scale, which is genuinely useful, but it makes ice that's noticeably softer and more prone to fusing into a slab in the bin, and at high sodium levels there's a faint taste. Ice machine manufacturers have warned about feeding fully softened water for years. If you have a softener, the better arrangement is an unsoftened line to the ice maker with polyphosphate treatment, rather than softened feed.

  3. Cloudy ice: the filter won't fix it

    This is the most-sold false promise in the category. Cloudiness in a cube is trapped air plus precipitated minerals, and the air is the bigger contributor by far. When water freezes inward from every side simultaneously — which is exactly what a bullet machine's prongs and a standard cube mould do — the dissolved gas gets driven to the centre and locked there as a white core.

    Filter that water to 8 ppm TDS with reverse osmosis and a bullet machine will still make cloudy bullets, because you haven't changed the freezing direction. What produces clear ice is slow, one-directional freezing that pushes impurities ahead of the ice front and out of the finished cube. That's a machine architecture, not a water treatment.

    Where RO genuinely earns its place is inside a machine that already freezes directionally. There, lower TDS means a cleaner-looking cube and less mineral haze at the tail of the freeze. Fitting it to a $120 bullet machine achieves nothing you can see.

  4. Cost per 1,000 lb, worked properly

    Ice is about 8.34 lb per gallon of water, so 1,000 lb of ice takes roughly 120 gallons through the machine. That single conversion is what makes filter pricing comparable, and almost nobody does it.

    A typical refrigerator cartridge rated 200 gallons over six months, at a $40–$55 replacement, works out to about $24–$33 per 1,000 lb of ice — but only if the ice maker is the only thing drawing through it, which it never is. Share it with the drinking-water dispenser and the real figure drops.

    An inline carbon cartridge rated 1,500 gallons at $25 lands nearer $2 per 1,000 lb. A polyphosphate cartridge at $30 for six months on a household ice maker producing 15 lb a day comes to roughly $11 per 1,000 lb. RO with prefilters amortised across membrane life sits in the $18–$35 range.

    The point isn't the exact figure — it's that inline cartridges are usually five to ten times cheaper per pound than proprietary refrigerator cartridges, and a plumbed countertop or undercounter machine can take one.

  5. The pressure trap with reverse osmosis

    Plumbed ice makers specify an inlet pressure range, and 20 to 120 psi is the near-universal figure. RO permeate does not naturally live in that range. It comes off a storage tank at maybe 5 to 40 psi, and that pressure falls as the tank empties.

    What happens next is a classic misdiagnosis. The inlet valve opens for its fixed fill time, less water arrives than the machine expects, and you get small, hollow, half-formed cubes. Owners test the valve, find it fine, and start suspecting the thermal system. The fault is upstream and it's pressure, not the valve.

    Two fixes. Fit a permeate pump or a small booster to hold at least 40 psi at the valve, or don't feed the ice maker from the RO tank at all. Manufacturers who allow RO feed generally state 40 psi minimum for exactly this reason — check the spec plate before you plumb it, not after.

  6. Portables are a different situation entirely

    A countertop portable has no plumbed inlet. You pour water into an open reservoir, so there is nothing to fit a cartridge to — the filtering has to happen before the water reaches the jug.

    That means a pitcher filter or a tap-mounted carbon unit, and both are perfectly adequate for taste. Neither addresses hardness, so in hard water your descale interval stays exactly where the hardness table puts it. Filtering does not buy you longer between cleanings unless the media is a sequestrant, and pitcher filters are not.

    If you're on very hard water and running a portable, the genuinely effective move is filling it with RO or distilled water from a jug. It's tedious, and it works — no scale reaches the evaporator at all. Whether that trade is worth it depends on how much you dislike descaling every three weeks.

  7. When no filter is the right call

    Soft municipal water under about 3.5 gpg with no chlorine complaint needs nothing. Descale quarterly, sanitise on the same schedule, and put the cartridge money toward a machine that isn't fighting a blocked vent.

    Skip the filter too if your only symptom is the pink film — that's a cleaning problem wearing a water problem's clothes, and buying a cartridge for it delays the actual fix by a month.

    And if what's really bothering you is output rather than water quality, filtering won't move that number either. Weigh 24 hours of production against nameplate first. Nine times out of ten a machine that's down on output has a dusty condenser or a hot kitchen, and no cartridge on any shelf addresses either one.

More guides