The short answer: For limescale, hardness matters, more precisely its carbonate portion; the SCAE warns of corrosion when alkalinity is too low (below 40 ppm CaCO3). For taste, a University of Copenhagen study found that bicarbonate mattered more (Rune et al., Journal of Food Science, 2026, one coffee, French press). Adding 240 mg/l bicarbonate changed the coffee clearly; calcium up to 264 mg/l and magnesium up to 80 mg/l did not do so on their own, although calcium in the duo-trio procedure came in just above the significance threshold at p = 0.054. The SCAA, SCAE and Wellinger give different target ranges for hardness, and how often you descale is stated in your machine’s manual.
Finding your hardness and setting it on the machine
Ask your water supplier first. The operator of a central water supply system must publish the water hardness online under § 9 of the German Detergents and Cleaning Agents Act (Wasch- und Reinigungsmittelgesetz, WRMG), along with the calcium, magnesium and potassium content in mmol/l and mg/l (Drinking Water Ordinance, 2023 version, § 46 para. 1 no. 3). These rules apply in Germany. This provision does not require bicarbonate or acid capacity. Siemens and Jura set the hardness in °dH (German degrees of hardness); if your supplier publishes mmol/l, convert at 5.608 °dH per mmol/l (SCAE table 1). Hamburg Wasser, the example further below, publishes the value per waterworks, not per city: so look for the works that supplies your house.
The indicator strip is the second route, above all for setting the machine. Three manufacturers describe it similarly and allow for a waiting time of one minute. De’Longhi dips the reaction strip for about one second, and after about one minute one to four red squares appear, each corresponding to one level. Siemens has you dip the strip briefly, let it drip off and wait one minute. With Jura you hold the Aquadur indicator stick under running water for one second, shake it off and read it after about one minute from the colour changes and the pack description.
For the EQ600, Siemens assigns four levels (1 to 7, 8 to 14, 15 to 21 and 22 to 30 °dH, level 4 from the factory) and writes that the correct setting is important so that the appliance prompts you to descale in good time. Our conclusion: the machine combines the set hardness with operating time or water volume; a value that is too low pushes the prompt back, and one that is too high brings it forward. The Philips LatteGo 5500 has five hardness levels, with level 4 for hard water set from the factory, and with the correct setting you do not descale more often than necessary, according to our German review. Jura allows 1 to 30 °dH. The other settings of a bean-to-cup machine are covered in the guide Bean-to-cup machine settings.
With a filter inserted, setting the hardness sometimes does not work at all. On the Nivona NICR 790, according to our German review, the hardness can only be set in operation without the filter, and on the Melitta CI Touch not at all with the filter. The Melitta Solo jumps to level 1 by itself when the PRO AQUA filter cartridge is inserted.
Total hardness, carbonate hardness, acid capacity
Total hardness is the sum of the alkaline earth ions, mainly calcium and magnesium (Spektrum Lexikon der Chemie, a German chemistry encyclopaedia); the Umweltbundesamt (German Federal Environment Agency) likewise describes water hardness as dissolved calcium and magnesium. Carbonate hardness, also called temporary hardness, is, according to the SCAE Water Chart (Wellinger, Smrke and Yeretzian of the ZHAW, the Zurich University of Applied Sciences; the document gives no year, the PDF metadata say 2016, and the file was uploaded in 2017), only the part of the total hardness that corresponds to bicarbonate. The rest is called non-carbonate or permanent hardness, mostly comes from sulphates or chlorides and does not form limescale. According to the encyclopaedia, the name temporary comes from the fact that on boiling the bicarbonates of calcium and magnesium decompose and settle as sparingly soluble carbonates, as scale.
The encyclopaedia considers it more exact to state the acid consumption (alkalinity) instead of the carbonate hardness, and the SCAE authors write that carbonate hardness is not necessarily the same as alkalinity. For the amount of limescale, the SCAE says: the lower of the two values, hardness or bicarbonate equivalent, determines how much limescale can theoretically form at most.
