After a motor trips on overload, an electronic motor protection relay latches a thermal-model value of 100% and blocks restart until the calculated winding temperature decays, so a brief power-cycle cannot be used as a workaround [S1].
The behaviour is governed by the relay class: bimetallic units rely on a physical bimetal cooling curve, while electronic overload relays and motor protection relays run a numerical thermal model in non-volatile memory that survives a control-power loss [S1][S2].
What Thermal Memory Actually Stores in a Motor Protection Relay
Thermal memory is a stored percentage representing the relay's calculated estimate of motor winding temperature as a fraction of trip threshold, and on a C440-style electronic overload it sits at 100% immediately after a trip event [S1]. The relay will not release its trip output, and on auto-reset models it will not permit a restart, until the stored value falls back below the unlock threshold, which mimics a real stator cool-down [S1][S7]. The mechanism is the digital equivalent of waiting for a solder pot or bimetallic strip to physically re-solidify, but it is programmable and is not a fixed minutes-per-ampere value [S3].
Electronic overloads such as the Allen-Bradley E100, E200, and E300 all expose a percent thermal capacity utilization diagnostic in their feature matrix, alongside percent full-load amperes, and that TCU register is what is preserved across a power cycle [S2]. In a motor protection relay like the Bulletin 857, settings, events, and indications live in non-volatile memory, so a trip, a control-power blip, and a re-energise will not silently clear the thermal model [S2]. The reference for the broader protection taxonomy is the protection relay encyclopedia entry.
Manual Reset vs Automatic Reset, and Why Power-Cycle Reset Is a Misconception
The physical reset knob on a thermal overload relay is a latch release, not a temperature knob: it does not cool the motor, it only re-arms the trip contact after the bimetal has bent back [S3]. On bimetallic units, most are wired so a manual reset is required and is mechanically blocked while the element is still hot, which is why a tripped overload cannot be re-closed by simply power-cycling the cabinet [S3]. On an E100 electronic overload, the 193-1EE variant is manual-reset-only across trip class 10 and 20, while the 193/592-1EF supports automatic or manual reset with trip class 10, 15, 20, or 30 [S2].
Auto-reset on a hot motor is exactly the failure mode thermal memory retention is designed to defeat: a Motor Control Center that auto-restarts on return of control power can re-energise an overheated motor and drive it straight into a second insulation-killing overload [S6]. The standard MCC mitigation is to keep thermal memory retention enabled and to add or extend a restart-inhibit timer, not to rely on the upstream breaker dropping out to clear the trip [S6]. For wider context on the relay tier below a full motor protection relay, the thermal relay entry covers the bimetallic and eutectic alloy variants this logic replaces.
Trip Class, Reset Mode, and How They Set the Cool-Down Window
Trip class sets the inverse-time trip curve the relay uses during an overload, with Class 10 being the fastest cold-trip, Class 20 the default for standard IEC motors, and Class 30 reserved for high-inertia loads such as fans and heavy conveyors [S2]. The same parameter also bounds the auto-reset cool-down, because most electronic overloads gate automatic reclosing on the TCU register falling below a configurable threshold within the trip-class time window [S2][S4]. E100, E200, and E300 all support phase loss, ground fault, jam, current imbalance, over- and under-voltage, and over- and under-power as standard, so the motor protection relay can hold its restart inhibit based on any of those latches, not only thermal [S2].
Setting the overload incorrectly defeats the protection just as effectively as removing it: NEC requires overload protection sized at 125% of motor full-load current, so a 60 A motor should be set at 60 A on a relay with the 125% factor built in, or 75 A on one without it, and a 50 A dial left in place will nuisance-trip under load rather than under protect [S5]. A thermal-memory value stored across a power cycle is only as trustworthy as the FLA setpoint that fed the model, which is why a motor protection relay that survives a brown-out but was set wrong will still restart into a fault on the first re-energise [S5].
Bimetallic vs Electronic: Why Drift, Not Memory, Is the Bimetallic Risk
Bimetallic thermal overload relays do not retain a numerical thermal memory, they retain heat in the bimetal itself, and that mass-based retention is exactly what makes them vulnerable to calibration drift after repeated heating cycles [S4]. Each trip mechanically stresses the bimetal, and over thousands of operations the element shifts its trip point, producing nuisance trips on cold starts and missed trips on real overloads [S4]. Electronic overloads eliminate the mechanical drift path because there is no bimetal to bend, but they replace it with a requirement to keep the non-volatile TCU register validated and to confirm the relay actually re-reads it after a control-power cycle [S1][S2].
Thermal memory retention is therefore not a single number, it is a stack: a percentage register that survives power loss, a trip-class-derived unlock time, and a reset mode that can be either operator-confirmed manual or fully automatic with inhibit [S2][S6]. The architectural choice cascades into the upstream device: a microcomputer protection on the feeder can add a second-stage thermal model, but the motor branch relay still owns the restart permission. Engineers sizing replacement overloads on legacy starters should treat any retrofit from a bimetallic to an E100-class electronic unit as a chance to also re-baseline FLA, trip class, and reset mode, because the new relay will only honour thermal memory as well as its parameters were entered.
Selection Criteria: When Retention Matters, When It Is Overkill
Thermal memory retention is a hard requirement on any motor that sits close to its insulation class limit and is fed from an MCC bus that can re-energise automatically after a control-power dip, which covers most continuous-duty pumps, compressors, and conveyors in chemical and water plants [S6]. It is overkill on small single-phase fractional-horsepower motors with a built-in line-level thermal switch, where a manual reset pushbutton on the motor frame is the intended re-arming path and there is no electronic state to preserve [S3]. It is also unnecessary on motors already protected upstream by a motor protection relay with its own non-volatile event log and a hardwired restart-inhibit contact.
Decision rules: (1) if the branch uses a bimetallic overload and a motor protection relay is already present, leave the bimetallic in manual reset and let the relay supervise restart. (2) If the branch is electronic-only, specify non-volatile thermal memory, automatic reset disabled by default, and a configurable restart-inhibit timer sized at 1.5 to 2 times the trip-class time. (3) On any VFD-driven motor running above 60 Hz, confirm the overload rating covers up to 400 Hz, which the E100 line does [S2]. (4) On nuisance-tripping bimetallics, treat repeated heating and cooling as a calibration drift symptom and replace the element rather than raising the FLA setpoint, since the latter silently sacrifices protection [S4].
Failure Modes and Trackable Signals After a Power Cycle
Two failure modes show up in the field. First, a motor that restarts into a hot winding because thermal memory retention was either not enabled or was cleared by a firmware reload; the visible signature is a second trip within seconds of the first re-energise, with a TCU register that does not match the relay's own cool-down curve [S1][S6]. Second, a motor that refuses to restart after a power cycle because the relay has held its 100% thermal value and the auto-reset window has not yet elapsed; the visible signature is a healthy-looking drive held off by an overload whose TCU register is decaying on schedule, and the corrective action is to wait the unlock time or invoke a controlled manual reset after a measured winding temperature [S1][S2].
Trackable signals going forward: NEMA MG-1 parts 14 and 20 revisions continue to tighten motor thermal-limit expectations, which puts pressure on relay vendors to extend TCU retention past the typical 5-minute cool-down into the 15 to 30 minute band; on the user side, the practical node to monitor is the percent thermal capacity utilization register on E100/E200/E300 and equivalent relays, read via Logix or DeviceLogix, and alarming on a sustained value above 80% as a pre-trip warning [S2]. For a cost-side read on the motors these relays protect, the 2026 electric motor pricing breakdown is a useful cross-check on the replacement-cost argument for keeping the protection healthy.