Bar scale (0.1 g)

Weighs the coffee going in and the drink coming out, so a recipe is a measurement the whole team can repeat rather than an opinion each barista holds privately.

Under the platform sits a load cell: a machined metal element with strain gauges bonded to it. Weight deflects the element by an amount far too small to see; the gauges stretch with it and their electrical resistance changes in proportion; a bridge circuit turns that change into a voltage and the electronics turn the voltage into a mass. Everything that separates a good bar scale from a bad one is downstream of that conversion — how quickly the reading settles, how far it drifts as the cell warms under a hot portafilter, and how well the firmware rejects vibration from the grinder and the knock box. The distinction operators miss is between READABILITY and ACCURACY: readability is the size of the last digit shown, accuracy is how close the number is to the truth. A scale can display to 0.1 g and be wrong by more than that all day.

• Decide first WHAT you are weighing, because that decides how many scales you need and where. A bar that weighs the dose only needs one scale at the grinder. A bar that weighs dose and yield needs a second one under the group, and the two are different shapes — a bench scale will not fit the clearance and a drip-tray scale is a poor bench. Buying one scale and moving it between the two positions is the compromise that quietly ends with nobody weighing anything.
• Readability is set by your tolerance, not by ambition. An instrument whose smallest division is the same size as the tolerance you are policing cannot police it — you need to see the error before it reaches the limit. Look at the dose tolerance your own SOP states, and require a scale whose last digit is comfortably finer than it.
• Settling time is the specification that decides whether the team actually uses the scale. A scale that takes a beat too long to show a stable number gets abandoned within a fortnight of opening, and an abandoned scale is worse than no scale because it silently converts your recipe into a guess while the equipment list still says you measure. Test this before buying, on the bar, at pace — not from the datasheet.
• Match sealing to position. The scale under the group lives in a puddle; the scale at the grinder lives in dust. Ask what the enclosure is rated to and assume the bar is harsher, because a scale is a consumable in this position and the honest question is how often you replace it, not whether.
• Team skill changes the answer more than volume does. A small, trained team can run a bench scale and a stopwatch and be consistent. A team that turns over needs the measurement built into the workflow so it cannot be skipped — that argues for a permanent scale under the group and, if the budget reaches, gravimetric dosing at the grinder.
• Buy the scale before the refractometer, without exception. Extraction yield is computed from a concentration and TWO masses; a refractometer with no scale beside it produces a number that cannot be turned into anything actionable. Operators reliably buy these in the wrong order because one of them looks like a laboratory and the other looks like a kitchen tool.
• Plan for a spare before you need it. This is the cheapest item on the bar whose failure stops the recipe, and lead time on a replacement is the real cost. One spare in the drawer is a smaller number than a week of unweighed shots.