Direct Answer

The meters on site use the DL/T 645 protocol — what is different when a gateway connects to them compared with Modbus devices? DL/T 645 is a dedicated communication protocol for multifunction energy meters; it addresses by data identifier, unlike Modbus which addresses by register; when connecting, you must read energy/demand/events by data identifier, and the gateway converts them into a unified model before uploading to the cloud. The points for the ESX intelligent edge gateway to connect to DL/T 645 meters: first understand the addressing difference, then recognize the data identifiers, configure the address range and timeouts — the gateway converts DL/T 645 into a unified data model for the cloud, and the upper platform does not need to care about the meter protocol.

A scene: a campus connected 48 multifunction meters and configured DL/T 645 with Modbus experience — no energy data read, frequent timeouts — DL/T 645 is not "another Modbus" — reading by register addresses and setting Modbus-style timeouts will both hit pitfalls — read the four differences before connecting, and the connection works on the first pass.

1. What Is DL/T 645: The Dedicated Meter Protocol

The definition of the standard — DL/T 645, the Communication Protocol for Multi-Function Energy Meters, is a Chinese power-industry standard: designed for multifunction energy meters — it defines how meters communicate with collection terminals/gateways; Its relationship with Modbus — DL/T 645 and Modbus are two protocols: Modbus is a general industrial protocol, DL/T 645 is a meter-specific protocol — they cannot be mixed; The role of the standard — it unifies meter communication: meters from different manufacturers communicate by the same protocol — one gateway driver connects meters from many manufacturers; Versions of the standard — DL/T 645-2007 is common on site: frame format and data identifiers correspond to the version — confirm the meter version before connecting.

In one sentence: DL/T 645 = a dedicated communication protocol for multifunction energy meters + different from Modbus + unifies meter communication — many manufacturers, one protocol, one gateway driver — confirm the version before connecting.

A standard example: meters from two manufacturers on one project: both support DL/T 645-2007 — one gateway driver connected 48 meters — no need to write a separate driver for each manufacturer. The value of DL/T 645 is "unification" — many meter manufacturers connected with one protocol — choose the protocol right, and the connection becomes simple.

2. Difference One: Addressing — Data Identifier vs Register

Modbus addressing — Modbus addresses by register: reading address 40001 means reading a certain register — the address is the position; DL/T 645 addressing — DL/T 645 addresses by data identifier: reading "energy" requires specifying the data identifier (DI) — the identifier means "what to read"; The nature of the difference — Modbus is "position addressing": knowing the position, you can read; DL/T 645 is "semantic addressing": only knowing "what data to read" gets you the data; The consequence of the difference — reading DL/T 645 with Modbus thinking: reading addresses 1, 2, 3 — no energy data — you must read by data identifier.

In one sentence: the addressing difference = Modbus reads by register position + DL/T 645 reads by data identifier — position addressing vs semantic addressing — reading a 645 meter by Modbus addresses gets no data.

An addressing example: a gateway read the 645 meter with Modbus addresses 1-10 — all values returned as zero — after switching to reading energy by data identifier, the data was normal. The addressing method decides "where to read" — a 645 meter must be read by data identifier — reading by position reads nothing — change the addressing mindset first.

3. Difference Two: Data Identifiers (DI) and Common Data Items

The meaning of data identifiers — the data identifier (DI) is the data-addressing code of DL/T 645: every data item has an identifier — the identifier decides "which item to read"; Common item one: energy — combined active/reactive energy: forward/reverse, time-of-use energy by peak/flat/valley — the source of billing energy; Common item two: demand — maximum demand and its occurrence time: important for load management — demand data has a "freezing" characteristic; Common item three: events — undervoltage, phase loss, power outage and power-on events: power-quality and anomaly tracing — events carry timestamps; Reading the items — read by identifier groups: one frame reads a group of items — more efficient than item by item.

In one sentence: data identifiers = three common groups, energy + demand + events — read by identifier, one frame per group — energy for billing, demand for load, events for tracing.

A data-item example: a platform needed three types of meter data: energy (time-of-use by peak/flat/valley), demand (maximum demand and time), events (undervoltage/outage) — the gateway read by identifier groups, three frames done. Data identifiers are the "catalog of what to read" — energy, demand and events each taken where they belong — read by the catalog, and the data neither mixes nor misses.

4. Difference Three: Meter Address Range and Networking

The meter address — every meter has one address: the range 0-9999 (four-byte BCD) — networking distinguishes meters by address; Configuring the address — meter addresses are set at the factory and changeable: check the address list before connecting — a wrong address reads the wrong meter; The networking method — one RS-485 bus can carry multiple meters: poll by address — the 485 bus has limits on distance and quantity; Networking notes — the 485 bus: distance generally within 1200 meters, the meter count per bus by its capability — exceeding the limits requires zones or additional gateways.

In one sentence: address range and networking = one address per meter + 485 bus polling + limited distance/quantity — with correct addresses and bus planning, many meters connect in order.

A networking example: 32 meters in a workshop on one 485 bus: polling addresses 1-32 at 9600 baud — one full round of energy reads took about 30 seconds. Networking is "calling roll by address" — with the address list checked and the bus capability calculated, many meters connect in order without conflict.

