The setpoint comes from the control system and can be changed at will, and the better units correct their own output.

A manual regulator is set by turning a knob, which adjusts a spring loading a diaphragm. It holds that pressure until somebody changes it. Its accuracy depends on the mechanism, and it drifts with supply pressure and flow.

A proportional regulator receives an electrical setpoint. It can be changed instantly, repeatedly, and differently for each product or each stage of a cycle - without anyone touching the machine.

Closed-loop units also measure their own output and correct it, which a manual regulator cannot do. That removes the droop that occurs when flow starts, and the drift that comes from supply variation.

So the difference is not only remote adjustment: it is that the delivered pressure becomes a controlled variable rather than a mechanical setting that happens to be approximately right.

Force in a cylinder is pressure times piston area, so setting the pressure sets the force the cylinder can develop.

For a given cylinder, the effective piston area is fixed. Controlling the pressure applied therefore controls the force directly and predictably.

That allows a single clamping or pressing station to apply different forces to different products, set from the recipe rather than by adjusting a manual regulator between batches - which removes both the changeover time and the risk of the adjustment being forgotten.

Two refinements matter in practice. The rod side of a double-acting cylinder has a smaller effective area because of the rod, so the force differs between directions for the same pressure. And friction in the cylinder and the mechanism means the force delivered at the tool is somewhat less than the theoretical figure, and varies with speed.

For accurate force, calibrate the relationship on the actual machine rather than relying on calculation alone.

Enough to fill the downstream volume at the required rate - and this is the specification most often got wrong.

A proportional regulator's accuracy figures describe how well it holds a setpoint. They say nothing about how much air it can pass.

If the downstream actuator draws more flow than the regulator can supply, the pressure sags the moment movement starts, recovers when it stops, and the control system sees an output that will not follow its setpoint under load. The valve is working correctly and is simply too small.

So size on the flow required at the working pressure - taking the peak demand when the actuator is moving, not the static condition - and check the manufacturer's flow figures at that pressure rather than the headline value.

Where a large flow must be controlled precisely, a small proportional regulator piloting a larger volume booster is the usual arrangement, giving accurate control with the capacity to deliver.

Quickly at the valve, with the downstream volume usually determining what the process sees.

The regulator itself can change its output in a short time, and manufacturers publish a step response figure. But that describes the pressure at the regulator outlet.

What the actuator experiences depends on how much volume has to be filled or vented to reach the new pressure. A small cylinder close to the regulator responds almost immediately; a large volume at the end of a long tube takes considerably longer, and reducing pressure can be slower still if the regulator vents through a small exhaust.

So for fast pressure control, mount the regulator close to the actuator and keep the connecting volume small - the same principle that governs valve placement.

Check whether the regulator can actively exhaust to reduce pressure or relies on the downstream volume leaking down. The difference matters greatly on a falling setpoint, and a non-venting regulator cannot reduce pressure quickly at all.

Good on closed-loop units, and specified as several separate figures that should be read together.

Manufacturers quote linearity - how closely output follows setpoint across the range; hysteresis - the difference between rising and falling; repeatability - how consistently the same command produces the same output; and sensitivity or resolution - the smallest change it can make.

A single accuracy percentage does not capture it.

Closed-loop valves perform substantially better on all of them, because the internal sensor corrects for the mechanism's imperfections and for external disturbances.

The figures are usually quoted as a percentage of full scale, which means absolute error is larger at the bottom of the range - so a regulator with a wide span used only at low pressures may be less accurate in absolute terms than a smaller-span unit. Choosing a pressure range close to the working requirement improves accuracy for that reason.

Force control, tension control, test benches, and any process where pressure is a recipe parameter.

Clamping and pressing applications use it to set force per product, and to vary force through a cycle - light to locate, firm to hold.

Web and wire tension control uses it to drive a brake or a dancer, adjusting continuously as reel diameter changes through a run.

Leak and burst testing uses it to apply and hold precise test pressures and to run programmed ramps.

Other uses include controlling paint and coating pressure, balancing and weight compensation systems, controlling the pressure to a pneumatic actuator on a process valve, and glue and dispensing applications where dispensed volume depends on pressure.

What unites them is that the pressure is a process variable rather than a supply setting, and that changing it under control is worth more than the additional cost of the component.

Clean air above all, plus periodic verification against a reference gauge.

The internal pilot passages and the sensing element are small and precise, and contamination is the principal cause of degraded performance. Symptoms are drift, increased hysteresis and sluggish response - all of which look like control problems.

So local filtration to the manufacturer's specification, changed on schedule, is the main maintenance task. Confirm whether lubricated air is acceptable; for many proportional units it is not.

Calibration should be verified periodically by comparing the delivered pressure against an accurate reference gauge at several points across the range, not just one. Drift at the top of the range with a correct zero is a common pattern.

On units with internal electronics, check that the signal scaling in the controller still matches the valve's configuration after any parameter change - a mismatch produces a machine that is consistently slightly wrong.