Overview
The question of "resolution" in a PROFINET context is frequently conflated with the protocol's data transport capacity. PROFINET itself is a deterministic, real-time Ethernet protocol that moves binary process data in defined I/O telegrams; it carries no inherent decimal point and no native engineering unit. The "resolution" of a measured value—the number of significant digits or smallest detectable change—is a property of the field device's transducer, its analog front end, and the data type declared in the device's GSDML. The controller (a Siemens SIMATIC S7-1500 CPU 1513-1 PN in this case) only consumes, scales, and stores the bits the device publishes.
This reference consolidates the limits that govern how much raw data a CPU 1513-1 PN can move on PROFINET, the maximum number of IO Devices and IO Modules it can address, and the engineering path required to bring a high-resolution field instrument—such as a weighing cell transmitter—into a STEP 7 / TIA Portal project without bottlenecking the bus.
The two commonly deployed hardware variants are 6ES7513-1AL02-0AB0 and 6ES7513-1AM03-0AB0, listed in the PROFIBUS & PROFINET Product Finder and the TIA Portal SIMATIC S7-1500 / ET 200MP manual collection respectively. The AM03 revision is the current production variant and the one documented in the Siemens cloud documentation for the ET 200MP / S7-1500 family.
PROFINET Architecture: Telegrams and Data Containers
PROFINET IO follows a producer/consumer model in which the IO Controller (the CPU) cyclically exchanges input and output data with IO Devices. Each IO Device exposes one or more slots, and each slot contains a subslot/module. The aggregated payload of all modules of one IO Device is the device's process data image—the value the controller reads or writes per cycle.
Three real-time classes matter for sizing:
- RT (Real-Time, Class 1)—cyclic data with priorities 0 (high) and 1 (low). Send clocks from 250 µs upward in practice; 1 ms is the practical minimum for most non-motion applications.
- IRT (Isochronous Real-Time)—scheduled with bandwidth reservation, used for motion and isochronous mode where jitter must be below 1 µs.
- NRT (Non-Real-Time)—standard TCP/IP/UDP/IP traffic for acyclic parameterisation, diagnostics, web server, and PROFIenergy.
The PROFINET IO telegram (V2.3+) carries an IO Data Container whose length is signalled in the first two octets. The controller reads this length to decode the slot structure. There is no implicit "resolution" parameter—every bit the device publishes is significant because PROFINET does not perform unit conversion in transit.
CPU 1513-1 PN Hardware Specifications
The CPU 1513-1 PN is part of the SIMATIC S7-1500 standard controller family with a single integrated PROFINET interface (X1 with an internal 2-port switch). Two order numbers are in active deployment:
| Order Number (MLFB) | Firmware | Work Memory (Code / Data) | Bit Memory | PROFINET IO Devices, max. | IO Modules, max. | Redundancy |
|---|---|---|---|---|---|---|
| 6ES7513-1AL02-0AB0 | V2.6 / V2.9 | 300 KB / 1.5 MB | 16 KB | 128 | 2 048 | MRP (manager/client) |
| 6ES7513-1AM03-0AB0 | V3.0 / V3.1 | 300 KB / 1.5 MB | 16 KB | 128 | 2 048 | MRP + S2 system redundancy |
Both variants ship with the X1 PROFINET interface supporting 100 Mbps full-duplex, MRP (Media Redundancy Protocol) client, and the PROFIenergy profile. Port 1 and Port 2 of the integrated switch are exposed on the front 35 mm shroud with pin assignments per IEC 61918. The newer 6ES7513-1AM03-0AB0 additionally supports the SYNC function for IRT and the S2 redundancy role. The S7-1500 / ET 200MP manual collection specifically documents the AM03 behaviour for "Failure of both redundancy connections and the PROFINET cable in the PROFINET ring" and "Failure of the two PROFINET cables in the PROFINET ring on the backup CPU," confirming the controller's PROFINET ring behaviour.
When used as MRP Manager, the CPU 1513-1 PN closes a ring of up to 50 PROFINET devices. In S2 mode (AM03) the controller and a backup CPU can each observe both ports; under a single-port failure the system continues to deliver process data without reconfiguration, satisfying the PROFINET conformance class CC-B (MRP) and CC-D (S2).
