Telemetry

What it is (precise)

A telemetry point carries three things: a value, a timestamp, and a quality flag saying whether the value is live, stale or bad. All three matter downstream.

The value is what people picture; the timestamp is what lets a market operator line a response up against the instruction that caused it, and the time-synchronization entry covers how the clocks are aligned to make that comparison honest; the quality flag is what stops a lost measurement from being read as a real zero, which is the difference between a resource that looks non-responsive and one that looks like it has a communications problem.

Points are produced by devices — PCS controllers, the BMS, protection relays, meters — marshalled by a remote terminal unit or protocol gateway, and published on a fixed schedule.

Telemetry travels one way, upward. The instructions that arrive from a market operator — a base point, a regulation signal, a curtailment order — travel the other way and belong to automatic generation control and the plant controller. Settlement is a third path with its own equipment: the revenue meter's registers, read on a slower schedule and held to a metering accuracy class, are what the invoice is built from.

The telemetered MW value is not the number that gets paid, and it is not meant to be. That is why an interconnection agreement lists metering and telemetry as separate requirements with separate acceptance criteria, and why the two are demonstrated separately during commissioning.

Telemetry is data and an obligation, not a box. The plant-side system that collects, displays and historises it is SCADA; the field device that marshals a defined subset onto the utility circuit is the remote terminal unit; the path itself may be fibre, a leased circuit or a licensed radio link.

Vendors supply those boxes, but the market accepts nothing until the points arrive in the required form. The distinction matters commercially because the boxes usually sit in the integrator's scope while the point list comes from the interconnection agreement and the market registration — two documents the integrator did not write.

The point list a storage plant actually sends

At the point of interconnection the electrical set is small: active power, reactive power, voltage, frequency and often current, all signed to a stated convention. Around it sit status points — breaker and disconnect positions, connection state, operating mode, and the alarm or fault words the plant uses to explain itself. Operators also want present capability rather than nameplate, because capability is what dispatch can rely on: available active power in each direction, available reactive power at the current operating point, and the reason for any shortfall.

Storage adds the signals that make it storage. A thermal unit reports one maximum; a battery plant has two, and they are not symmetric. Maximum discharge power and maximum charge power move independently with state of charge, cell temperature, converter derates and whatever the BMS is currently allowing. State of charge is the signal that makes the pair interpretable — an operator seeing 100 MW of discharge capability needs to know whether four hours of energy stands behind it or four minutes.

That is the reason SOC became a telemetered quantity at all: FERC Order 841 (2018) required each FERC-jurisdictional RTO and ISO to build a storage participation model that respects the resource's physical and operational characteristics, and the models that followed pulled SOC out of the plant and onto the wire. ERCOT sits outside FERC jurisdiction and arrives at the same place through its own protocols, which require energy storage resources to telemeter state of charge.

Two North American standards put a floor under the content. IEEE 1547-2018, which governs distribution-connected distributed energy resources wherever it is adopted, devotes Clause 10 to interoperability: a DER in scope must expose a local interface supporting at least one of IEEE 1815 (DNP3), IEEE 2030.5 or SunSpec Modbus, and must make a defined monitoring set readable — active power, reactive power, voltage, frequency, operational state and alarm status among them.

IEEE 2800-2022 addresses transmission-connected inverter-based resources and sets plant-level measurement and monitoring expectations, with the specific list settled with the transmission system operator. Vendors publish against these: SMA's IEEE 2800 statement for its SC / SCS UP-US family lists an AC sampling rate of at least 6 kHz for the transient time frame of that standard's Table 2. Measurement inside the converter is orders of magnitude faster than anything telemetered outward, and the two rates answer different questions.

Cadence, protocol and where the latency actually sits

Cadence is a market rule, and it is jurisdiction-specific enough that carrying an assumption across a border is a design error. As of mid-2026, PJM's regulation rules run on a two-second telemetry cycle for resources following the signal; CAISO's direct-telemetry business practice calls for values updated at least every four seconds; ERCOT collects resource telemetry from qualified scheduling entities under its own Nodal Protocols and Operating Guides.

In the EU, Regulation (EU) 2016/631 — the RfG network code — creates the obligation but delegates the content, leaving the relevant TSO or DSO to specify which signals are exchanged and how often, which is why two projects inside the same synchronous area can carry different point lists.

