Coffee scales by routine: which features earn counter space

Resolution: a 0.1 g step measured against an 18 g dose

Resolution is the smallest weight the display will show, and the practical effect is the rounding it forces. A scale that displays to 0.1 g rounds every reading by up to half a step, so ±0.05 g. On the 18 g dose that Breville lists for the Bambino and the Bambino Plus, that rounding is ±0.28% of the dose; on the 22 g dose listed for the Dual Boiler, ±0.23%; on a 15 g pour-over dose at 1:15, ±0.33% [1][2][3][6].

A 0.01 g display moves the same rounding to ±0.005 g, or ±0.03% of an 18 g dose. Both are far below the 1 g step that matters at the other end of the shot: on the published 18 g in, 36 g out recipe, one extra gram of yield is 2.8% of the beverage [3]. Resolution therefore separates scales by an order of magnitude that is smaller than the differences a barista acts on, which makes 0.1 g sufficient for espresso and pour-over and 0.01 g useful mainly where doses are tiny or where a single bean's weight is the point.

Capacity: the weight of a 1.25 L brew

Capacity decides what the platform can hold without being re-tared mid-recipe. Take the batch case first. A 1.25 L cycle brewed at 60 g per litre — the automated-filter strength in the published filter comparison — needs 75 g of coffee, so the water and grounds together come to about 1,325 g before the carafe is counted [4][6]. A scale rated to 2 kg covers that with margin; a 500 g scale does not.

Pour-over for one is smaller than it looks. At 1:15 with 300 g of water, the load is 20 g of coffee plus 300 g of water, and the server underneath adds a few hundred grams; a 1 kg platform covers the whole assembly [6]. The espresso case is smaller again: 18 g of coffee and a 36 g beverage sit comfortably inside 500 g even with the cup on the platform [1][2]. Cupping, at 0.055 g/mL — 8.25 g per 150 mL — is the smallest recipe in the set and the least demanding on range [8].

Timer and response against a 25-second window

An espresso protocol extracts in 25 s ± 3 s at about 9 bar, which is the number a scale's timer has to be able to follow [7]. Three seconds is 12% of that window, so a timer that counts in single seconds and starts from the first drop is doing real work; a timer that lags by a second or two eats a meaningful share of the allowance before the shot even reaches the target weight.

The pour-over case is the opposite. Pacing a 1:15 brew by hand takes minutes rather than seconds, and the useful relationship on the display is cumulative weight against elapsed time, because that is what tells you whether the bed is draining faster or slower than the last brew [6]. Response behaviour — how quickly the number on the screen follows the stream — matters most in the espresso window, where the yield lands inside a few seconds, and least in a batch brew that is finished unattended.

Water resistance and power at the edge of a 9 bar counter

Espresso work happens beside a machine that pushes water at roughly 9 bar and, on the Bambino Plus, vents steam at 266 °F (130 °C) [2]. Splashes and steam sit on the same bench as the scale, so a sealed top plate is the specification that decides whether a reading survives the session; the same logic applies to a pour-over kettle that tips.

Power is the unglamorous one. A scale that is charged by USB-C and a scale that takes AAA cells both work, and the routine difference is whether the unit has to leave the counter to be fed. What matters operationally is that the scale does not die mid-shot: a 25 s extraction with a display that blacks out at 20 s costs the whole reading and the recipe built on it [7].

Update rate: the grams per second the display has to keep up with

Resolution is a static number, but weighing is a moving one, and the flow rate behind a brew is what the display has to track. A 1:15 pour-over at the published 20 g dose puts about 300 g of water through the bed, and that water leaves the kettle across three to four minutes — roughly 1.3 to 1.7 g per second. An espresso recipe of 18 g in and 36 g out runs the output side in 25 seconds, which is about 1.4 g per second at the cup. Both brews therefore move the reading in tenths of a gram several times a second, so the number on the screen is only usable if the platform updates far faster than the eye reads it. That is the reason a scale's response or update specification sits next to its resolution rather than somewhere further down the product page: a display that lags by a full second costs you roughly 4 per cent of a 25-second shot window, and on a pour-over it hides the moment the bed starts to stall.

The specification table, read against the routine

The table is a filter rather than a ranking. Cross off any scale whose capacity cannot hold the routine in its row; then check that its resolution rounds your typical dose by well under a gram; then confirm the timer resolves the window you actually brew in. Anything left over — housing material, app support, colour — is a preference, not a specification.

Which two lines to read first

Read resolution and capacity before anything else, because those two decide whether the scale can weigh the recipe at all. An 18 g dose on a 0.1 g display rounds by ±0.05 g, which is well inside the tolerance a 36 g yield is written with; a 1.25 L batch needs a 2 kg platform because the load passes 1,300 g [1][3][4][6]. Timer and response come next, and they are the two specifications that separate a scale built for a 25 s window from one built for a five-minute pour [6][7]. Water resistance and power come last, and they are the two that decide whether the unit is still working in month twelve.

For context on what the rest of the bench costs, pour-over hardware — drippers, kettles and servers — is bought as separate listings at one retailer, which puts the scale in perspective as the item that changes every measurement you make [5].