Hamburg Wasser shows what this looks like in a real analysis, for the Schnelsen groundwater works (2025 analysis values, the analysis applies only to this waterworks): total hardness averaging 13.96 °dH (minimum 13.7, maximum 14.2), carbonate hardness 9.9 °dH, acid capacity to pH 4.3 of 3.54 mmol/l, plus 85 mg/l calcium and 9 mg/l magnesium. Under the law this water is medium. Calculated with the SCAE factors (7.147 mg/l calcium and 4.334 mg/l magnesium per °dH), calcium and magnesium give 13.97 °dH, which matches the stated total hardness. The acid capacity corresponds to roughly 216 mg/l bicarbonate, and with the SCAE factor of 2.804 °d per mmol/l, 3.54 mmol/l is roughly 9.9 °d, that is, the stated carbonate hardness. Between total and carbonate hardness lie a good 4 °dH, so for limescale in this water at most 9.9 counts.
Limescale and corrosion: what concerns the machine
According to the SCAE chart, limescale mainly consists of calcium carbonate, and its formation depends on carbonate and calcium. The Stadtwerke Augsburg write in an undated information sheet that significant limescale deposits and scale only occur when water is heated above 60 °C. Our conclusion: in a coffee machine, limescale sits mainly where heating takes place, that is, in the boiler, thermoblock and heating element. Siemens does, however, also mention limescale deposits in the water tank (EQ600 manual, chapter on determining water hardness), so the tank is not exempt.
The SCAE authors describe what scale does as follows: poorer heat transfer, blocked valves and nozzles. Their worked example assumes 100 drinks a day of 0.1 litre each, rinsing included. At 30, 100 and 200 mg limescale per litre of water, 110 g, 365 g and 730 g of limescale can form per year. The manufacturers’ figures are more specific, but undocumented. durgol puts the extra consumption, based on internal trials, at 8.1 per cent energy per millimetre of limescale. In the manual of the Stelio Aroma, WMF writes that limescale leads to longer brewing time, poorer coffee quality, energy loss and a shorter service life. Jura states that narrowed pipes lengthen the heat-up time and increase electricity consumption.
The counterpart is corrosion, and it has different causes from limescale. According to the SCAE it threatens at alkalinity below 40 ppm CaCO3 and with a high chloride, sulphate or nitrate content.
For health, hard water is harmless according to the Umweltbundesamt: with a higher calcium and magnesium content it is not harmful but to be viewed positively.
What studies and expert bodies say about taste
The Deutscher Kaffeeverband (German Coffee Association) writes in its FAQ that a coffee drink consists of about 98 per cent water, that salts and minerals are important flavour carriers, and that the water should contain oxygen and minerals and be heated quickly. Hard, limescale-containing and chlorinated water is bad, it says, because it has a negative effect on taste. Appliance manufacturers also comment on taste: Jura writes that limescale in small amounts is a flavour carrier, while too much leads to a flat aftertaste, and according to Siemens the limescale content influences aroma and crema. Brita separates the two: total hardness does not influence taste, but temporary hardness does, namely in food and drinks made with hot water. According to our German review of the water filter DLSC002, De’Longhi speaks on the German product page of the intense taste and natural aroma of a good coffee.
One possible explanation comes from a computational model. We only know Hendon, Colonna-Dashwood and Colonna-Dashwood (2014, Journal of Agricultural and Food Chemistry 62(21), 4947 to 4950) from the abstract; the full text was blocked. Using density functional theory, the authors calculated how strongly sodium, magnesium and calcium ions bind to five coffee acids, caffeine and eugenol, and derived from this a mechanism and an ideal mineral composition. As reported by Rune et al., the work remained theoretical and no sensory data were collected to check it; in the model magnesium binds more strongly to flavour compounds than calcium, which suggests better extraction.
