Brazilian meter makers are not short of sensor options. They are short of time. Most projects we see arrive with the meter mechanics fixed and the communication module already chosen, and the sensing layer gets picked last, usually by comparing quiescent current on two datasheets.
That is the wrong order. Three local constraints decide more than any spec table: certification, frequency band, and water loss.
ANATEL homologation. Any radio product sold in Brazil needs ANATEL homologation. For a finished meter it is normally held by the meter maker or the importer, not by the sensor supplier. Ask early who holds it and whether your module choice is covered. A module that is cheap and low-power but has no path to homologation is not cheap.
Frequency band. LoRaWAN in Brazil runs on AU915. NB-IoT is deployed mainly on B3; confirm coverage with the operator before locking the design. Wireless M-Bus, which dominates European tenders, is not the default here.
Water loss. Non-revenue water in Brazil has stayed above 30% for years. Utilities therefore care about minimum-flow registration and tamper evidence. Both are sensing-layer problems, not radio problems.
Pulse weight and rotation speed. Residential G15-G25 meters usually run 1 L per pulse. At DN25 overload (7.87 m3/h) that is 2.18 revolutions per second. If your sensor tops out at 2.2 rps, you have no margin. Check the number with Q4, not nominal flow.
Resolution. A 0.1 L version outputs ten pulses per revolution with 1 L per turn. At 2.2 rps that covers 7.9 m3/h, enough for DN15/DN20, not for a DN50.
Sensing gap. Our module datasheet specifies a stable measuring distance below 8 mm. The 12 mm figure quoted in some presentations is the maximum sampling distance after enlarging the coil, not the production setting.
Magnetic immunity. Formal testing shows normal metering with a 10,000 Gauss magnet brought close to the module.
Quiescent current at your reporting interval. A figure without an interval is meaningless. Our wM-Bus module is below 20 uA with a 30-second transmission interval. The NB-IoT module is below 18 uA at one report per day. Two numbers, two duty cycles.
Removal alarm. The module can flag separation from the meter body. In retrofit projects this is often the only tamper signal the utility gets.
Our LoRaWAN module: 10 uA quiescent, 130 mA transmit, 12 mA receive, -136 dBm sensitivity, bands including AU915. Our NB-IoT module: 18 uA quiescent, below 300 uAh per upload, B3/B5/B8, IP68, 6-year life at one report per day.
The NB-IoT arithmetic, on a 3.6 V / 2.4 Ah ER14505 with 20% reserve: standby 18 uA x 24 h = 0.432 mAh/day; one upload 0.300 mAh/day; total 0.732 mAh/day. 1,920 / 0.732 is about 2,620 days, roughly 7.2 years. The datasheet says six years, which is the honest number once self-discharge and weak-signal retries are included.
Change the requirement to one report every six hours: 0.432 + 1.2 = 1.632 mAh/day, about 3.2 years. Same battery, four times the traffic, less than half the life. Decide the reporting interval with the utility before you quote a battery life.
For LoRaWAN the per-report cost depends on spreading factor and payload. A single uAh figure without the SF is a guess.
Reed switches are cheap and well understood, but a magnet silences them. In a market where a meaningful share of water is unaccounted for, that is not a theoretical weakness.
The inductive module uses PCB coils to track a metal target on the impeller. There is no magnetic circuit to saturate, so a strong external magnet does not stop the count. It also outputs forward and reverse pulses separately, which is what a utility needs to flag backflow and reconcile suspicious consumption.
The trade-off is mechanical: the gap matters, the target must be concentric, and assembly tolerance shows up in signal strength. We expose signal strength as a production metric for exactly this reason.
Most Brazilian volume is still mechanical. Retrofitting is usually cheaper than replacing, and the sizing rule is always the same: calculate revolutions per second from Q4, then pick the speed tier.
If a project asks for 0.1 L resolution, confirm the meter Q4 before promising it. Resolution and speed are a trade-off, not a menu.
Send the meter type and diameter, the Q4 figure, the pulse weight you want, the reporting interval, and whether the site has mains power. We will tell you which speed tier and module fits, and what battery life that actually gives, before a quote.
