IMDHMIControls LibraryShop main store

Selection guide

Selecting an S7-1215C CPU for Modbus TCP and OPC UA

PLC programming Published 8 min read

Short answer

Yes, the S7-1215C carries a Modbus TCP client to the subordinate controller and an OPC UA server for the edge gateway, once firmware V4.4 or higher and the SIMATIC OPC UA S7-1200 Basic option (6ES7823-0BA00-2BA0) are on the order; the two PROFINET ports settle wiring, not capacity, and at this scale both protocols together load the port by a hundredth of a percent.

On this page
  1. Is the S7-1215C enough for a Modbus TCP client and an OPC UA server?
  2. What does an S7-1200 need before its OPC UA server will run?
  3. Can the OPC UA server deliver what the edge gateway asks for?
  4. How do you size the Modbus TCP client load on the S7-1200?
  5. Do the two PROFINET ports mean two separate networks?
  6. Do four digital inputs and outputs need expansion modules?
  7. What exactly goes on the order for this control?
  8. Questions asked

Yes, on two order conditions. The S7-1215C runs the Modbus TCP client toward the subordinate controller as standard and carries the OPC UA server for the edge gateway from firmware V4.4 onward, provided the SIMATIC OPC UA S7-1200 Basic option, order number 6ES7823-0BA00-2BA0, sits on the order beside it. The two PROFINET ports settle wiring, not capacity: the arithmetic below puts both protocols together at about a hundredth of one percent of a port. What actually decides this selection is the delivered firmware, the gateway's update-rate requirement, and the electrical type of the four digital points.

Is the S7-1215C enough for a Modbus TCP client and an OPC UA server? #

Yes — the S7-1215C is a supported candidate for both duties at once, on two order conditions: firmware V4.4 or higher on the delivered CPU, and the SIMATIC OPC UA S7-1200 Basic option, order number 6ES7823-0BA00-2BA0.

Modbus TCP client duty is native to the S7-1200 family: no option, no gateway box and no extra license sits between your program and the subordinate controller's registers. The OPC UA server is the conditional half — it exists in the CPU only from a specific firmware and needs an orderable option — so the bill of materials, not the port count, is where this selection is decided.

Requirement in your architecture What the S7-1215C gives you What actually decides it
Modbus TCP client to the subordinate controller Native client instructions, no option needed One connection and one poll cycle; negligible load at this scale
OPC UA server for the edge gateway Server in the CPU from firmware V4.4, enabled by SIMATIC OPC UA S7-1200 Basic (6ES7823-0BA00-2BA0) Gateway's monitored items and interval against the documented sampling intervals
Four digital inputs and outputs Onboard I/O, no signal module Electrical type of each point, not the count
Two networks on two ports Wiring flexibility on one cell subnet Whether you truly need separated subnets, a network decision

The rest of this article walks each row in the order you meet it.

What does an S7-1200 need before its OPC UA server will run? #

Firmware V4.4 or higher on the CPU plus the SIMATIC OPC UA S7-1200 Basic option named above — with either one missing, there is no server to configure in the project.

Siemens documents that S7-1200 CPUs as of firmware V4.4 are equipped with an OPC UA server, and the support documentation states the same threshold from the other direction: OPC UA communication is supported from firmware version 4.4 upward. Two consequences for the order. Check the firmware of the exact unit you are quoted, because older stock can arrive below V4.4 and then needs a firmware update through the engineering tool before any server exists to configure. And carry the Basic option as its own line item, because nothing on the CPU's nameplate tells you it is there. The server is configured in the same project as the control logic, so plan the interface while you plan the program, not after.

Can the OPC UA server deliver what the edge gateway asks for? #

For sensor values sampled at 100 ms, 250 ms or 500 ms from a server interface you model yourself — yes; for automatic exposure of every CPU and DB tag, or updates below 100 ms — no.

The documented system limits for the S7-1200 OPC UA server at firmware V4.6 list possible sampling intervals including 100 ms, 250 ms and 500 ms. That set is the gate your gateway passes or fails: its required publishing interval has to land inside the documented set for the firmware you run, so pull the interval out of the gateway specification before the CPU ships, not during commissioning.