Measurements on coffee that we have found do not support this. Bratthäll, Figueira and Nording (Heliyon 10(5), e26625, 2024) added magnesium and calcium chloride before and after brewing and measured citric, malic, lactic and quinic acid with GC-MS and NMR. At concentrations typical of drinking water they found only limited differences in acid content, and clearer ones at ten times higher concentrations. Because adding before and after brewing mostly gave similar contents, it seems likely that acid extraction proceeds independently of water composition, although interactions after brewing could still influence taste. The SCAE authors, too, could not confirm the assumed effect of hardness on extraction yield (Hendon et al. 2014) in their own series of trials with a coffee refractometer. By their own account, they nevertheless saw a clear influence on flavour balance and aroma.
Marco Wellinger (ZHAW) writes in SCA Magazine 25 (issue 9, 2019 according to the page metadata) that there are no clear answers on total hardness. Observations mention over-extracted taste above 250 ppm CaCO3 and under-extracted taste below 40, but measurements suggest that total hardness has an insignificant influence on extraction efficiency between 20 and 250 ppm CaCO3.
The most recent work we found is by Rune et al. (University of Copenhagen, Journal of Food Science 91(9), e71434, September 2026, online on 16 September 2026). They brewed a medium-roast Colombian coffee in the French press, 50 g coffee to 900 g water (about 55.6 g/l), 92 °C, 4 minutes, and compared pure water with additions of calcium, magnesium and bicarbonate. These were accompanied by duo-trio difference assessments in two sessions with 31 and 25 untrained people and two descriptive sensory analyses with 10 trained panellists each.
Hardness alone produced no noticeable taste differences, neither through calcium up to 264 mg/l nor through magnesium up to 80 mg/l; for calcium only the lowest level (15 mg/l, p = 0.022) was significant, with no dose-response pattern, and the highest (240 mg/l) came in just above at p = 0.054 (the authors write that it approaches significance). Bicarbonate, by contrast, changed the cup: 120 mg/l was just significant (p = 0.049), 240 mg/l clearly so (p below 0.001). In the second descriptive analysis the coffee with 264 mg/l calcium did not differ in any attribute from the coffee made with pure water, whereas an addition of 300 mg/l bicarbonate lowered acidity on the 15 cm scale from 12.0 to 2.9, fruitiness from 10.2 to 2.4 and complexity from 8.7 to 4.6, while bitterness rose from 6.9 to 9.5. The 300 mg/l is the amount added; in the water the authors titrated 253 mg/l for this sample, and 286 mg/l after correcting for the method’s under-recovery (13 per cent). The pH of the coffee rose from 4.73 (pure water) to 5.19 (table 6 of the study, same coffee); for three further coffees it averaged 5.53 with 300 mg/l bicarbonate against 4.88 with pure water (table 7). The authors interpret the effect through buffering: bicarbonate neutralises organic acids. It was added as sodium bicarbonate, so the sodium came into the cup too, and the study describes no control with sodium alone.
For context: by our conversion with the SCAE factors, 120, 240 and 300 mg/l bicarbonate are 5.5, 11.0 and 13.8 °d alkalinity. The authors give 200 to 400 mg/l bicarbonate as typical for Danish tap water and 110 mg/l calcium and 20 mg/l magnesium as typical for Copenhagen tap water. The water from Schnelsen, with roughly 216 mg/l bicarbonate, lies at the lower edge of this Danish range.
The authors name the study’s limits themselves: one coffee, one roast level, one brewing method (by their statement, the findings cannot be transferred to other origins and roast levels), no measurement of TDS and extraction yield (so small extraction differences are not ruled out), no survey of consumers. When calcium and bicarbonate were mixed, calcium carbonate precipitated and lowered the effective bicarbonate. Their conclusion: the narrow hardness windows in brewing water recommendations reflect machine protection rather than a demonstrable taste advantage, while the range for bicarbonate remains sensorially meaningful. Experienced cuppers, they concede, might perceive subtle differences.
This fits older sensory work that Rune et al. cite (second hand; we have not opened the originals): Lockhart, Tucker and Merritt (1955) and Pangborn, Trabue and Little (1971) described brews from distilled water as too sour and brews from water with a high carbonate content as flat and bitter.
