Short answer: an M-Bus to wireless water meter adapter works, but it is a bus master, not a battery-powered dongle. It has to power the M-Bus loop, so it needs mains power where the concentrator sits. If there is no mains power, the practical choices are a meter-side wireless module or an add-on sensing layer.
We build meter-side sensing and communication modules, not bus masters. So this guide separates the three retrofit routes, shows the arithmetic that decides between them, and says plainly where our modules fit. Every figure below comes from our own datasheets, not from a generic industry chart.
If the wired interface is still undecided, compare the two options side by side: Wired M-Bus vs RS485 for smart water and gas meters covers wiring, bus power, addressing and data model.
| A. Bus adapter (gateway) | B. Meter-side module | C. Add-on sensing layer | |
|---|---|---|---|
| Best for | Building or campus already wired for M-Bus, no backhaul | Meter already has a pulse, RS485 or M-Bus slave port | Mechanical meter with no electronic interface |
| Power | Mains or DC supply | 3.6 V lithium cell | 3.6 V lithium cell |
| Work on the meter | None | Wiring, possibly a case opening | PCB and metal target inside the meter |
| Cost per point | Low — one gateway serves a group of meters | Medium | Medium to high |
| Main risk | Bus load, master conflict | Sleep current, RF environment | Assembly tolerance, seals |
Selection rule: mains power and one large group of meters, route A; scattered meters that already have an interface, route B; old mechanical meters with nothing, route C.
Wired M-Bus is a current-loop bus. Under EN 13757-2, one slave counts as one unit load, 1 UL = 1.5 mA, and a segment carries up to 250 UL. The master holds the loop at about 36 V and drops it to about 24 V when a slave draws current.
The current adds up quickly. 100 meters draw 150 mA. 250 meters draw 375 mA. Size the master supply for that current plus headroom; 500 mA is the usual starting point. A device that has to source 375 mA cannot run for ten years on a lithium cell.
| Item | Value |
|---|---|
| Load per slave | 1 UL = 1.5 mA |
| Limit per segment | 250 UL — add a repeater to split the bus |
| Typical speed and distance | 2400 bps, about 1 km depending on cable and load |
| Current for 100 meters | 150 mA |
| Current for 250 meters | 375 mA |
Two field problems cause more failures than the circuit itself.
Master conflict. If the old concentrator is still polling, the new adapter becomes a second master on the same segment. The symptoms are dropped frames, jumping meter IDs and readings that work intermittently. Either take the old master off the bus completely or let the adapter take over the whole topology. Do not leave both running.
Polling time. Reading 250 meters one by one at 1–2 seconds each takes 4–8 minutes per round. If the utility expects a report every 15 minutes, there is very little room left for retries and exception handling. The value of the adapter is not “it converts to wireless”; it is that it can finish the whole segment inside the reporting window.
To be clear about scope: we make meter-side metering and communication modules, not bus masters. For a whole-building retrofit, the gateway comes from a concentrator vendor. Our modules come in when each meter already has a pulse output, an RS485 port or an M-Bus slave port.
A meter-side module does not power the bus. It sits on the meter’s interface, which is why a battery design works. But “ten-year battery life” has to be calculated from sleep current and reporting frequency, not repeated from a brochure.
Our NB-IoT water meter module datasheet gives <18 µA static current, <300 µAh per typical report, one report per day by default, and a nominal battery life of 6 years. With a 3.6 V / 2.4 Ah ER14505 and 20% capacity held back:
| Item | Calculation | Per day |
|---|---|---|
| Standby | 18 µA × 24 h | 0.432 mAh |
| One report per day | 0.300 mAh × 1 | 0.300 mAh |
| Total | 0.732 mAh | |
| Usable capacity | 2,400 mAh × 80% | 1,920 mAh |
| Theoretical life | 1,920 ÷ 0.732 | 2,623 days ≈ 7.2 years |
The datasheet says 6 years while the arithmetic gives 7.2. That is the right direction: self-discharge, low temperature and retries in weak signal all need margin.
Customers often ask for a report every 6 hours instead. The same cell then looks like this: 0.432 + (0.3 × 4) = 1.632 mAh/day, and 1,920 ÷ 1.632 ≈ 1,176 days, or about 3.2 years. Four times the report rate takes more than half the life out of the same battery. Say this before the quote, not two years after installation when batteries start failing in batches.
The other two modules use the same arithmetic, with one trap each.
The LoRaWAN module has <10 µA sleep current, 130 mA transmit at 20 dBm, 12 mA receive, -136 dBm sensitivity, and supports CN470, EU433, EU868, US915, AS923 and AU915. Energy per report depends on the spreading factor. SF12 has a long airtime and SF7 a short one; the difference can be an order of magnitude. When someone quotes you “µAh per frame”, ask which SF and which payload length it was measured at.
The wM-Bus module datasheet says <20 µA, but the bracket contains the important part: 30 s transmit interval. That 20 µA is a whole-device average that already includes frequent transmission. It is not the same measurement as the NB-IoT figure of 18 µA at one report per day. Putting the two numbers side by side without the intervals produces the wrong conclusion every time.
So when you ask about module power, do not ask only “what is the sleep current”. Ask for the reporting interval, the airtime and the retry count as well.
On this route the hard part is not the circuit. It is rotation speed and mechanics.
