How a pneumatic espresso machine works

The pressure problem

Espresso is defined by pressure, and pressure is the one thing a portable machine cannot borrow from a wall socket. The industry answer for a century has been a pump: 9 bar, delivered by a motor, a boiler and about fifteen kilograms of steel.

A pneumatic machine takes a different route. It stores energy as compressed gas and releases it through a regulator. One 8 g cartridge of liquid CO₂ becomes roughly 4.1 litres of gas at atmospheric pressure — and because the gas is stored at 57 bar at 20 °C, it carries that pressure with it wherever it goes. No motor, no battery, no arm.

The interesting engineering is not the gas. It is everything between the cartridge and the puck.

Stage 1 — regulation: from 57 bar to 9

A full CO₂ cartridge sits at its saturated vapour pressure: about 57 bar at 20 °C, rising to roughly 75 bar at 30 °C. Espresso wants 9 bar. That is a six-to-eight-fold reduction, held stable while gas is flowing and the cartridge is chilling from expansion.

This is why a pneumatic machine lives or dies by its regulator. Three things matter:

1. Setpoint accuracy. A regulator that wanders 2 bar across a shot is a machine that cannot reproduce a recipe.
2. Flow behaviour. Regulators droop under load. As gas flows, the outlet pressure sags unless the design compensates. Droop is what you taste as a shot that starts fast and finishes slow.
3. Relief and venting. Every pressurised path needs a relief valve and a controlled way to dump residual pressure before the chamber is opened.

A regulator is not a valve. A valve lets gas through; a regulator holds a pressure while it does so. If a machine's marketing never mentions a regulator, ask what holds 9 bar.

Stage 2 — thermal mass replaces the boiler

A café machine has a boiler with a PID and a couple of litres of water acting as a thermal flywheel. A portable pneumatic machine has a few hundred millilitres — or less — and no active control loop.

So where does the heat go? Into the puck.

Rough calculation: 18 g of ground coffee at 20 °C, raised to roughly 94 °C during extraction, with a specific heat capacity near 1.8 kJ/(kg·K), absorbs about 2.4 kJ. If 100 g of brew water at 94 °C supplies that energy, its own temperature falls by roughly 5–6 °C — before any losses to the basket or the chamber walls.

That single number explains most bad portable espresso:

• A cold, dry basket steals heat from the first, most soluble part of the shot.
• A cold cup removes another degree or two before you taste anything.
• Wind and altitude increase losses at exactly the moment the machine is least able to compensate.

The fix is procedural, not electronic: preheat the chamber, the basket and the cup with the same water you are about to brew with, and work fast.

Stage 3 — pressure is not flow

Nine bar is a setpoint, not a guarantee of good extraction. What the puck actually sees depends on resistance:

A pneumatic machine makes the pressure column constant. It does nothing about the other four. If the shot runs in 12 seconds, the answer is grind, not pressure — and if it chokes completely, the answer is also grind.

Stage 4 — the gas budget

Buyers ask "how many shots per cartridge" and sellers answer with a number that never holds in the field. The arithmetic is easy to do properly.

• 8 g of CO₂ is 0.18 mol. At standard conditions that expands to about 4.1 L.
• A 36 ml shot brewed at 9 bar consumes the equivalent of 0.32 L of gas at atmospheric pressure.
• Ceiling: about 11 shots. Add purge cycles, chamber headspace and regulator losses, and realistic yield lands at 5–7.

Two practical consequences:

1. Cartridge cost per shot is a real operating cost, and it scales with shot size and purge discipline — not with the machine's price.
2. Cold weather shrinks usable output. Vapour pressure falls with temperature, so a cartridge left in a freezing car delivers less pressure headroom and more droop.

What the architecture cannot do

Honesty is cheaper than a returned machine.

• No steam. Milk needs a separate tool.
• No throughput. One chamber, one puck, one shot at a time, plus a cooldown between shots.
• No infinite repeatability. Without a boiler and a PID, temperature varies with ambient conditions.
• A consumable. Gas is a supply chain you now participate in.

If any of those are dealbreakers, a battery-powered portable pump or a small home machine with a boiler will serve you better — and that is a legitimate outcome, not a failure of the category.

What to check on any pneumatic machine

Whether you buy from Gilka or anyone else, these are the questions that separate an engineered product from a rebadged experiment.

• Is there a regulator, and is its setpoint published? A machine that only quotes "up to 15 bar" is quoting a cartridge, not a brewing pressure.
• Is there a relief valve and a documented depressurisation step?
• Is the gas path rated for food-grade CO₂ and does the manual say so?
• What is the water volume per shot, and is it fixed or user-set? Fixed volume trades flexibility for repeatability.
• Which parts are consumables — gaskets, seals, baskets — and what do they cost per year?
• What is the mass and packed volume, not the marketing "pocket size" claim.

Published specifications for the Gilka One are being finalised. We will publish the sheet before any pre-order opens, and we will label every figure as measured or targeted.