Sediment and carbon filtration
Strips particulate, chlorine and the taste and odour compounds that travel with it, so the water tastes of nothing before the coffee is added to it.
Two stages doing two unrelated jobs, and they are routinely treated as one. The sediment stage is mechanical: a depth or pleated medium that physically captures particles larger than its rating, protecting everything behind it from abrasion, blocked solenoids and a scoured pump. The carbon stage is not filtration at all — it is ADSORPTION. Chlorine and organic taste and odour molecules stick to the enormous internal surface area of activated carbon as water passes through it, and two consequences follow that operators miss constantly. The first: adsorption is a finite-capacity process. The medium fills up. When it is full the water still flows, still looks clean, still leaves no visible sign of anything, and is no longer being treated — which is why capacity is rated in volume rather than in months, and why a calendar-only change schedule is a schedule for a bar whose volume nobody counted. The second, and the reason this entry exists in a category about scale: CARBON DOES NOTHING TO HARDNESS. Calcium and magnesium pass straight through it, unchanged and undiminished. A carbon filter protects the TASTE of the water; it does not protect the BOILER. A bar can have genuinely excellent-tasting water from a carbon cartridge and still be destroying an element month by month, and nothing in the cup will warn anybody. Removing chlorine also removes a disinfectant residual, which is precisely the point for flavour and worth understanding for hygiene: everything downstream of the carbon is now an unprotected surface, so a line that stagnates over a closed weekend is a line worth flushing before service.
- Fit carbon as the baseline on every bar, and never as the answer to hardness. It belongs on the line because untreated chlorine ruins the taste of everything and because it protects the membrane or resin of whatever sits behind it. It does not touch calcium and magnesium, so a carbon-only chain on a hard supply is a bar with lovely water and a dying element. What the supply test said is what decides whether anything else is needed.
- Size the cartridge on VOLUME against your own draw, not on the calendar and not on the size of the venue. Capacity is rated in litres passed, and litres passed depends on drink mix, ice, tea and how many machines share the line — a small bar with an ice machine and a boiler on the same cartridge can outrun a larger bar without them.
- Check flow at PEAK SIMULTANEOUS demand, not against one machine's figure. The espresso machine refilling while the boiler refills while the ice machine calls is the real load, and it happens at exactly the moment nobody can afford a slow refill. A cartridge chosen on capacity alone can be right on volume and wrong on flow.
- Treat pressure drop as an operational specification and not a footnote. Restriction rises as the cartridge loads, so the flow you commissioned is the best flow the system will ever have. If the machines are marginal on day one they will be worse by the change interval.
- Insist on a head with an isolating valve you can reach and a flow meter you can read without moving anything. A cartridge that requires the counter to be pulled out to be changed is a cartridge that gets changed late, and every argument in this category depends on it being changed on time.
- Decide deliberately WHICH lines are filtered. The espresso machine and the hot water boiler are obvious. The ice machine is the one that gets forgotten, and ice is a food that goes straight into the glass — untreated ice will carry the taste of the building into every cold drink on the menu, and scale will do to the ice machine's evaporator what it does to a boiler.
- Put the change in the maintenance schedule with a volume, an owner and a date, and log a reading at each change. A carbon cartridge gives no warning: it does not slow, it does not discolour the water, it simply stops adsorbing. The schedule is the only mechanism that catches it.
- Confirm the connection, the available supply pressure and any backflow requirement with a licensed plumber before ordering, and let the local authority govern. What a site can deliver in pressure is a property of the building, not of the cartridge, and it decides whether the chain you have drawn is buildable at all.
Upstream of everything and downstream of nothing except the supply test that specified it. Physically it belongs under the counter or in the back-of-house run, close enough to the machines that pressure is not lost in a long branch and accessible enough that a change is a five-minute task rather than a job. Every other item in this category sits behind it: a membrane fed unfiltered chlorinated water is a membrane with a short life, and resin is no different.
- Treating carbon as scale protection. It protects taste; hardness passes straight through it, and this single misunderstanding accounts for more scaled boilers than any other.
