Wireless Battery-Free Bearing Temperature Measurement for High-Performance Electric Powertrains

In elite motorsport, performance margins are measured in milliseconds. Detecting component anomalies early can make the difference between winning and losing.

In Ferrari Sports Management’s hybrid power unit, the temperature of the MGU-K bearing became a critical indicator of rising friction and early degradation. At the same time, it remained extremely difficult to measure directly.

A battery-free wireless telemetry solution changes that. It enables direct, real-time measurement of bearing temperature on the rotating component itself. This once inaccessible signal becomes useful data for predictive performance and reliability.

Introduction

The MGU-K is a key component of Formula 1 hybrid power units, introduced in 2014, that functions as both an electric motor and generator. It captures kinetic energy during braking to charge the battery.

In modern motorsport power units, performance margins are so narrow that early detection of emerging anomalies can determine whether a race is completed or prematurely terminated. 

In this context, Ferrari Sports Management identified bearing temperature inside the MGU-K as a critical predictive indicator of frictional increase and incipient degradation—yet one that is fundamentally inaccessible with conventional instrumentation. Rotating geometry, metallic enclosure, and severe electromagnetic interference prevent direct and timely observation of bearing-level thermal behavior. 

To overcome this limitation, Radio6ense engineered a custom, battery-free wireless temperature telemetry system based on passive RAIN UHF technology. It is capable of operating directly on the rotating bearing without cabling or local power. 

The system leverages Radio Frequency IDentification (RFID) principles in the 860–960 MHz band. This enables battery-free sensing capabilities. We engineered it to maintain a stable near-field data link that is resilient to severe electromagnetic interference and metallic shielding. 

Through native integration with DewesoftX via open-source openDAQ SDK, we measure the temperature of the rotating bearing. This previously unreachable signal becomes a fully synchronized measurement channel, continuously correlated with electrical and mechanical quantities in the same test and measurement environment. 

We validated the solution through test-cell experiments on a Ferrari F1 electric power unit reproducing realistic on-track profiles. Instead of post-event diagnostics, the measurements reveal early and bearing-level anomalous thermal signatures that precede friction growth and mechanical degradation—signatures entirely invisible to a conventionally mounted PT1000 probe or IR sensor. 

By enabling accessible, real-time, bearing-level temperature measurements inside a rotating electric machine, this work shows how hybrid wired–wireless sensing within Dewesoft X provides actionable, physics-based insight for predictive maintenance, performance optimization, and data-driven decisions that directly influence on-track performance. 

Motivation and background: why bearing temperature monitoring matters more in today’s electric powertrains 

Electric powertrains are fundamentally changing how mechanical subsystems operate. Compared to internal combustion engines, modern electric machines run at higher speeds, experience sharper torque changes, and are increasingly designed for extreme power density to meet efficiency, range, and packaging demands. As a result, they face thermal, mechanical, and electromagnetic stresses that were far less pronounced in earlier automotive platforms.

Within this evolving context, rolling-element bearings have become critical to both performance and reliability. Elevated mechanical loads, rapid load reversals from regenerative braking, and high-frequency electrical phenomena induced by fast-switching inverters simultaneously affect these. Stray currents, voltage gradients, and EMI can accelerate lubrication breakdown, promote surface damage, and trigger premature degradation—even in otherwise well-designed systems. 

Among the various quantities monitored through continuous monitoring, temperature at the bearing itself remains one of the earliest and most reliable physical indicators of abnormal behavior.

Localized thermal rises precede friction growth, lubrication degradation, misalignment, and other incipient failure mechanisms. However, such signatures only retain diagnostic value if they are captured exactly at the bearing and with sufficient temporal fidelity—requirements that are incompatible with the architecture of a rotating electric machine. 

The machine structure makes direct access to bearing-level temperature extremely difficult. The bearings are enclosed inside compact metallic assemblies with complex thermal behavior. These assemblies are also exposed to vibration, fast transients, and severe electromagnetic interference.