Wellinger explains the mechanism like this: the perceived acidity of a cup corresponds to the extracted acids minus the alkalinity of the water, and alkalinity matters more than pH. The beverage ratio differs between preparations by up to a factor of 10 (ristretto 1.5, filter 15), so with espresso preparations only about a tenth of the alkalinity per coffee acid is available compared with filter coffee, and espresso tolerates more of it. The SCAE authors add: above 100 ppm CaCO3 alkalinity, neutralising the coffee acids with bicarbonate produces a lot of carbon dioxide, which lengthens the extraction time and can lead to over-extraction, and with more sodium the effect intensifies. According to them, water with high total hardness can also give a high-quality filter brew.
Target values for brewing water: three numerical ranges, two sources without a target value
| Source | Hardness | Alkalinity | Other details |
|---|---|---|---|
| SCAA standard (Beeman, Songer, Lingle 2011, presented by Emma Sage, SCAA, 2013) | Calcium hardness 17 to 85 mg/l, target 51 to 68 mg/l (our conversion: 0.95 to 4.8 and 2.9 to 3.8 °dH) | Total alkalinity about 40 mg/l | pH 6.5 to 7.5 (target 7), TDS 75 to 250 mg/l (target 150), chlorine 0, sodium below 30 mg/l (target 10 mg/l) |
| SCAE Water Chart (ZHAW authors, summary point 4) | Total hardness 50 to 175 ppm CaCO3 (roughly 2.8 to 9.8 °d, 2.9 to 9.8 in the document) | 40 to 75 ppm CaCO3 (2.2 to 4.2 °d) | pH 6.5 to 8, odourless, free of chlorine, hypochlorite and chloramines, electrical conductivity below three times the alkalinity |
| Wellinger, SCA 25 Magazine, issue 9 (proposal from Wellinger’s research for the ongoing discussion in the SCA standards committee, not an adopted standard) | from 20 ppm CaCO3, lower limit justified purely on sensory grounds | up to 150 ppm CaCO3 | for espresso only |
| Rune et al. 2026 (difference assessment, one coffee, French press) | Calcium up to 264 mg/l and magnesium up to 80 mg/l without noticeable effect | 120 mg/l added bicarbonate just perceptible, 240 mg/l clearly perceptible | no target values, TDS not measured |
| Deutscher Kaffeeverband (German Coffee Association, FAQ) | hard, limescale-containing water is bad | no statement | no figure, no study |
The conversion into °dH is ours, at 17.85 mg/l CaCO3 per °dH according to SCAE table 1. Calcium hardness (SCAA) and total hardness (SCAE) are not the same, because total hardness also includes magnesium, and whether the SCAA values are meant as CaCO3 is stated expressly in the article only for alkalinity. The SCAE gives its values as a recommendation that combines technical and sensory considerations. The graphically displayed core zone is expressly derived for espresso machines and brew boilers, and its edges follow corrosion and limescale limits and the minimum line of Colonna-Dashwood and Hendon. The SCAE authors also write that the data are not sufficient to describe the effect of total hardness and alkalinity on aroma and taste more precisely; the chart is a framework for further work.
Our assessment: none of these sources documents a hardness window that you would have to hit. Wellinger sees hardly any influence on extraction between 20 and 250 ppm CaCO3. For taste we would watch the bicarbonate and not let it get high: the 120 mg/l at which Rune et al. just found a difference correspond to roughly 98 ppm CaCO3, and the SCAE gives 100 ppm as the threshold for the risk of over-extraction.
