A self-powered electronic motor protection relay harvests its operating energy directly from the current transformers on the protected line, so the relay is fully active only when load current is present; an auxiliary-powered device draws from an external 24-250 VDC (or universal AC/DC) station supply, keeping logic, metering, and communication ports live even at zero load.
The choice is no longer either/or on most modern numerical platforms: vendors now ship dual- or self-powered units (Fanox SIA-B, Schneider Electric) that default to CT energy and switch to auxiliary DC when present, while full-featured motor relays such as the SEL-710 are still engineered around an external supply for thermal model accuracy, RTD inputs, and IEC 61850.
What "self-powered" actually means in a numerical relay
Self-Power defines a supply mode in which the electronic protection relay takes all operating energy from current transformers installed on the line, with no auxiliary voltage required [S1]. The relay's internal switch-mode power supply rectifies and stores that CT secondary current; for medium-voltage units, the CT can be a standard 1 A or 5 A type, or a "specific" CT whose non-standard secondary gives higher saturation margin and a wider measurement range [S2].
Because every joule in the relay is stolen from the protected line, trip coils (typical 50-200 VA class) are impractical: self-powered units drive a low-energy striker of roughly 100 mJ, with the spring charged while the breaker is closed [S2]. If a coil-type breaker is already installed, an external energy-accumulator adapter charges in about 10 seconds from any available auxiliary and holds roughly three days of autonomy for one trip after supply loss [S2].
Where self-powered design still loses to auxiliary supply
Self-powered designs face three documented limits. First, switching onto a fault: with no pre-load, the relay must cold-start from the fault current itself, so the effective operate time is the declared operate time plus the relay's start-up time [S3]. Second, test-set interaction: the integrated SMPS presents a non-linear load, so a nominal 1 A sinusoidal injection from a relay test set can be read as a much higher or lower current at the relay terminals, and conventional primary injection is the only fully reliable verification [S3]. Third, the zero-load blind spot: with no current in the protected feeder, the relay is de-energized, so sustained undervoltage, phase reversal, and many thermal functions cannot run continuously [S3].
These are the exact gaps that auxiliary-powered motor relays are designed to close. The SEL-710 runs a full enhanced thermal model with locked-rotor, time-between-starts, starts-per-hour, antibackspin, and load-loss elements, and adds up to 10 internal RTDs (12 with the SEL-2600 module), PTC thermistor input, underpower, reactive power, and loss-of-potential on the optional voltage card [S4]. Standard communications include Modbus RTU, MIRRORED BITS, IRIG-B, and optional IEC 61850, Modbus TCP/IP, DeviceNet, and single or dual copper or fiber Ethernet ports [S4]. None of that is feasible from CT energy alone.
Decision matrix: self-powered, dual-powered, or auxiliary-powered
For an RMU or pole-top feeder below roughly 800 kVA, or any retrofit where no DC battery is present, a self- or dual-powered unit such as the Fanox SIA-B, SIA-C, or Schneider Electric's Self/Dual-Power mode is the rational pick: it eliminates the battery, the charger, and the annual maintenance visit, and the auxiliary-supply function can be added later for trip-coil breakers via the TCM-class adapter [S1][S2][S6].
For a 400 V or 6 kV motor control center driving a critical pump, compressor, or fan, an auxiliary-powered numerical relay such as the SEL-710 is the rational pick: continuous thermal model, RTD-based stator temperature, differential overcurrent (87M), load jam, underpower, and grid-friendly protocols (IEC 61850, Modbus TCP) all require stable supply voltage and continuous logic power [S4]. For dual-powered architectures, the TI TIDA-00229 reference design shows the canonical topology: a wide-input DC-DC converter fed from a rectified CT secondary, with a MOSFET shunt regulator and Zener-clamped auxiliary DC input, generating stable 12 V and 3.3 V rails to the host MCU [S5].
Application fit, not ideology
Self-powered relays are now common down to 100 kVA distribution transformers, versus the traditional 800 kVA threshold, because the same economics that justified them for Smart Grid kiosks (lower cost, no battery) keep moving downward [S3]. For motor applications, however, the dominant self-aligning-bearing of decisions is thermal-model continuity: a motor drawing nameplate current and then stalling can be detected only if the relay was already running, and that requires either pre-fault load (the very condition the relay cannot guarantee) or an external supply.
The practical rule of thumb is therefore straightforward. Specify self- or dual-powered CT-fed relays for new MV feeder bays, RMU retrofits, and any substation where you are explicitly trying to retire the battery. Specify auxiliary-powered microcomputer protection for motors above ~75 kW, for any application that needs continuous thermal modeling, RTD-based stator protection, differential (87M), or IEC 61850, and accept the battery or station-service burden as the cost of those features. For the middle ground, the dual-powered topology with a CT-fed primary rail and an auxiliary DC backup gives both, and reference designs such as TI TIDA-00229 make the implementation straightforward [S5].
For cabling decisions that sit alongside this relay choice, the XLPE insulated power cable spec map lines out the voltage, conductor, and insulation classes that pair with each MV motor feeder. Trackable signals: confirm whether your shortlisted relay's stated start-up time (cold-start from fault current) is published and where it is tested, and whether the vendor's IEC 61850 data model includes the motor-specific LN classes (MMXU, MMTR) your substation IED needs.