The second documented behavior shapes your engineering budget rather than the hardware: the server supports the standard SIMATIC server interface and does not support automatic publishing of CPU and DB tags. Every value the gateway reads has to be placed into the modeled interface deliberately, one by one, so a 200-variable gateway means 200 modeled entries regardless of CPU speed. And "sends sensor data" is not yet a specification: a subscription at 500 ms reports each variable twice per second, a polled read behaves differently, and the two load the server differently. If the edge gateway itself is still open in the project, an industrial mini PC beside the controller can host the OPC UA client and the local historian without putting a PC on the office network. If the gateway's list and interval fit the documented set, the 1215C carries the server side; if they do not, no CPU headroom fixes it — the requirement moves the server role to the S7-1500 class.

How do you size the Modbus TCP client load on the S7-1200? #

By transactions per second, not by port count: an illustrative poll of 20 holding registers every 250 ms over one connection works out to 676 bytes per second on the wire, about 0.005 percent of one 100 Mbit/s port.

Illustrative inputs, so re-run them with your real data map: the subordinate controller exposes 20 holding registers, you poll all of them every 250 ms over one connection, and the gateway subscribes to 40 variables with a 500 ms publishing interval.

  • One read transaction, function-code-03 style, on 20 registers: request = 7 bytes of MBAP header + 5 bytes of function code, address and quantity = 12 bytes; response = 7 bytes of MBAP header + 1 byte function code + 1 byte byte count + 20 x 2 bytes of data = 49 bytes.
  • Application bytes per transaction: 12 + 49 = 61 bytes.
  • Framing: each frame carries 14 bytes Ethernet + 20 bytes IP + 20 bytes TCP = 54 bytes of overhead; two frames per transaction add 2 x 54 = 108 bytes.
  • Wire bytes per transaction: 61 + 108 = 169 bytes.
  • Transaction rate at a 250 ms cycle: 1 / 0.25 s = 4 per second.
  • Modbus wire load: 4 x 169 = 676 bytes/s = 5408 bit/s, about 5.4 kbit/s, which is 5408 / 100 000 000 x 100 percent, about 0.005 percent of one 100 Mbit/s port.
  • OPC UA side: 40 variables x 1 / 0.5 s = 80 value reports per second, at roughly 10 bytes encoded each = 800 bytes/s, about 6.4 kbit/s.
  • Both links together: 5.4 + 6.4 = about 11.8 kbit/s, about 0.012 percent of one port.

Read that as a decision, not trivia: the wire and the ports are nowhere near binding at this scale, so stop sizing the network. What consumes an S7-1200 is executing the client instructions and serving the subscription inside the scan, and 4 transactions per second plus 80 values per second is small duty for this family. One Modbus peer is the smallest configuration the family's client is built for; the per-firmware connection ceiling is a system-manual number, and nothing in this architecture depends on approaching it. One boundary no sizing number covers: neither link is a safety function — interlocks that must stop the machine stay in hardwired or fail-safe I/O, so a stalled Modbus response or a dropped gateway session degrades operation, never protection.

Do the two PROFINET ports mean two separate networks? #

No — treat them as wiring, not routing: the subordinate controller and the edge gateway can share one cell subnet, and the two ports give you placement and line-topology flexibility on that subnet.

The ports buy you a daisy chain without an extra switch: the controller chain on one port, the gateway side on the other, or any mix that suits the cabinet. What the port count cannot buy is a conclusion about capacity — the arithmetic above cleared that axis three orders of magnitude ago. If your plant standard requires the Modbus cell and the OPC UA path on separated subnets, that is a network architecture requirement and it changes the parts list, a CPU or router documented for that separation, not a load you can compute away. The question of how much traffic one Ethernet interface can share is the one behind checking whether one S7 CPU PN port can run Modbus TCP and S7 comms together, and there too the answer turns on configuration, not on the number of jacks.

Do four digital inputs and outputs need expansion modules? #

No — four inputs and four outputs, even read as four points total, sit inside the onboard digital I/O of an S7-1215C with margin, so the count does not move the CPU selection.