PROFINET Data Limits: Segment, Device, and Slot
The PROFINET IO specification (IEC 61784-2 / IEC 61158) caps the user data per IO Device and per controller. The numbers below are the contractual limits and the working envelope a TIA Portal engineer can rely on when planning a topology:
| Scope | Maximum User Data | Notes |
|---|---|---|
| Per PROFINET segment (cumulative across all IO Devices) | 8 192 bytes input + 8 192 bytes output (8 KiB each direction) | Throughput-budget driven; the practical ceiling before update time degrades. |
| Per IO Device (input) | 1 440 bytes | PROFINET IO limit. Slots may be modular or compact; aggregate must fit. |
| Per IO Device (output) | 1 440 bytes | PROFINET IO limit. Independent of input budget. |
| Per slot / module | 254 bytes input + 254 bytes output | Slot-level cap; typical modules use 2–32 bytes. |
| Number of slots per device | 256 (incl. head module slot 0) | Defined in the GSDML. |
| Number of IO Devices on CPU 1513-1 PN | 128 addressable; practical ≈ 64–80 | Limited by the device table in the CPU's technical data and the cycle-time budget. |
| Number of IO Modules on CPU 1513-1 PN | 2 048 | Total across all IO Devices. |
The 1 440-byte per-device figure is the same in both directions. A typical weighing cell with 2-byte status + 4-byte gross weight + 4-byte tare + 4-byte net weight consumes 14 bytes—less than 1 % of a single device's budget. The CPU 1513-1 PN can therefore host hundreds of high-resolution weighing instruments on one PROFINET line without exhausting the device budget; the wire-level bandwidth and the send-clock setting become the real limiters, not the protocol's count of devices.
PROFINET Topology Reference
Signal Resolution: Where the Limit Lives
PROFINET is bit-exact. The smallest delta a CPU can detect is 1 least-significant bit (LSB) of the data type signalled by the IO Device's GSDML. The choice of data type and the engineering range of the device jointly set the resolution visible to the application:
| Data Type (TIA Portal) | Width (bits) | Codomain | Distinct Codes | Typical Use |
|---|---|---|---|---|
| BOOL | 1 | 0 / 1 | 2 | Digital I/O, status bits |
| BYTE / USINT | 8 | 0 … 255 | 256 | Flags, simple counts |
| WORD / UINT | 16 | 0 … 65 535 | 65 536 | Status words, 16-bit ADC |
| INT | 16 | −32 768 … 32 767 | 65 536 | Signed 16-bit field values |
| DWORD / UDINT | 32 | 0 … 4 294 967 295 | 4.29 × 10⁹ | Counters, 32-bit ADC |
| DINT | 32 | −2 147 483 648 … 2 147 483 647 | 4.29 × 10⁹ | Signed 32-bit process values |
| REAL | 32 | ±3.4 × 10³⁸ | ~7 decimal digits | IEEE-754 single precision |
| LREAL | 64 | ±1.8 × 10³⁰⁸ | ~15 decimal digits | IEEE-754 double precision |
| LWORD / LINT / ULINT | 64 | ±9.2 × 10¹⁸ | 1.84 × 10¹⁹ | High-resolution counters / timestamps |
For a weighing cell transmitter, the published "resolution" is usually given in display counts (e.g., 100 000 d for a 50 kg cell = 0.0005 kg/d). To get the wire-side resolution right, specify a transmitter that publishes its primary value as a 32-bit signed integer (DINT) or 32-bit float (REAL) at the full 100 000-d count. TIA Portal will then place the value into a tag of the corresponding type, and no further bit-level math is required to read individual counts.
Resolution Math: From Counts to Engineering Units
The wire-level step (1 LSB) becomes a physical step through the device's published scale factor. Use:
ΔEng = (Range_max − Range_min) / (2^N − 1)
For a 32-bit signed value (N = 31 significant bits, codomain size = 2³¹) scaled to 0–50 000 kg:
ΔEng = 50 000 / (2^31 − 1) ≈ 2.328 × 10⁻⁵ kg (≈ 0.023 g per LSB)
This is the smallest deterministic step the controller can register. Mechanical noise, thermal drift, and ADC quantisation in the transmitter typically dominate; the bus is rarely the limiting factor for weighing applications. For perspective, a 24-bit ADC (common in weighing transmitters) yields 16 777 216 codes, which over a 50 000 kg range equates to ≈ 0.003 kg/LSB at the ADC. The PROFINET wrapper adds no further quantisation if the transmitter publishes the full 24-bit count in a 32-bit slot.
Engineering Workflow in TIA Portal
After installing the device's GSDML (TIA Portal V18 / V19 / V20) the slot structure is fixed. To verify the data path and the resolution budget:
- Open the device's Properties → Device view; confirm the slots list matches the GSD.
- Open Properties → IO tags; map each slot to a tag name and PLC data type (BYTE, INT, DINT, REAL, etc.).
- Open Properties → PROFINET interface → Real-time settings; set the send clock (1 ms for 1 kHz update, 4 ms typical for weighing).
- Open Properties → Module parameters; on a weighing transmitter, set the integer scale, units (kg, g, t), and the status-word bits to match the device's manual.
- Compile the hardware configuration and download. In the online view, force the IO tags to verify wire-level data integrity.
