A merging unit is the device that sits between conventional instrument transformers and the digital world of IEC 61850. It takes the analog current and voltage signals from current transformers and voltage transformers, samples them at a precise rate, time-stamps each sample, and publishes them onto the process bus as digital Sampled Values. Every protection IED, bay controller, or meter that needs to know what the power system is doing gets its measurements from the merging unit — not from analog wiring.
Understanding what a merging unit does, how it publishes data, and how the IEC 61850 model represents it is essential for anyone designing, commissioning, or troubleshooting a digital substation process bus.
Table of Contents
Why Merging Units Exist
Traditional substations run analog wiring from current transformers and voltage transformers to every protection and measurement device in the bay. A bay with five relays means five separate pairs of CT and VT secondary wiring runs, each carrying the same analog signal. That wiring is expensive, difficult to test, impossible to modify without physical rewiring, and a source of errors during installation.
A merging unit eliminates this. One device acquires the CT and VT signals for a bay, digitizes them, and publishes them over Ethernet. Any device on the process bus — protection IEDs, bay controllers, meters, fault recorders — receives the same digital sample stream. Adding a new subscriber means adding a subscription in software, not running new copper.
The benefits are practical: reduced copper wiring, simplified testing, easier engineering changes, standardized digital interface, and deterministic timing through synchronization.
What a Merging Unit Does — Step by Step
1. Signal acquisition The merging unit connects directly to the secondary outputs of current transformers and voltage transformers. It accepts conventional CT and VT secondary levels — typically 1 A or 5 A for CTs, 100 V or 110 V for VTs — or the low-level analog outputs of electronic instrument transformers.
2. Analog-to-digital conversion The input signals are sampled by analog-to-digital converters at a fixed rate. The sampling must be precise and consistent. Jitter in the sampling process directly affects measurement accuracy and protection performance.
3. Time synchronization Each sample is time-stamped. For the samples to be usable by protection IEDs — especially those using phasor-based algorithms or differential protection across different bays — all merging units in a substation must sample at exactly the same instant. This requires synchronization to a common time reference, typically GPS-disciplined IEEE 1588 PTP or an IRIG-B signal.
4. Sampled Value publishing The digitized, time-stamped samples are packaged into Sampled Value messages and published over the process bus as Ethernet multicast frames. Subscribers receive the stream continuously.
5. Status and supervision The merging unit also monitors its own health and the health of the connected instrument transformers — fuse failures, communication loss, synchronization loss — and makes this information available over the station bus via MMS.
IEC 61850 Model Inside a Merging Unit
IEC 61850 models a merging unit as a physical device containing one or more logical devices, each with a defined set of logical nodes. The relevant logical nodes are:
LPHD — Physical Device Information
Every physical device has an LPHD logical node. It carries the device nameplate (PhyNam) and the physical device health status (PhyHealth). For a merging unit, PhyHealth reports the overall operational status of the hardware — whether the device is functioning correctly, degraded, or in alarm. These are mandatory attributes.
LLN0 — Logical Node Zero
LLN0 is the logical node that hosts the Sampled Value control blocks — one MSVCB (Multicast Sampled Value Control Block) or USVCB (Unicast Sampled Value Control Block) per SV stream. The data sets that define which samples are published are also defined here.
TCTR — Current Transformer Logical Node
TCTR represents one current transformer channel. It carries the sampled current values (Amp, of type SAV — Sampled Analogue Value) as the mandatory measured value. The samples are transmitted as engineering values — true primary current values, already corrected for transformer ratio. Correction factors (Cor, AngCor) and winding ratio (Rat) are optional settings used for maintenance purposes when an external conventional CT is connected.
Additional optional attributes include rated current (ARtg), rated frequency (HzRtg), external equipment health (EEHealth), and operation time (OpTmh). A separate TCTR instance is used for each phase — typically three for a three-phase bay, plus one for neutral if required.