The requirement wording is ambiguous between four of each and four in total, and the decision survives both readings. What does need settling before wiring is electrical type: 24 V DC inputs on the DC variants, relay versus transistor outputs, commoning groups, and whether any point needs a high-speed function that pins it to specific terminals. Those are datasheet checks against the exact variant ordered; none of them moves the CPU choice. If the point count grows later, signal modules extend the station without revisiting the CPU or either protocol.

What exactly goes on the order for this control? #

Two items beyond the CPU itself: confirmation of firmware V4.4 or higher on the delivered unit, and the SIMATIC OPC UA S7-1200 Basic option, order number 6ES7823-0BA00-2BA0.

  • Modbus TCP client duty: order nothing beyond the CPU. One connection, one poll cycle; re-run the arithmetic above with the real data map and file it.
  • OPC UA server duty: the firmware condition plus the option, and a gateway whose monitored items and interval fit the documented sampling set.
  • Network: one cell subnet; the second port is placement. A two-subnet requirement is a network decision that changes the parts list, not a CPU upgrade.
  • Digital points: onboard I/O; settle the electrical type of each channel at wiring design.
  • Growth: the escalation trigger is the OPC UA side — more modeled variables, additional clients, intervals below the documented set. That is where the S7-1500 class in the same portfolio takes over, and it is a server-limit decision, not a wire-speed one.

For the architecture as stated — one Modbus peer, one edge gateway, four digital points — order the S7-1215C with the option, and spend the saved sizing effort on the server interface model, because that is the only part of this project that grows with scope.

Questions asked

Does the OPC UA server also run on the fail-safe S7-1200 CPUs?

Yes. The firmware V4.4 OPC UA server covers the S7-1200FC fail-safe CPUs in addition to the standard S7-1200C CPUs, so a fail-safe variant of the same architecture keeps the server for the edge gateway. The two order conditions, firmware level and the Basic option, are unchanged.

What if the edge gateway needs updates faster than 100 ms?

Then the requirement, not the CPU, decides. The documented sampling intervals for the S7-1200 server at firmware V4.6 list 100 ms, 250 ms and 500 ms among the possible values, so a gateway demanding 50 ms subscriptions has no documented setting to use. That is the trigger to size an S7-1500 class CPU for the server role instead of stretching the 1215C.

How many Modbus TCP client connections can one S7-1200 handle?

The documented maximum is published per CPU and firmware in the S7-1200 system manual, and this project needs exactly one connection, the subordinate controller. One peer on one poll cycle is the smallest configuration the family's Modbus client is built for, so the ceiling never enters this selection.

Does a dropped gateway connection stop the Modbus polling?

No. The Modbus client and the OPC UA server are independent communication functions in one CPU: polling continues on its cycle while the server waits for the gateway to reconnect, and a slow Modbus response does not pause the server. Program a fault routine for each link separately, and treat a lost gateway as a data-path event, not a control event.

Keep reading

  • Selection guide

    Connecting ControlLogix to RS232 and RS485 serial devices

    Convert the RS232 devices to RS485 at the electrical layer, then pick one master for the whole pair: 1734 ASCII modules for fixed-length or delimited frames, an in-chassis Modbus RTU module for a pure Modbus bus, or an EtherNet/IP-to-serial gateway with a protocol builder when variable-length checksummed frames share the wire with Modbus RTU. One master serializes every transaction behind the 3.5-character silent gap, 3.65 ms at 9600 baud.

    Industrial networks

  • Selection guide

    How to size an S7-400 CPU against a PLC-5/40 scan time

    No S7-400 CPU maps one-to-one to a PLC-5/40 by instruction timing, and none needs to: every CPU from the 414 upward is faster on logic execution. Select on memory, I/O and communications capacity, then accept the choice only when the translated program's measured worst-case cycle lands at or below two-thirds of the measured PLC-5/40 scan.

    PLC programming

  • How-to

    Can one S7 CPU PN port run Modbus TCP and S7 comms together

    Yes. A 414-5H's integrated PN interface runs a Modbus TCP server for the DCS and S7 communication for the operator station on the same port, because both are TCP-based services on one IP stack. What decides the project is the CPU's connection-resource total, the station's transport type, the loaded response delay against each timeout — and on an H pair, the fact that Modbus sessions break at switchover unless you use the redundant option on a CP 443-1 pair.

    Industrial networks