- Optionally assign the IO Device's PROFINET name and IP from TIA Portal's "Assign PROFINET device name" wizard.
Sample SCL snippet for scaling a 32-bit signed integer to engineering units when the device sends a raw count in the %IW area:
// Raw DINT from PROFINET input area
// "iwWeightRaw" is mapped from slot 1, subslot 1
#rGross_kg := DWORD_TO_REAL(UDINT#0)
+ DWORD_TO_REAL("iwWeightRaw")
* (50 000.0 / 2 147 483 647.0); // unsigned map
#bValid := "iwWeightRaw".%X15; // bit 15 = status valid
For a REAL type (32-bit float), no scaling is required in the PLC; the engineering value arrives in the tag as written by the device.
Practical Example: Weight Cell Integration
A 50 t load cell cluster with 100 000 d display resolution and a PROFINET transmitter (e.g., a Siemens SIWAREX WP241 / WP251 or a HBM FIT / AED class device) is integrated as follows:
- Slot 0: device head module (empty data)
- Slot 1: 4 bytes input — gross weight (DINT, 0.5 kg/LSB typical)
- Slot 2: 4 bytes input — net weight (DINT)
- Slot 3: 2 bytes input — status word (UINT)
- Slot 4: 2 bytes input/output — control word / commands (UINT)
Total IO footprint: 12 bytes input + 2 bytes output per device. The bus cycle at 4 ms send-clock consumes ~28 µs of line time per device, leaving ample headroom for 30+ load cells on a single PROFINET line of the CPU 1513-1 PN. Even with 128 devices of the same type populated, the total is 1 536 bytes input + 256 bytes output — well within the 8 192-byte per-segment budget and inside the 1 440-byte per-device limit per slot aggregate.
Cycle-Time Budget
Use the following dimensional check before commissioning:
T_total = Σ (T_device_update × N_devices) + T_switching + T_guard
With N_devices = 30 at T_device_update = 100 µs, a fast IRT-capable switch, and 200 µs of guard band, T_total ≈ 3.2 ms, comfortably under a 4 ms send clock. The same formula with 128 devices at 100 µs gives ≈ 13 ms, which mandates a 16 ms send clock to stay below the watchdog. This is the typical engineering trade-off: device count vs. update rate.
PROFINET Conformance Classes and Their Impact
The CPU 1513-1 PN supports the following PROFINET conformance classes (per IEC 61784-2):
- CC-A — RT, basic diagnostics, non-realtime mixed. Default for the AM03.
- CC-B — adds MRP redundancy, device-level diagnostics, and SNMP. Required for ring topologies.
- CC-C — adds IRT bandwidth reservation. Required for motion and isochronous applications.
- CC-D — adds S2 system redundancy. Only the 6ES7513-1AM03-0AB0 supports this; the 6ES7513-1AL02-0AB0 does not.
For weighing and other non-motion applications, CC-B with a 4–8 ms send clock is usually sufficient. The send clock is a global parameter on the controller's PROFINET interface and applies to all ARs (Application Relationships) on the segment.
Cable and Topology Constraints
Although this is not strictly a PROFINET protocol limit, the physical layer sets the upper bound on device count and segment length. For copper PROFINET (100BASE-TX, Cat 5e or better):
| Topology | Max Devices per Segment | Max Cable Length (copper) | Notes |
|---|---|---|---|
| Line (no switch) | Limited by device count | 100 m device-to-device | Single point of failure; only CC-A. |
| Star (managed switch) | Up to 128 | 100 m segment, 6 hops max | CC-A / CC-B; standard for cabinets. |
| Ring (MRP) | Up to 50 recommended | 100 m segment | CC-B; < 200 ms reconfiguration. |
| Ring (MRPD) | Up to 50 | 100 m segment | CC-B; zero-loss reconfiguration with IRT. |
| Redundant S2 | Up to 128 | 100 m segment, 2 controllers | CC-D; AM03 only. |
For installations exceeding 100 m between devices, use PROFINET over fibre (100BASE-FX) with SC or LC connectors via a media converter or a PROFINET switch with integrated fibre uplinks. The CPU 1513-1 PN does not have native fibre ports; an external switch is required.
Verification and Diagnostics
After download, run the following checks before handing the system over to operations:
- Online → Diagnostics → PROFINET diagnostics: confirm the device state is green; no submodule diagnostics pending.
- Watch table: monitor the raw IO tags, apply a known load to the cell, and verify the LSB-to-kg conversion matches the device's published spec.
- Trace: record the input image over 10 s and verify the cyclic update; check for any glitches or hold-last-value events.
- Web server of the CPU ("http://<CPU-IP>/") → "Diagnostics" tab to inspect port counters, discard rate, jitter, and active ARs.