TVTR — Voltage Transformer Logical Node
TVTR represents one voltage transformer channel. It carries the sampled voltage values (Vol, type SAV) as the mandatory measured value, again as true primary voltage values. Optional attributes include fuse failure status (FuFail — important for protection IEDs that need to detect VT failure), rated voltage (VRtg), rated frequency (HzRtg), and phasor correction factors.
A separate TVTR instance is used for each voltage channel — phase-to-earth or phase-to-phase depending on the transformer configuration.
MMXU — Measurement Logical Node (Optional)
Some merging units also implement MMXU, which calculates power system quantities from the sampled values — phase currents (A), phase-to-earth voltages (PhV), phase-to-phase voltages (PPV), active power (W), reactive power (VAr), apparent power (VA), power factor (PF), and frequency (Hz). These calculated values are published over the station bus via MMS for metering and monitoring applications.
MMXU is optional in a merging unit — its presence depends on the product and the application. When present, it allows the merging unit to serve as both a process bus SV publisher and a station bus measurement source.
Sampled Value Control Block — How Publishing Is Configured
Publishing of Sampled Values is controlled by the Sampled Value Control Block (SVCB). For multicast transmission — the standard mode for process bus — this is the MSVCB. Its key attributes:
SvEna — enables or disables transmission. When TRUE, the merging unit publishes the SV stream. When FALSE, transmission stops. This is the primary on/off switch for the SV stream.
MsvID — a system-wide unique identifier for this SV stream. Subscribers use this to identify which MSVCB a received message belongs to.
DatSet — the reference to the data set containing the samples to publish. The data set is defined in LLN0 and contains references to the SAV data attributes of the TCTR and TVTR logical nodes.
ConfRev — configuration revision counter. Increments when the data set configuration changes. Subscribers check this to detect that the merging unit’s configuration has changed since they were commissioned.
SmpRate — the sample rate, in units of samples per nominal period. This is not an absolute frequency — it is samples per power system cycle. At 50 Hz with SmpRate=80, the merging unit publishes 80 samples per cycle, or 4000 samples per second. At 50 Hz with SmpRate=256, it publishes 12800 samples per second.
OptFlds — optional fields included in each SV message: RefrTm (refresh time), SmpSynch (synchronization status), and SmpRate.
The SV Frame — What Goes on the Wire
Each Sampled Value message is a standard Ethernet frame with the following structure:
EtherType: 0x88BA — the registered EtherType for IEC 61850-9-2 Sampled Values. Receiving devices filter by this value.
Destination MAC address — multicast. The range 01-0C-CD-04-00-00 to 01-0C-CD-04-01-FF is reserved for IEC 61850 Sampled Values. The first three octets (01-0C-CD) are IEEE-assigned for IEC 61850. The fourth octet distinguishes the service: 01 for GOOSE, 02 for GSSE, 04 for multicast sampled values. The last two octets are assigned per stream during engineering.
APPID — a 16-bit application identifier. The reserved range for sampled values is 0x4000 to 0x7FFF. The default value of 0x4000 indicates no APPID has been configured — every SV stream in a system should have a unique APPID assigned during engineering.
VLAN tag (IEEE 802.1Q) — mandatory. Default VLAN priority for Sampled Values is 4. The VID should be different from the GOOSE VLAN to allow separate bandwidth allocation.
APDU — the payload, encoded using ASN.1 Basic Encoding Rules. It contains:
- noASDU — the number of Application Service Data Units (ASDUs) concatenated in this frame. Multiple ASDUs can be concatenated to reduce frame overhead at high sample rates.
- svID — the SV stream identifier from the MSVCB
- datset — the data set reference (optional)
- smpCnt — the sample counter. Incremented with every new sample. When the merging unit is synchronized and the synchronization pulse occurs, smpCnt resets to 0. The sample coinciding with the sync pulse is assigned smpCnt=0.