What the RfG does fix numerically is on the control side: from type A upward, a power-generating module must have a logic interface able to cease active power output within five seconds of an instruction arriving at that port. Treat every figure in this paragraph as a pointer to a current rule document rather than as a constant — market manuals are revised on their own schedule.

Protocol choice follows the same regional pattern. In North America the utility-facing link is usually DNP3, standardised as IEEE 1815, with ICCP/TASE.2 (IEC 60870-6) between control centres. In the ENTSO-E area IEC 60870-5-104 fills that role.

IEC 61850 appears as the information model and substation bus, its 7-420 part covering distributed energy resources specifically, while Modbus TCP dominates at device level almost everywhere — the datasheets confirm it, with Power Electronics listing Modbus TCP as the communication protocol across the Freemaq PCSK and PCSM families, EPC Power's CAB1000 offering Modbus TCP, CAN and Modbus RS-485, and SMA's controllers speaking Modbus TCP upward toward SCADA.

A plant therefore normally spans at least two protocol domains with a gateway between them, and that gateway is where point maps get mistranslated.

The equipment end of the chain is not the bottleneck, and the sizes are worth knowing. EPC Power's CAB1000 brochure quotes command latency of 1 ms over CAN and 3 ms over Modbus TCP/IP. Against a two- or four-second telemetry cycle the converter's contribution rounds to nothing; the budget is spent on polling intervals, gateway hops, protocol translation and the wide-area link.

High-rate recording is a separate duty and does not substitute for operational telemetry: Power Electronics' technical note comparing its PSAMS recorder with its Fast Logger quotes sampling up to 2,000 Hz for PSAMS and up to 1,000 Hz in the CAISO-oriented configuration. Kilohertz recording answers what happened during a disturbance; seconds-cadence telemetry answers what the plant is doing right now, and a project can be complete on one and short on the other.

Why it is a market-qualification item

Telemetry is not an IT detail because three separate documents each impose part of it, and the controls contractor writes none of them. The interconnection agreement specifies the signals delivered to the transmission or distribution operator, the path that carries them and often an availability figure for that path. The market registration package specifies what the resource must send to be eligible for each product it wants to offer into.

In North America, NERC's operating standards close the loop: under the TOP-003 and IRO-010 data-specification requirements, the Balancing Authority, Transmission Operator and Reliability Coordinator issue data specifications, and the entities named in them are obliged to supply the listed data. Telemetry is how a generator owner discharges that obligation, so a missing point is a compliance matter rather than a preference.

It is also gated by test. Before a resource clears for a product the operator normally wants an end-to-end demonstration through its own systems: points mapped and scaled correctly, values that move when the plant moves, sane timestamps, and a response to an issued test signal inside the product's window. That is a dated commissioning event on the critical path to commercial operation, sitting alongside the other acceptance tests.

Once the plant is operating, performance scoring in pay-for-performance regulation markets is computed from the telemetered response against the issued signal, which puts telemetry quality directly upstream of the payment. A plant whose values arrive stale or scaled wrong can score as a poor performer while behaving correctly at the terminals.

Common pitfalls

Two defects account for most of the pain, and point-to-point testing finds both where document review does not. The first is scaling and units — a point mapped in kW where the receiving system expects MW, or a per-unit value referred to the wrong base, produces a number that is plausible and wrong. The second is sign convention.

A storage resource is both generator and load, and the BMS, the PCS, the plant controller and the market system can each carry a different polarity for charging, so the site needs one written convention and an end-to-end check against a known injection. The sign-convention entry works through the failure modes; the one that bites hardest here is the pair of errors that cancel during a discharge-only witness test and reappear later on the charge side.

Then there are the failure modes of the link itself. Ask what the receiving system sees when the path drops — a held last value, a flagged stale value or a zero — and whether the plant's own behaviour changes once it can no longer be seen. Ask what the state-of-charge point actually means, because a percentage of usable energy, a percentage of nameplate and an absolute MWh figure are three different signals sharing one label, and the market's definition is the one that governs.

Watch aggregation too: the operator normally receives plant-level values, but a single derated block sets real capability, so the owner's internal telemetry set should be granular enough to show the limiting unit even where the market-facing set is not.

Finally, settle the point list early. Adding a signal that was never mapped means touching device configuration, the RTU database, the gateway translation and the operator's own model, then repeating the witnessed test — a sequence measured in weeks when it surfaces at commissioning and in hours when it is caught at design review. The list is available long before the equipment is, because it comes from the interconnection agreement and the market registration, and both exist on paper while the containers are still being built.