A traffic checkpoint looks simple from the driver's seat: pull over, blow into a device, move on. Behind that interaction is a two-stage testing structure designed to move a large number of people through quickly while producing results that survive challenge for the small number who fail. Understanding how it works is useful whether you are equipping a police unit or briefing a fleet on what its drivers will encounter.
Why checkpoints use two stages
The two demands on a checkpoint pull in opposite directions. Screening has to be fast, because a queue of stationary vehicles on a live road is its own hazard. Evidence has to be rigorous, because a result that leads to prosecution will be examined in detail.
No single device does both well. So enforcement separates them: a rapid preliminary screen decides who needs closer attention, and a slower evidential test produces the record for those who fail it.
Stage one: preliminary screening
The first test is designed for throughput. Passive and non-contact devices can sample the air near a driver's face as they speak, without a mouthpiece and without the driver leaving the vehicle, returning a pass or alert indication in seconds. Screening batons serve the same purpose in a handheld form and are easy to operate from outside a car window.
Because there is no mouthpiece to change between subjects, per-test cost is effectively zero and there is no consumable to run out of mid-shift — a practical consideration on a long deployment. The trade-off is that these devices, typically semiconductor or non-contact sensors, indicate the likely presence of alcohol rather than delivering a defensible figure. That is exactly what stage one needs to do.
Stage two: the evidential test
A driver who fails the screen is tested again on a calibrated fuel cell analyzer. This is where the number that matters is produced.
Fuel cell sensors respond specifically to ethanol rather than to volatile compounds in general, so the result is not vulnerable to the cross-sensitivity arguments that a semiconductor reading invites. The device requires a proper deep-lung sample rather than a token puff, and should reject an inadequate blow instead of reporting a low figure from the wrong part of the breath.
Before this test, a waiting period applies — typically 15 to 20 minutes during which the subject consumes nothing. This is not procedural formality. Alcohol still present in the mouth from a recent drink, from reflux, or from an alcohol-containing product sits on top of the true deep-lung reading and inflates it substantially. The observation period is what separates a measurement of the subject's blood alcohol from a measurement of what is on their tongue.
The record is the evidence
An evidential device earns its price at the moment the result is captured, not the moment it is displayed. A printed slip or a logged digital record carrying the reading, the date and time, and the device identity is a durable artefact. A number an officer read off a screen and wrote in a notebook is testimony about a reading, which is a materially weaker thing.
Devices with integrated or Bluetooth printers produce that artefact at the roadside, and on-device logging retains a copy that can be exported later. For a unit running regular deployments, that log doubles as an operational record: how many tests were conducted, when, on which device, by whom.
Calibration discipline
The first line of attack on any breath result is the instrument's calibration history, and it is the easiest one to defeat with good record-keeping.
Every evidential device should carry a documented history: the dates of each accuracy check and calibration, the certified standard used and its lot or certificate details, the readings taken before and after any adjustment, and the person who performed the work. Devices should rotate through calibration on a schedule with spares available, so that taking a unit out of service never leaves a deployment short.
A lapse is expensive in a way that is easy to underestimate. If a device is found to have been out of calibration, every result it produced since its last valid calibration becomes arguable — not just the one being contested.
Operator practice
The instrument is only part of the result. Officers need to be consistent about the observation period, about ensuring a full sample is delivered, about changing mouthpieces between subjects on devices that use them, and about recording the result the same way every time. Inconsistent practice across a shift produces results of inconsistent quality, and defence arguments find that variation quickly.
Ambient conditions matter too. Devices have specified operating temperature ranges, and a unit left on a dashboard in direct sun or used in extreme cold may fall outside them. Warm-up and recovery times between tests are real constraints at a busy checkpoint — plan device numbers around them rather than rushing samples.
Equipping a unit
A working checkpoint kit is a tiered one. Non-contact detectors and screening batons handle stage one at volume: AlcoBreath's AT-7000, AX5, A-18 Rapid and the A19 Chita 1 and Chita 2 batons are built for this role. Stage two calls for a fuel cell analyzer with printing and logging — the A-9000, A-100, A200 and A-65 sit in that class, and the A15 Finger Print adds biometric identification where tying a result to a specific person matters.
Match the ratio to your deployment: several screening devices per evidential unit reflects how the traffic actually flows, since most drivers clear the first stage and never need the second.