- confRev — current configuration revision of the MSVCB
- smpSynch — TRUE when samples are synchronized to a clock signal, FALSE when not
- smpRate — current sample rate (optional)
- sample — the actual data values from the data set, one entry per data set member
Sampling Rates — 80 and 256 Samples Per Cycle
Two sampling rates are commonly used in process bus implementations:
80 samples per nominal period — the standard rate for protection applications. At 50 Hz this gives 4000 samples per second; at 60 Hz, 4800 samples per second. This rate provides sufficient bandwidth for fundamental frequency protection algorithms and is widely supported by IED vendors.
256 samples per nominal period — used for power quality and metering applications requiring harmonic analysis. At 50 Hz this gives 12800 samples per second. The higher rate captures up to the 128th harmonic, well beyond protection requirements. Some merging units publish both rates simultaneously on separate MSVCB instances.
The SmpRate attribute in the MSVCB specifies which rate is in use. Subscribers must be configured to match the publisher’s rate.
Time Synchronization — The Critical Requirement
Sampled Values are only useful if the samples from different merging units are aligned in time. A differential protection function comparing current from two ends of a line must know that sample #1234 from merging unit A and sample #1234 from merging unit B represent the same instant in time.
This requires all merging units to synchronize their sampling to a common time reference.
IEEE 1588 PTP — the preferred method in modern IEC 61850 substations. PTP distributes time over the Ethernet network with sub-microsecond accuracy. Each merging unit acts as a PTP slave, receiving time from a grandmaster clock. When synchronized, smpSynch is set to TRUE in the SV message and smpCnt resets to 0 at each synchronization pulse.
IRIG-B — a legacy time distribution method using a dedicated timing cable. Still used in many substations, but being replaced by PTP as IEC 61850 deployments mature.
When synchronization is lost, the merging unit sets smpSynch to FALSE. Subscribers receiving a stream with smpSynch=FALSE should treat the samples with caution — they are still valid for local measurements but cannot be used for functions requiring cross-bay time alignment, such as differential protection or synchronized phasor measurement.
Process Bus Architectures
IEC 61850-9-2 describes several ways to arrange the process bus:
Per-bay architecture — each bay has its own process bus segment with its merging unit. Separate station-wide bus carries aggregated data. Provides isolation between bays but requires switches or routers to share data across segments.
Multi-bay segments — one process bus segment covers several bays. Reduces infrastructure but increases traffic per segment.
Station-wide bus — all devices on a single process bus. Simplest architecture but highest data rate requirement. The merging unit SV streams from all bays compete for bandwidth on the same network.
Function-oriented segments — bus segments aligned with protection zones, minimizing cross-segment data transfer. Requires routers at zone boundaries but reduces unnecessary traffic.
The choice of architecture affects latency, bandwidth planning, VLAN design, and redundancy requirements.
Merging Unit vs. Conventional CT/VT — Key Differences
| Conventional CT/VT | Merging Unit | |
|---|---|---|
| Output | Analog (1A/5A, 100V) | Digital Sampled Values over Ethernet |
| Wiring | Dedicated copper per device | Shared Ethernet network |
| Signal distribution | Point-to-point hardwiring | Multicast to all subscribers |
| Time alignment | Not inherent | Synchronized via PTP or IRIG-B |
| Adding subscribers | New copper wiring run | Software subscription |
| Testing | Injection per relay | Network-level injection |
| Failure detection | Difficult | smpSynch, PhyHealth, FuFail flags |
What to Check During Commissioning
SvEna — confirm it is TRUE and the stream is being published before connecting subscribers.
smpCnt — verify it is incrementing correctly and resetting to 0 at the synchronization pulse. A non-resetting smpCnt indicates synchronization loss.
smpSynch — must be TRUE for protection applications requiring time alignment. If FALSE, find the synchronization issue before commissioning differential or distance protection.
confRev — must match the configured value on all subscribers. A mismatch means the merging unit was reconfigured after the subscribers were commissioned — recheck data set membership and order.
