All guides

Knowledge baseConnect

Connecting data sources

Adding equipment and reading it: Modbus TCP, OPC UA, IEC 60870-5-104, MQTT and Sparkplug B, and datalogger files — with the time-zone settings that keep timestamps right.

Every source ends up in the same place: one history per tag, watched by the same alarms and drawn in the same charts. This guide walks through connecting each kind. The Data sources page lists every detail each one supports.

The common steps

  1. Equipment → Add device. Give it a code, a name, an alarm category and how it is read: files, Modbus TCP, OPC UA, IEC 60870-5-104 or MQTT. The protocol can be changed later on the Communication tab.
  2. Communication tab. Fill in the connection and the poll interval (per device, from 1 second to 1 day), then press Test connection — it probes the device without saving anything. A number outside its allowed range is refused with a message naming the field.
  3. Tag Mapping tab. Add the device’s tags with their address and data type, and link each to a system tag. Many tags at once can be imported from a CSV with Import CSV. See Tags, virtual tags and deadband.
  4. Monitoring & Alarms tab. Set the communication timeout — how long the device may stay silent before it is shown as offline — and, if you want to be told, the communication-lost alarm.

Each protocol runs in its own service, so a misbehaving device on one protocol never stops the others. Reading PLCs, MQTT and files is part of every licence.

Modbus TCP

SyncTide is the Modbus client: it reads holding registers, input registers, coils and discrete inputs, and never writes.

  • Connection: host, port (502 by default), unit ID, timeout and an address offset for devices that count from 1.
  • Addresses: classic 40001 / 30001 / 10001 / 00001 numbering, or explicit hr:100, ir:0, di:5, coil:12 (0-based).
  • Data types: int16, uint16 and bool on one register; int32, uint32 and float on two; int64, uint64 and double on four; or one bit of a register. The byte order (ABCD, CDAB, BADC, DCBA) is set per device.
  • Scaling: value × scale + offset, per tag.
  • Neighbouring tags are read in one request, so hundreds of tags take a handful of requests. Polled once a minute or faster, the connection stays open between polls.

Some PLCs close idle Modbus sessions and then refuse new ones for a while (could not connect now and then). Turn on the keep-alive option of the device’s connection so SyncTide holds the session open.

OPC UA

SyncTide is an OPC UA client on an opc.tcp:// endpoint, signing in anonymously or with a user name and password.

  1. Enter the endpoint, for example opc.tcp://192.168.1.50:4840, and test the connection.
  2. On the Tag Mapping tab, browse the server: walk its namespaces and node tree, see each variable’s value and type, and add the ones you want — no NodeIds to type by hand.
  3. Choose how values arrive: read on the poll interval (all of a device’s tags in one request), or subscribe and let the server send changes. With subscriptions, the tag’s deadband is also sent to the server, so unchanged values never cross the link.

NodeIds can also be typed: ns=2;s=Flow, ns=2;i=1234. Timestamps come from the server (the source time, else the server time, else the time of receipt). If the server answers BadIdentityTokenInvalid, it wants a user name and password.

IEC 60870-5-104

For utility RTUs. SyncTide is the controlling station: it keeps one connection open per device (port 2404 by default), asks for a general interrogation on connect, then takes the RTU’s spontaneous events as they come.

  • Common address (ASDU): it must match the RTU’s. If the RTU connects but no values arrive, check the common address first.
  • Points: one information object address (IOA) per tag, with its type — short floats, scaled and normalised values, single points, or any monitoring type named exactly.
  • Time: the RTU’s time tags (CP56Time2a) are kept, so events buffered during an outage arrive with their own time. Those time tags carry no time zone: set the zone of the RTU’s clock on the device.
  • Points the RTU sends that are not mapped are written once to the log with their type, so they can be mapped without another tool.

MQTT and Sparkplug B

SyncTide subscribes to an MQTT broker — your own, or the one in our cloud.

  1. Register the broker once in Configuration → Runtime → MQTT: host, port, user name and password, TLS.
  2. On the device, pick the broker and a topic prefix; SyncTide subscribes to prefix/# with the QoS you choose (0, 1 or 2).

Payloads can be one value per tag topic ({"value": 12.3}), several tags in one JSON message, a list of values with their own timestamps, or a bare number. Sparkplug B (NBIRTH, DBIRTH, NDATA, DDATA) is decoded automatically, with metric names and aliases from the birth messages and the metric timestamps kept.

Datalogger files

For dataloggers and systems that export CSV or Excel files, often over a cellular link and often late.

  • Where the files go: each equipment has its own folder (by default under the datalogger folder chosen at install, named after the equipment code). Your FTP or SFTP server, or a network share, drops the files there; you can also upload files from the browser. In our cloud, dataloggers send by FTP, FTPS or SFTP.
  • Formats: CSV with the delimiter detected (comma, semicolon, tab or pipe) and Excel .xlsx (first sheet). One column per tag, or long rows of time, tag and value; a units row is recognised.
  • Numbers: decimal comma or point; on/off, true/false and yes/no read as 1/0.
  • Safe reading: a file is read only once it has stopped changing, so a slow upload is never read half-way. A file sent twice is recognised by its content and skipped, and overlapping files never store a value twice.
  • Late files are filed at their own times, and alarms still fire for the crossings inside them.
  • Files that cannot be read are set aside with the reason and listed in the platform. They are tried again when the file, the equipment settings or SyncTide change, or when you retry them.

Check the first import of a new file layout. An ambiguous date such as 03/04/2026 is read day first. If a system writes month-first dates without saying so, a misread date silently lands rows on the wrong days rather than failing.

Time zones

Two separate settings answer two different questions. Getting them wrong never corrupts data, but it shifts the whole history on screen.

Setting Where What it answers
Timestamp time zone Communication tab How to read the timestamps arriving from this device
Site time zone Metadata tab Where the equipment stands — used to show its data

Which timestamp time zone to choose:

  1. The device sends UTC — choose UTC.
  2. The device sends local time and follows summer time by itself — choose the named zone, for example Europe/Lisbon.
  3. The device sends local time and never changes its clock (it was set by hand) — choose the matching fixed offset, for example Etc/GMT-1 for UTC+1 (the sign of the Etc/GMT names is reversed).

Changing a time zone never rewrites stored data, but it changes how every existing value is shown, in charts, reports and exports. Change it deliberately, and tell whoever reads the reports.