meterradio

902–928 MHz · ISM band · pre-flight check

Will an RTL-SDR actually read your meter?

Most guides start after the part that decides everything. Check before you spend $40 on a dongle that may have nothing to listen to.

Look up your FCC ID

The definitive answer. Every radio sold in the US carries an FCC ID on its faceplate, and the filed grant states the frequencies it is authorised to use. 991 meter radios are in here, straight from the FCC filings.

Try:

Can’t read the plate? Describe the meter

Less certain than the FCC ID, but it settles most cases — and it will tell you when there is nothing to look up.

1 — Which meter?

What the filings actually say

Every US-certified meter radio, counted from the FCC equipment-authorisation database.

991meter radios on file
679receivable at 902–928
312transmit somewhere else

That last number is the one that costs people weekends. Roughly one meter radio in three is not in the band every tutorial tells you to sweep — it is on licensed UHF around 450–470 MHz, in the old 950–960 utility band, on 2.4 GHz, or on an inductive pad that never transmits at all. Sweep 902–928, hear nothing, and the obvious conclusion is wrong.

Before you spend the $40

Four things that cost people a weekend each. None appear in the standard tutorial.

A spinning metal disk means no radio. Ever.

An induction-disk meter is purely mechanical — the disk is driven directly by the current it measures. There is no transmitter, no firmware, and no frequency to find. The utility reads it by sending a human to look at it. This one check disqualifies more setups than everything else combined. More on electric meters.

The antenna matters more than the dongle.

The stock whip in a cheap SDR kit is the usual culprit behind “I hear nothing.” A dead or wrong-length element looks exactly like a meter that does not transmit. Swap the radiator before concluding anything — in one deployment this cost a full day of chasing software bugs that were never there. What to buy instead.

You will hear your neighbours loud and clear.

Picking up half a dozen strong meter IDs that are not yours is normal, not a win. It proves the receiver works, which is genuinely useful — but the interesting one may still be silent. How to tell yours apart.

Not every utility at an address transmits.

Gas, water and electric are three separate decisions made by three separate utilities, often decades apart. A house can easily have one chatty meter and two silent ones. Qualify each independently.

Protocols you might decode

What the open tools understand, and what each message actually carries.

MessageTypicallyWhat it carries
SCMgas, water, electricTotal consumption — the basic meter reading
SCM+gas, water, electricTotal consumption, longer meter IDs, better precision
IDMelectricDifferential consumption across 47 prior intervals at 5 minutes each — the richest payload of any type
R900water (Neptune)Total consumption plus a leak flag
R900BCDwaterConsumption as binary-coded digits

Most ERT-family devices transmit in “bubble-up” mode: no wake-up tone is needed, they simply broadcast every few seconds and you listen. Full protocol reference.

Field log — one address, three utilities

A real survey across the full 902–928 band. Note that the majority result is silence.

UtilityMeter faceResultFrequencyProtocol
GasModule on dialsDecoded912.6 MHzERT / SCM
WaterMechanical registerSilentnone
ElectricInduction diskSilentnone
NeighbourDecoded~915 MHzERT / SCM
NeighbourDecoded~915 MHzERT / SCM

One of three own-meters transmits. Two neighbours came in stronger than the target. A typical outcome, not a failed install.

Why this page exists

Every RTL-SDR meter tutorial assumes you already have a transmitting meter. The question people ask first — “will this work at my house?” — is scattered across a GitHub wiki, a meters.md file, FCC filings and forum threads. Nobody has put a straight answer in one place.

The FCC publishes no API for this. Its search sits behind bot protection, and the frequency data exists only inside individual grant records. Assembling it into something you can actually query took a browser-driven harvester and a lot of patience with a nested HTML table from 1998. How the dataset was built.