5. Difference Four: Communication Rate and Timeout

Setting the rate — DL/T 645 commonly uses 1200/2400/4800/9600 baud: the rate must match the meter — a rate mismatch fails the communication; The limit of the rate — meter communication rates are generally lower than industrial devices: small data volumes, different real-time requirements — the gateway polling cycle must adapt; Setting the timeout — the inter-frame timeout is set by the protocol characteristics: the 645 frame has a fixed format and response time — too short a timeout falsely judges failure, too long slows polling; The relationship between rate and timeout — the rate decides "how long a frame takes", the timeout decides "how long to wait before declaring failure" — matched, and the polling is stable.

In one sentence: rate and timeout = baud rate matches the meter + timeout by the frame characteristics — the rate decides the frame duration and the timeout decides failure — matched, and the polling is stable.

A rate example: a gateway defaulted to 9600 connecting to a 2400-baud meter — 90% communication failures — after changing the gateway to 2400, it stabilized. A wrong rate breaks everything — aligning the baud rate is the first step — no matter how accurate the timeout, a wrong rate is useless.

6. Common Mistakes When Connecting

Mistake one: reading by Modbus registers — reading the 645 meter with addresses 1/2/3 — no energy data — the right way: read by data identifier; Mistake two: not checking the address list — meter addresses do not match the configuration — wrong-meter data read — the right way: check every meter's address before connecting; Mistake three: inconsistent rates — gateway and meter baud rates differ — communication failures — the right way: configure by the meter's actual rate; Mistake four: mismatched timeout — too short a timeout falsely judges failure, too long slows polling — the right way: set the timeout by the 645 frame characteristics.

In one sentence: the four connection mistakes = reading by registers + addresses unchecked + inconsistent rates + mismatched timeout — avoid the four pits and the connection works on the first pass.

A mistake example: a failed-connection investigation: read by Modbus addresses (wrong), address list unchecked (wrong), baud 9600 vs 2400 (wrong) — after correcting the three errors, the energy data was normal. Connection failures are mostly "habit errors" — Modbus thinking, careless addresses, default rates — ask the four questions and the problem surfaces.

7. The Unified Data Model After Gateway Conversion

The meaning of conversion — the gateway reads the DL/T 645 data and converts it: meter-protocol data → a unified data model — the upper platform does not care about the meter protocol; The fields of the unified model — the unified model contains standard fields: energy, demand, events, timestamp and meter number — the fields align with what the platform needs; The mapping of the conversion — 645 data items map to the unified model: energy → the energy field, events → the event field — the mapping table is configured once; The value of the conversion — the platform uses one data model to connect multiple device types: 645 meters and Modbus devices upload in a unified way — the protocol difference is digested by the gateway.

In one sentence: gateway conversion = 645 data into the unified model + standard fields + the mapping table configured once — one platform model connects many device types — the protocol difference is digested at the gateway layer.

A conversion example: a platform connected both 645 meters and Modbus meters: the gateway converted both protocols into the unified "energy/demand/events/timestamp" model — the platform read both meter types with one set of fields. The gateway is the "translator" — it translates 645 into one common language — the platform sees only the unified model and never touches protocol details.

8. Back to FEXLINK: How the ESX Intelligent Edge Gateway Connects

The ESX intelligent edge gateway has a built-in DL/T 645 driver: it reads energy/demand/events by data identifier and converts them into a unified upload. The configuration checklist: ① confirm the meter protocol and version — DL/T 645-2007 → ② check the meter address list — every address correct → ③ configure the baud rate and timeout — matching the meter → ④ configure the data-identifier mapping — energy/demand/events into the unified model → ⑤ verify the platform connection — data on the cloud, fields aligned — the 645 meter connection is done in one step by the gateway.

One summary: the points of DL/T 645 connection = addressing by data identifier + recognizing energy/demand/events + address-range networking + rate/timeout matching + unified gateway conversion — DL/T 645 is a meter-specific protocol whose addressing differs from Modbus — understand the four differences before connecting, and the gateway converts and uploads in a unified way. DL/T 645 is not "another Modbus" — read by identifier, configure by version, and the gateway digests the difference — the connection works on the first pass and the data uploads steadily.

One reminder: when connecting to a 645 meter fails, do not suspect the meter first — check four things: is it read by data identifier? Is the address list checked? Is the baud rate consistent? Is the timeout matched? — after the four questions, the problem surfaces. The pits of 645 connection are almost all "stepped on with Modbus habits" — read by identifier, check addresses, set the rate and configure the timeout — with the four steps right, the connection works.

Scope and Limitations

  • The conclusions in this article are intended for LV distribution and electrical-safety sites. Standard numbers (DL/T 645) refer to their currently effective editions, and product parameters follow the latest FEXLINK selection handbook and product knowledge base.
  • Quantitative criteria (protection zone, discharge and residual voltage, earth resistance, thresholds and grading) hold under the preconditions stated in this article. Where the site voltage level, earthing arrangement, environmental category or load profile differs, the figures must be recalculated. Specific configurations involving ESX intelligent edge computing gateway should be verified against actual circuit parameters.
  • Specific product models and parameters are subject to the latest selection handbook and product knowledge base; this article does not constitute selection or procurement advice.
  • Standard numbers cited here refer to the currently effective editions; when a standard is revised, the related quantitative conclusions must be re-checked.
  • Deployment and implementation for specific projects must be carried out by qualified engineers based on site survey and design drawings.

Frequently Asked Questions

DL/T 645 Meter Protocol

The meters on site use the DL/T 645 protocol — what is different when a gateway connects to them compared with Modbus devices