Hardness ranges and manufacturer scales
The law counts three hardness ranges and calculates in mmol calcium carbonate per litre: soft below 1.5, medium 1.5 to 2.5, hard above 2.5 (WRMG § 9 para. 2). With the factor of 5.608 °dH per mmol/l, those are 8.4 and 14 °dH. On its page about water hardness (as of 28 April 2025), the Verbraucherzentrale (German consumer advice centre) still works with the earlier classification, and WMF and Siemens have four levels with limits at 7, 14 and 21 °dH. Other manufacturers set their own limits: for the Melitta Solo and CI Touch they lie at 10, 15 and 20, for the AromaFresh X, by contrast, at 7, 14 and 21, and for Nivona at 9, 14 and 19. The Lexikon der Chemie lists a five-level classification.
An example: 9 °dH is medium under the law, soft on the Melitta CI Touch, level 2 on Siemens and level 1 on Nivona. How many °dH one level means on De’Longhi or the Sage Barista Express is not stated in the manuals we read. So on the machine, always set the value from the scale given in its manual, not the range under the WRMG.
| Scale | Classification | Source |
|---|---|---|
| WRMG § 9 para. 2 | soft below 8.4 °dH (below 1.5 mmol/l), medium 8.4 to 14 (1.5 to 2.5), hard above 14 (above 2.5) | Law, converted at 5.608 °dH per mmol/l |
| Earlier classification (Verbraucherzentrale) | Hardness range I 0 to 7 °dH soft, hardness range III 14 to 21 °dH hard | Page on water hardness, as of 28 April 2025 |
| WMF Stelio Aroma | soft 0 to 7, medium above 7 to 14, hard above 14 to 21, very hard above 21 °dH | Manual, 1 °dH corresponds to about 0.18 mmol/l |
| Siemens EQ600 | Level 1: 1 to 7, level 2: 8 to 14, level 3: 15 to 21, level 4: 22 to 30 °dH, level 4 from the factory | Manual |
| Melitta Solo | Level 1: 0 to 10, level 2: 10 to 15, level 3: 15 to 20, level 4: above 20 °dH, level 4 from the factory | Manual, according to the review |
| Melitta CI Touch | soft 0 to 10, medium 10 to 15, hard 15 to 20, very hard above 20 °d | Manual, according to the review |
| Melitta AromaFresh X | below 7, 7 to 14, 14 to 21, above 21 dH, level 4 from the factory | Manual, according to the review |
| Nivona NICR 790 | indicator strip with four levels: 0 to 9, 10 to 14, 15 to 19, from 20 °dH, level 3 from the factory | Review (FAQ on water hardness) |
| Sage Creatista Plus | Level 0 below 3, level 1 above 4, level 2 above 7, level 3 above 14, level 4 above 21 dH | Manual |
| Sage Barista Express | Sage levels, not stated in °dH | Manual, according to the review |
| De’Longhi Magnifica Plus | four levels by strip (1 to 4 red squares), without °dH | Manual |
| Philips LatteGo 5500 | five hardness levels, level 4 for hard water from the factory | Manual, according to the review |
| Jura E6 | 1 to 30 °dH adjustable | Manual |
| Spektrum Lexikon der Chemie | very soft 0 to 4.2, soft up to 8.4, moderately hard up to 16.8, hard up to 28, very hard above that | Encyclopaedia, copyright notice of the page 1998, no publication date in the text |
Descaling intervals according to the manufacturer
There is no general figure, and the manufacturers do not even calculate in the same unit: in months (WMF, Emerio, Senseo), in days (Sage), in cups (Nespresso), in servings (Sage Creatista), in brewing cycles (Moccamaster) or in litres per filter cartridge (Jura). Gaggia and Senseo do not graduate by hardness at all. For Nespresso the relationship between cups and hardness can be checked by calculation: 300 cups of 40 ml at 20 °dH are 12 litres, 600 cups at 10 °dH are 24 litres, and litres times hardness gives 240 in both cases. For Nespresso and the Sage Creatista Plus, the interval therefore depends on hardness and water volume at the same time, and a calendar period is only right there if a similar amount runs through every day. The Melitta AromaFresh X signals descaling via a CALC LED, based on the set level.