APPID — must be unique per SV stream. Duplicate APPIDs on the same network cause subscribers to misidentify streams.
Multicast MAC address — must be in the range 01-0C-CD-04-00-00 to 01-0C-CD-04-01-FF. Verify no address conflicts with other SV streams.
FuFail (TVTR) — verify that VT fuse failure detection is working before commissioning voltage-dependent protection functions.
PhyHealth (LPHD) — verify the merging unit is reporting healthy status before relying on its output for protection.
IEC 61850-9-2LE vs IEC 61869-9
This distinction matters for procurement, commissioning, and interoperability. The two are often confused.
IEC 61850-9-2LE — The Implementation Guideline
9-2LE is not an IEC standard. It stands for IEC 61850-9-2 Light Edition — a vendor implementation guideline developed by the UCA International Users Group to address the lack of specific implementation constraints in the original IEC 61850-9-2 standard. The original 9-2 left too many degrees of freedom: it did not mandate a specific data set structure, sample rate, or number of ASDUs per frame. Different vendors made different choices, and interoperability suffered.
9-2LE solved this by defining fixed conventions:
- Fixed data set — 8 samples per ASDU: four currents (Ia, Ib, Ic, In) and four voltages (Va, Vb, Vc, Vn), each with a quality word
- Fixed sample rates — 80 samples per nominal period for protection, 256 for metering
- noASDU = 1 — one ASDU per frame, meaning one frame per sample
- Mandatory SmpSynch — synchronization status always included
9-2LE became the de facto standard for process bus implementations throughout the 2000s and early 2010s. Most legacy merging units and protection IEDs on the market today implement 9-2LE, not the formal IEC standard.
IEC 61869-9 — The Formal Standard
IEC 61869-9 is a proper IEC standard — part of the IEC 61869 series on instrument transformers. It defines a digital interface for instrument transformers and establishes a standardized process bus profile that supersedes the 9-2LE guideline.
Key differences from 9-2LE:
- Data set — IEC 61869-9 defines its own data set structure. It uses a different data model aligned with IEC 61869 instrument transformer definitions rather than the IEC 61850-7-4 TCTR/TVTR model.
- Quality encoding — quality bits are defined per IEC 61869-9 conventions, which differ from the IEC 61850-7-3 quality encoding used in 9-2LE implementations.
- Additional status information — IEC 61869-9 includes more detailed status and supervision data reflecting the instrument transformer’s operating condition.
- Sample rates — compatible with 9-2LE rates (80 and 256 per nominal period) but with additional options.
Interoperability Considerations
A merging unit publishing 9-2LE frames and a protection IED subscribing to IEC 61869-9 frames are not directly interoperable. The data set structure, quality encoding, and ASDU layout differ. In a mixed-vendor project, confirm which profile each device supports before finalizing procurement.
New projects should specify IEC 61869-9. Legacy projects upgrading existing 9-2LE infrastructure need to verify IED firmware support before changing the merging unit profile.
Network Bandwidth — Sizing the Process Bus
SV streams generate continuous, high-rate traffic. Bandwidth planning is a required engineering step before finalizing the network architecture.
Per-Stream Bandwidth
A typical SV frame is approximately 150–160 bytes including Ethernet overhead. The frame rate depends on the sample rate:
| Sample rate | Frequency | Frames/second | Approx. bandwidth |
|---|---|---|---|
| 80 s/c | 50 Hz | 4000 | ~5 Mbit/s |
| 80 s/c | 60 Hz | 4800 | ~6 Mbit/s |
| 256 s/c | 50 Hz | 12800 | ~15 Mbit/s |
| 256 s/c | 60 Hz | 15360 | ~18 Mbit/s |
These are per-stream figures — one merging unit, one MSVCB. A substation with 10 bays, each with one merging unit at 80 s/c, generates approximately 50 Mbit/s of SV traffic on the process bus.