- Changing on the calendar when capacity is rated in volume, on a bar whose volume changes with the season.
- Skipping the sediment stage and letting the carbon do both jobs, then wondering why the interval calculated at commissioning never held.
- Setting the bypass at commissioning and never re-checking it, so a supply everyone calls treated is quietly part untreated.
- Filtering the espresso machine and leaving the ice machine on the raw supply, when ice goes straight into the glass.
- Choosing a cartridge on capacity and never checking flow at peak simultaneous draw, then diagnosing the slow refill as an undersized machine.
- A plumbed supply with an isolating valve reachable without dismantling the counter, and enough pressure at peak simultaneous draw to feed everything behind the cartridge. Available pressure and the connection itself are a licensed plumber's determination and the local authority governs. Which lines are filtered is a design decision made from the supply test, not from the machine manuals.
- None in normal running, but a cartridge change releases the water standing in the housing. Somewhere to catch it, or a change that happens over a tray, is the difference between a routine task and a task people avoid.
- The cartridge body plus the swing room to remove it — a filter that fits the cupboard but cannot be unscrewed inside it has not fitted. Measure the removal arc, not the housing.
- Access to the head, the bypass and the meter without moving equipment, and enough light to read the counter. Anything installed behind a fixed appliance is changed late and read never.
- Sediment pre-filter
- A depth or pleated cartridge sitting first in the chain, before anything else.
- The cheapest cartridge in the chain protecting the most expensive ones behind it. Skip it and the carbon does the sediment's job as well as its own, exhausts early, and the change interval you calculated stops being true. On a supply fed from a building tank or an older network this is not optional.
- Granular activated carbon
- Loose carbon granules in a cartridge body, with water flowing through the packed bed.
- High flow and low pressure drop, which matters when the espresso machine, the boiler and the ice machine all call for water at once. The trade is channelling: water can find a low-resistance path through the bed and leave partly untreated, so the taste result is less consistent than the same volume of carbon in a block.
- Carbon block
- Carbon compressed into a solid form, so every path through it is the same length.
- No channelling and finer particulate capture in the same stage, so the water is more consistent and often needs no separate sediment cartridge on a clean supply. The cost is restriction: pressure drop is higher and rises as it loads, and a restricted supply shows up not as bad taste but as a machine that cannot refill during a rush.
- Filter head with bypass and flow meter
- The mounting hardware the cartridge screws into, carrying an isolating valve, a blending bypass and a volume counter.
- The counter is what converts the change interval from a calendar guess into a measured decision, and on a bar whose volume varies by season that is the difference between changing early and changing late. The bypass lets a deliberate proportion of untreated water past — a treatment decision wearing the appearance of a plumbing detail. A bypass set at commissioning and never revisited is one of the quietest ways a treated supply drifts.
- The building comes back into the cup — a chlorine or swimming-pool note in the americano and the tea before anyone notices it in the espresso.
- The carbon has reached capacity. It gives no other warning: flow does not fall, the water does not discolour, adsorption simply stops.
- Drink quality across the whole menu at once, and it is diagnosed last because every other variable is unchanged. The long drinks show it first because there is the most water in them.
- Refills slow at peak; the boiler takes longer to recover and the machine sits waiting on water at the worst moment.
- A loaded sediment stage, or a carbon block chosen on capacity without checking flow at simultaneous demand.
- Throughput, precisely when throughput is the only thing that matters — and it reads as an undersized machine, so the money goes to the wrong place.
- Water tastes clean and correct, and the boiler scales anyway.
- Carbon-only treatment on a hard supply. Adsorption does not touch calcium and magnesium, and nothing in the cup reports what is happening to the element.
- Money and downtime, cumulative and delayed. This is the failure the whole category exists to prevent, and it is the one a carbon filter is most often wrongly believed to have already solved.
- Nobody can say how much water has passed the cartridge, and the change date keeps moving.
- No flow meter, a calendar-only schedule, and no named owner for the task.
- The protection you paid for becomes a matter of luck, and both errors cost: changed early is money thrown away, changed late is an untreated supply nobody knew was untreated.