| Machine | Manufacturer’s statement | Note |
|---|---|---|
| WMF Stelio Aroma | soft 0 to 7 °dH every 6 months, medium above 7 to 14 every 3 months, hard above 14 to 21 every 6 weeks, very hard above 21 monthly | Guide values for daily household use |
| Emerio CME 122933 | at average water hardness every 3 to 4 months | Frequency depends on hardness and use, according to Emerio |
| Gaggia Classic Evo Pro | every 2 months of machine use | not graduated by hardness |
| Senseo Original Plus | flat rate every 3 months | independent of water hardness, according to Versuni |
| Moccamaster KBG Select | every 100 brewing cycles or per pack of filter paper | more often with hard water (manufacturer FAQ) |
| Grundig KVA 6230 | at least once a month | or more often depending on water hardness, an LED signals descaling |
| Sage Barista Express | hardness 4 every 90 days, hardness 6 every 60 days | The unit of the levels is not in the manual |
| Nespresso Essenza Mini | by cups of 40 ml: hard (20 °dH) 300, medium (10 °dH) 600, soft 1,200 | no descaling indicator described in the manual, you have to keep count yourself |
| Sage Creatista Plus | by servings of 40 ml: below 3 °dH 4,000, above 4 °dH 3,000, above 7 °dH 2,000, above 14 °dH 1,500, above 21 °dH 1,000 | Table in the manual |
| Melitta CI Touch | without filter every 3 months, with filter only when prompted | Manual, according to the review |
| Jura E6 | no descaling prompt with the CLARIS Smart+ inserted | Cartridge by litres, see filter table below |
| Severin KA 4810 | no interval prescribed | Manual: no hardness setting, no limescale filter |
Fault patterns: symptom, cause, change
The table lists only causes that a named source supports. What we conclude from them is marked as a conclusion. Further causes of bitterness and sourness in the cup are covered in the guide Coffee tastes bitter or sour.
| Symptom | Possible cause | What you change |
|---|---|---|
| Coffee seems flat, little acidity, rather bitter | A lot of bicarbonate: in Rune et al. 2026 less acidity and fruitiness, more bitterness; according to the SCAE, perceived acidity falls with alkalinity | Look up the acid capacity or bicarbonate in your supplier’s analysis, if it is listed there. Our assessment: for manual brewing, take water with less bicarbonate; for machines, the manufacturers’ requirements in the section Bottled water in the tank apply |
| Coffee tastes too sour | Very low-mineral water: Lockhart 1955 and Pangborn 1971 (as cited by Rune et al.) described brews from distilled water as too sour | No distilled or demineralised water (manufacturer requirements in the section Bottled water in the tank) |
| The descaling prompt comes too late or too often | The set hardness does not match the water (Siemens: the correct setting is needed so that the prompt comes in good time; Philips: so that you do not descale more often than necessary) | Look up the value with the supplier or measure with the strip, then set it again in the machine’s scale |
| Brewing time becomes much longer, the machine makes more noise | Limescale: WMF requires descaling at the latest when brewing time is considerably longer or noise increases | Descale according to the manual |
| Steam too wet (Sage Creatista Plus) | The troubleshooting table in the manual names the water: cold drinking water, never heavily filtered, demineralised or distilled | Choose water according to the manual |
| Hardness cannot be set | Filter inserted (Nivona NICR 790, Melitta CI Touch) | On the Nivona, set the hardness in operation without the filter |
Water filters: manufacturer statements, Federal Environment Agency and consumer advisers
Brita describes the working principle like this: in ion exchange, dissolved ions are swapped for other ions of the same charge, for example two hydrogen ions for one calcium ion, and the plastic beads reduce carbonate hardness and total hardness. The activated carbon removes substances that spoil smell and taste, such as chlorine, without changing the mineral content. Filtered water is not distilled; the process demineralises only partly. By the SCAE’s classification, a softener (exchange for sodium or potassium) lowers the total hardness without changing the alkalinity. A decarboniser (exchange for protons) lowers hardness and alkalinity by the same amount, and a buffered one leaves about 50 ppm CaCO3 of residual alkalinity so that the water does not become acidic and corrosive. Our conclusion: the hydrogen exchange example named by Brita corresponds to the SCAE’s description of a decarboniser, so the alkalinity would fall along with it.