- PG / PC Ethernet statistics: confirm zero CRC errors, zero late collisions, and zero symbol errors on the PROFINET port.
- PROFINET name and IP audit: each device's PROFINET name must match the project; mismatches are the single most common commissioning fault.
Troubleshooting Matrix
| Symptom | Likely Cause | Action |
|---|---|---|
| Value jumps by 2× or 256× between cycles | Data type mismatch (DINT read as INT, or byte swap) | Check slot → tag mapping; toggle byte order in the IO Device properties; verify INTEL / BIG_ENDIAN attribute in the GSDML. |
| Always zero or always saturated (0x7FFFFFFF) | Status bit not set; device in "service" / "calibration" mode | Read the status word; issue the "operate" command from the control word; confirm zero / span calibration is released. |
| Bus error "IO Device failure" (red) | PROFINET name / IP conflict, or device power loss | Re-assign the device name from TIA Portal; verify the topology editor and the physical port wiring; check the device's 24 V supply. |
| Cycle time exceeded (jitter / watchdog) | Send clock too short for the IO footprint | Increase the PROFINET update time in the controller's properties; check the number of ARs; reduce the device count per controller. |
| Negative or unchanging reading despite applied load | Tare active; calibration zero not released; wrong sign convention | Send a "zero / tare reset" command via the control word; re-zero mechanically; verify the engineering range sign. |
| Random spikes in the reading | EMC on the PROFINET cable; poor shield bonding | Verify Cat 5e/Cat 6 shielding, equipotential bonding, and separation from VFD cables (≥ 200 mm). |
| Device drops off the ring after switchover | MRP not configured; device does not support MRP | Enable MRP Manager on the CPU; verify all ring participants are MRP-capable; check the role assignment (Manager / Client). |
Glossary of Key Terms
- AR (Application Relationship)—the cyclic data channel established between the controller and a device; consumes one of the controller's AR slots.
- CR (Communication Relationship)—a record-level channel within an AR, used for acyclic read/write of parameters and diagnostics.
- GSDML—XML device description file (General Station Description, Markup Language) that defines the slot structure, parameters, and diagnostics of a PROFINET device.
- Send clock—fundamental cycle time of the PROFINET IRT/RT schedule, set on the controller and shared by all ARs.
- Watchdog—time within which the controller must refresh a device's output data; if missed, the device enters fail-safe state.
- Submodule—synonym for a slot-level module that publishes a defined IO length.
FAQ
What is the maximum data size per PROFINET IO device on a CPU 1513-1 PN?
The PROFINET IO specification caps the user data of a single IO Device at 1 440 bytes for inputs and 1 440 bytes for outputs. The CPU 1513-1 PN (6ES7513-1AL02-0AB0 and 6ES7513-1AM03-0AB0) accepts up to 128 such devices in its device table and 2 048 IO Modules in total.
How many decimal places can PROFINET transmit?
None. PROFINET does not carry a decimal point; it transports raw bits. Decimals are a property of the engineering unit set in the field device's GSDML and the data type (REAL, LREAL, DINT) selected in TIA Portal.
Is the CPU 1513-1 PN the resolution bottleneck for a weighing cell?
No. The controller reads whatever bit-width the transmitter publishes. For 0.001 kg resolution, specify a transmitter that outputs a 32-bit float (REAL) or a 32-bit signed integer scaled to the required engineering range; the 1 440-byte per-device limit is not approached.
What is the maximum PROFINET send clock on the CPU 1513-1 PN?
250 µs in IRT mode with the SYNC function enabled (firmware V2.6 or higher on the 6ES7513-1AL02-0AB0; V3.0 / V3.1 on the 6ES7513-1AM03-0AB0). RT operation supports 1 ms as the practical minimum, with 4 ms typical for non-motion applications such as weighing.
Does the CPU 1513-1 PN support PROFINET redundancy?
Yes. Both PROFINET ports of the X1 interface participate in MRP on every variant. The 6ES7513-1AM03-0AB0 additionally supports S2 system redundancy, allowing seamless switchover in a ring or paired-controller configuration, as documented in the S7-1500 / ET 200MP manual collection for the AM03.
How is the per-device 1 440 B budget distributed across slots?
Each slot may consume up to 254 bytes input + 254 bytes output. A typical weighing cell with 14 bytes total IO fits in a single slot; a complex modular IO Device with 8 submodules may approach 1 440 bytes without exceeding the slot cap.
What happens if the cumulative segment payload exceeds 8 192 B?
The PROFINET update time degrades proportionally. For deterministic behaviour, keep the segment payload under 8 KiB in each direction and adjust the send clock to match the resulting cycle time. Above this budget, isolate the segment via a managed switch.