Why Gigabit Ethernet Is Often Preferred
100 Mbit/s Ethernet can handle SV traffic in small substations, but the margin disappears quickly as bay count increases. A 100 Mbit/s segment shared between 10 or more SV streams, GOOSE, and management traffic is approaching its practical limit. Gigabit Ethernet is increasingly the baseline for process bus design, not because individual streams require it, but because the aggregate traffic and headroom for future expansion does.
noASDU Concatenation
IEC 61850-9-2 allows multiple ASDUs to be concatenated into a single Ethernet frame (noASDU > 1). This reduces frame count and overhead at the cost of latency — a frame with noASDU=8 at 80 s/c is only sent 500 times per second instead of 4000 times, but each frame carries 8 samples worth of data. For protection applications, noASDU=1 is strongly preferred to minimize latency. For metering applications, higher noASDU values are acceptable.
PRP and HSR Impact
When PRP or HSR redundancy is used on the process bus, every SV frame is duplicated. PRP sends two copies on two parallel LANs simultaneously. HSR sends two copies in opposite directions around the ring. The effective bandwidth consumed on each network path doubles.
For a 10-bay substation at 80 s/c on PRP, each LAN carries approximately 50 Mbit/s of SV traffic. This is the primary reason Gigabit Ethernet is recommended for PRP process bus designs rather than 100 Mbit/s.
Redundancy on the Process Bus
Protection-grade process bus requires zero-recovery-time redundancy. A merging unit that loses network connectivity — even for milliseconds — can cause protection IEDs to lose measurements and enter degraded states. Two redundancy approaches are used:
PRP (Parallel Redundancy Protocol)
The merging unit has two Ethernet ports, one connected to LAN A and one to LAN B. Every SV frame is sent simultaneously on both LANs. The subscribing IED receives both copies and discards the duplicate — exactly the same mechanism as PRP for GOOSE. If LAN A fails completely, frames continue arriving on LAN B with zero interruption. The IED detects the loss of duplicate frames through smpCnt gaps but continues operating on the surviving stream.
PRP is the most common redundancy method for process bus because it provides true zero-recovery-time and is straightforward to implement.
HSR (High-Availability Seamless Redundancy)
In HSR, the merging unit connects to a ring and sends each SV frame in both directions simultaneously. Subscribers receive both copies and discard the duplicate. A single ring break does not interrupt communication — the surviving path continues delivering frames.
HSR avoids the need for two separate LAN switches but adds forwarding delay at each ring node. For SV streams, this forwarding delay accumulates per hop, which must be accounted for in the latency budget for protection functions. Mode X (counter-duplicate suppression) is not applicable to SV streams — both copies must always circulate the full ring.
What Happens When Synchronization Is Lost
Redundancy protects against network failure but not against synchronization loss. If the PTP grandmaster fails and no backup grandmaster takes over within the timeAllowedToLive window, all merging units transition to smpSynch=FALSE. The SV stream continues — samples keep flowing — but subscribers can no longer use cross-bay time alignment. Local protection continues; differential protection and phasor-based functions that depend on synchronized samples should raise an alarm and may need to be blocked until synchronization is restored.
Design the PTP architecture with the same redundancy discipline as the SV network itself.
Conclusion
A merging unit is the foundation of the IEC 61850 process bus. Without it, digital protection IEDs have no measurements. With it, the entire bay becomes accessible to any subscriber on the network — protection, metering, monitoring, and fault recording — all from a single, synchronized, digital source.
The IEC 61850 model built around TCTR, TVTR, LLN0, and LPHD is what makes merging units from different vendors interoperable. The MSVCB controls what gets published and when. The SV frame carries the samples with the timing information subscribers need to act on them safely.
Get the synchronization right, size the network for the actual SV traffic load, choose between 9-2LE and IEC 61869-9 deliberately, and verify the configuration before enabling protection functions. The process bus works exactly as designed when all of these are in place.