| Filter | Manufacturer’s statement | Source |
|---|---|---|
| Brita MAXTRA PRO | up to four weeks and 150 litres, at 10 to 12 °dH replace after 150 litres, earlier with harder water or more consumption, at least every four weeks | Brita FAQ |
| JURA CLARIS Smart+ (see Jura E6) | up to 65 l at 1 to 5 °dH, up to 60 l at 6 to 10, up to 55 l at 11 to 15, up to 50 l at 16 to 20, up to 45 l at 21 to 25, up to 40 l at 26 to 30 °dH | Jura product page |
| Sage Barista Express | every 90 days or 40 litres, the 40 litres apply at hardness 4, at hardness 2 up to 60 litres | Manual, according to the review |
| Siemens BRITA Intenza | 50 litres per cartridge or at most 2 months, aroma ring position A at hardness level 1 and 2, B at level 3, C at level 4 | Review |
| Philips AquaClean | up to 625 cups per filter, replace after 3 months at the latest | Review |
| De’Longhi DLSC002 | replace after about 2 months, at most 3 weeks when idle with the filter inserted | Review |
| Melitta Solo | with the PRO AQUA filter cartridge, replace every two months | Review |
The harder the water, the fewer litres a cartridge manages. Our German review of the Philips AquaClean points out that the capacity depends physically on litres times degree of hardness, and Philips advertises 5,000 cups without descaling, which according to the review only holds after eight filters. For the Jura E8 with a 1.9 litre tank capacity without a filter, the up to 65 and up to 40 litres of the CLARIS Smart+ give, by the review’s calculation, roughly 34 tank fills per cartridge in soft water and 21 in very hard water, as a rough guide, because JURA does not state how much fits in with the cartridge inserted. On its US site De’Longhi gives about 20 per cent less frequent descaling per fresh filter; the German site and Siemens speak only of less frequent descaling, without a figure. The filter does not replace descaling (De’Longhi, Siemens), and before descaling it must come out of the tank. Jura goes further: with the CLARIS Smart+ inserted, no descaling prompt appears on the E6, according to the manual.
The Umweltbundesamt and the Verbraucherzentrale take a more sober view of filters. The Umweltbundesamt regards additional treatment of drinking water in the building as generally unnecessary and advises against additional purification or treatment devices. Bacteria could multiply in filters that are not replaced regularly, and overloaded filters or ion exchangers could release previously removed substances again in higher concentration. The exception: for very hard water (calcium 2.5 mmol/l or more), DIN 1988-200 provides for softening or stabilisation if the hot water is heated above 60 °C. That concerns the drinking water installation in the house, not the jug filter. The Verbraucherzentrale writes that from a health point of view you do not generally need a filter for tap water. Anyone who uses a table water filter should follow the instructions, replace the cartridge regularly, clean the filter regularly and keep it in the refrigerator so that the treated water does not become contaminated with germs. Filters could also break through unnoticed and release retained substances back into the water in concentrated form.
Stiftung Warentest (Germany’s consumer testing foundation) examined six jug filters and two replacement filters in test 7/2022. The report is behind a paywall; we know only the freely readable introduction. The filters promise soft water and, according to the foundation, manage this for only a few litres; water of 16 to 17 °dH was poured in, which according to the foundation is roughly the German average. In the simulated holiday scenario, bacteria were found in all jugs, and in one model also mould.
Our assessment: for the machine, a filter is an option. With Siemens, from an editorial point of view, hard water from level 3 upwards speaks more for the filter and level 1 more against it, as our German review of the Siemens BRITA Intenza sets out. The Elizabeth from Lelit comes with a resin filter for water softening (35 l, replace after 14 tank fills or 4 months), and according to our German review a resin filter in the tank holds back limescale. For a jug filter, the hygiene rules recommended by the Verbraucherzentrale apply.