Our non-magnetic metering modules come in three speed tiers: 2.2, 10 and 64 revolutions per second. Those numbers were not chosen by feel. They come from the overload flow of the largest meter in each class. Three examples:
Residential G15/G20/G25, one pulse per litre. DN25 has an overload flow of 7.87 m³/h, which is 2.18 L/s, or 2.18 revolutions per second. That is why the first tier is 2.2 rev/s.
Commercial DN50–DN150, one pulse per 10 litres. DN150 has an overload flow of 312.5 m³/h, or 87 L/s. Divided by 10 litres, that is 8.7 rev/s — the reason the large-diameter tier is 10 rev/s.
Direct impeller sampling with a metal target on the impeller. The limit is 28 revolutions per litre. At DN25 overload that is 61 rev/s, which is where the 64 rev/s version comes from.
The high-resolution version follows different logic: 10 pulses per revolution at 0.1 L per pulse, so one litre per revolution. At 2.2 rev/s that is 7.9 m³/h, enough for DN15 and DN20.
Recheck this for every project: calculate revolutions per second from Q4 overload flow first, then match the module’s speed tier. If you select on nominal flow only, the meter will lose pulses during the overload test. The same principle is covered in our guide to 0.1 L high-resolution mechanical water meters.
Two mechanical numbers matter as well. Stable sensing distance is under 8 mm, which is the datasheet figure; 12 mm is the maximum sampling distance after adjusting the coil diameter, so do not mix the two. For magnetic immunity, work to 30,000 Gauss and above — the meter keeps counting with a magnet held against it. The PCB side of this design is covered in the WF8002 inductive sensing design guide.
The WF8002 metering chip comes in three versions, and static current rises with speed. Do not order on “≤4 µA” alone; that number belongs to the residential version only.
| Version | Maximum speed | Static current |
|---|---|---|
| Residential | 2.2 rev/s | ≤4 / ≤5 µA |
| Commercial | 10 rev/s | ≤11 µA |
| High-speed | 64 rev/s | ≤42 µA |
Matching the residential 4 µA figure to a DN150 running at 8.7 rev/s is the most common mismatch we see. Speed first, current second.
One more field detail: the module outputs a signal-strength value and separate forward and reverse pulse counts. Production uses the signal strength to judge assembly clearance; service uses the two counters to check whether the module and the main board agree. When electronics and mechanics do not match, these two numbers tell you whether the problem is assembly or the board, which is cheaper than reworking the whole meter. If the host MCU interface is the open question, start with the WF8002 pulse output guide.
The selection logic is plain. If the utility already has wM-Bus concentrators and works to OMS/EN 13757, use wM-Bus. If there is a LoRaWAN gateway and the project wants its own network, use LoRaWAN. If there is no gateway, the project will pay for a SIM, and you need IP68 and remote upgrade, use NB-IoT. The EN 13757 architecture itself is covered in this wM-Bus integration guide.
Here is a wM-Bus payload example worth aligning before integration: forward flow uses VIF 0x13 in 0.001 m³, battery voltage uses VIFE 0x47 in 0.01 V, and the error flags carry valve state, low battery, meter removal and magnetic insertion. Agreeing these fields with the platform team up front is faster than capturing frames afterwards.
| # | Check | How to judge it |
|---|---|---|
| 1 | Existing bus topology | How many segments, how many slaves, does any segment exceed 250 UL? |
| 2 | Ownership of the old master | Who polls after the retrofit? The old master must go offline. |
| 3 | Power at the gateway | Is mains power available, and can it deliver 500 mA peak? |
| 4 | Meter interface | Pulse, RS485, M-Bus slave port, or nothing? |
| 5 | Pulse weight | Litres per pulse, and revolutions per second at Q4. |
| 6 | RF environment | Pit lids, basement walls, metal enclosures. |
| 7 | Reporting interval | What the business asks for versus what the battery can support. |
| 8 | Seals and verification | Does opening the case or removing the meter affect local verification? |
One note on item 8. Leaving the metering chamber untouched usually does not trigger a new type approval. Anything that involves drilling the case, breaking a seal or changing the totaliser interface should go to the certification body first. Do not decide that on the sales call.
No, not while it is the master. The master supplies the loop; it cannot also draw its operating power from it. If there is no mains power at the concentrator, go back to route B or C.
The standard limit is 250 UL per segment. In practice you derate for cable gauge, voltage drop and polling period. 100–150 meters is a comfortable engineering range.
Three numbers decide it: sleep current, energy per report and reports per day. Our NB-IoT module at 18 µA, 300 µAh and one report per day is rated at 6 years. At one report every four hours it is a little over three.
Use route C and add a sensing layer. Measure the pointer diameter, confirm the rotation speed and check the installation space before choosing the module. Our minimum pointer diameter is 18 mm, with a stable sensing gap under 8 mm.
If the metering chamber is untouched, it usually does. If seals or the case structure are involved, follow the local metrology authority’s decision.
Send us the meter type and size, the existing bus topology and slave count, the meter interface, the required reporting interval, whether mains power is available, and a few site photos. With those we can name the route and the module without guessing.
Contact us at hangzhouweifeng.cn/pages/contact-us, or email Jing@hangzhouweifeng.cn. Tell us the retrofit case in one paragraph; we will reply with the route, the module and the numbers behind them.