Bottled water in the tank
With bottled water, the machine manufacturers’ requirements are strict. In its FAQ, Nivona advises against filling the tank with mineral water to avoid limescale, because mineral water also contains calcium and magnesium; it recommends tap water, regular descaling and the CLARIS filter. Philips, Saeco and Gaggia require in their safety booklets never to fill the water tank with warm, hot or mineral water, as also stated in the German review of the Gaggia Classic UP. Jura and Tchibo exclude carbonated mineral water, De’Longhi sparkling and distilled water, Sage heavily filtered, demineralised and distilled water. For levels 4 or 5 on the Sage indicator strip, by contrast, the manual of the Sage Barista Touch strongly recommends an alternative water source.
With manual brewing without a machine, no manufacturer requirement binds you. There we would look at the bicarbonate, not at calcium and magnesium. How to find the value on a mineral water label is explained in the FAQ below.
Descalers: citric acid, vinegar or manufacturer products
On the products, the sources contradict each other. The Verbraucherzentrale writes that kettles and coffee machines can be descaled without problems with diluted citric acid, and the Stadtwerke Augsburg say the same for deposits in kettles and coffee machines. For the aluminium Moka Express, Bialetti prescribes: fill the boiler to just below the valve, add two teaspoons of citric acid or vinegar, assemble the pot completely, run one cycle without coffee grounds, pour away the liquid, rinse with running water. The pot is expressly not dishwasher safe. The manual of the Severin KA 4810 requires 80 ml of vinegar essence in the tank and the jug filled with water to mark 10, followed by a rinse of the appliance.
On the other side, for the EQ600, Siemens excludes pure citric acid, vinegar, vinegar-based products and phosphoric acid and permits only its own tablets. WMF excludes substances based purely on vinegar or citric acid, saying suitable products contain inhibitors to protect seals and surfaces, and that with an unsuitable product there is no warranty claim. In its FAQ, Nivona writes that you should never use citric acid, vinegar or formic acid, as this destroys the sealing rings. Philips and Saeco, Gaggia, Grundig, Senseo, Nespresso and Moccamaster have also excluded vinegar. We would use the product from the manual. If it says nothing, the diluted citric acid from the Verbraucherzentrale and the Stadtwerke Augsburg is the only independent statement we found, and when in doubt you ask the manufacturer.
Two descalers from our German reviews work with other acids. The De’Longhi EcoDecalk is a lactic acid descaler in a 500 ml bottle for five applications of 100 ml each, and De’Longhi names it as suitable for all coffee machines. The durgol swiss espresso contains amidosulfonic acid instead of citric acid, is according to the manufacturer not intended for filter machines and kettles, and requires in its instructions that the machine be set to the local water hardness. Cleaning tablets are something different: the German review of the Jura 3-phase cleaning tablets explains that complexing agents keep hardness formers in solution in the rinse water but do not remove a limescale layer that has grown over months. How a descaling programme runs is explained in the guide Descaling and cleaning. Our German ranking of cleaning and care products compares eight of them, including descalers and water filters.
What the sources leave open
We have not measured or tasted anything; everything here comes from the named sources or is marked as a conclusion, assessment or our own calculation. They leave some things open, and we have not filled the gaps. None of the five that comment on taste (Kaffeeverband, Jura, Siemens, Brita, De’Longhi) documents the statement in the material we read, and we found no independent check that a water filter improves coffee taste. Melitta, Nivona, De’Longhi and Philips give no accuracy for their indicator strips, and on the price and accuracy of strips from a hardware store we found no source. Whether a cartridge (Brita Intenza, CLARIS, AquaClean, DLSC002) lowers the hardness by sodium or by hydrogen exchange is not stated expressly by any manufacturer, and nobody quantifies the effect in °dH. Whether descaling too often strains the seals is not documented by the sources we read; Nivona and WMF only warn against descalers with unsuitable ingredients.