As a result, conventional sensors can provide only indirect high temperature data. The signal is spatially averaged and temporally filtered, which hides the early, localized anomalies engineers need to detect.

The main limitation is not sensor accuracy. It is access to the measurement point itself.

Although this measurement challenge comes from automotive engineering, specifically from motorsport, it reflects a much broader industrial need. In motorsport, components operate at the limit, so detecting anomalies early is critical.

The same challenge is now appearing across many industries. As electric powertrains and other high-power rotating machines become faster, more compact, and more power-dense, it is becoming increasingly important to measure what is happening inside sealed rotating subsystems.

Access to these physically meaningful measurements is no longer a niche requirement. It is becoming essential for diagnostics, validation, and reliability engineering across transportation, industrial automation, and energy systems.

Stakeholders 

Our work emerged from a tightly coupled engineering collaboration between end users, wireless sensing developers, and Test & Measurement platform providers. 

Ferrari F1 ERS (Energy Recovery System) team – end user and customer

Ferrari defined the measurement challenge during the development and validation of a high-performance electric power unit. They identified bearing-level temperature in the MGU-K as a critical early indicator of rising friction and potential failure. 

Based on this need, they established the functional requirements and integration constraints for embedding sensors directly into the component.

Radio6ense – wireless sensing technology provider 

Radio6ense designed and engineered the custom passive, battery-free RAIN RFID temperature sensors for the bearing and the associated wireless telemetry solution. 

This creation included the electromagnetic co-design of rotating and stationary antennas for operation within the MGU-K under severe mechanical and EM/EMI constraints, as well as the development of software components to expose the sensing data as structured measurement channels. 

Dewesoft Italy – test & measurement platform and integration support 

Supported the openDAQ-based integration of third-party wireless sensing technology into the Dewesoft data acquisition ecosystem.

Temperature measurement for bearing-level diagnostics in the MGU-K 

During the development and validation of the F1 car electric power unit, bearing temperature in the MGU-K (Motor Generator Unit – Kinetic) emerged as a critical indicator of lubrication state, friction evolution, and early-stage degradation. 

From an engineering perspective, bearing temperature is one of the most informative indicators of emerging failure mechanisms—provided you can measure it at the right location and time. 

At the time of this work, the best available and currently implemented temperature measurement solutions for the MGU-K relied on two established sensing approaches, each representing the state of the art within its respective measurement paradigm. 

• Infrared (IR) temperature sensing offered the most physically representative reference measurement under controlled conditions. When line-of-sight access to the bearing surface was available, IR measurements provided valuable insight into localized thermal behavior. We therefore used the sensors as a benchmark in dedicated test configurations. However, the intrinsic requirement for optical access restricted IR sensing to open-bench or partially disassembled setups, with the shaft removed and the assembly exposed. 

As a result, IR measurements could not be applied to a fully assembled power unit and were inherently incompatible with realistic test cell operation, on-vehicle testing, or in-race conditions. 

• Conventional wired temperature sensors, such as PT1000 probes, represented the only viable solution for routine testing on a fully assembled power unit and therefore constituted the baseline implementation in the Ferrari test cell. These sensors are robust, accurate, and well-suited for integration into harsh environments. However, due to installation constraints, they can only be mounted on stationary structural components in proximity to the bearing. 

Consequently, the measured temperature is strongly influenced by heat-conduction paths and structural thermal inertia, yielding spatially averaged and temporally smoothed signals. Under identical operating profiles, this indirect measurement approach lacks the sensitivity needed to resolve localized bearing-level thermal deviations and to discriminate between nominal and degraded conditions early. 

Taken together, these considerations highlight a fundamental limitation. The available sensing solutions either provide physically representative data in non-representative configurations (IR) or representative data in non-representative configurations (wired sensors). 

This limitation defined the core engineering challenge Ferrari posed to Radio6ense: to enable a physically representative, bearing-level temperature measurement directly on the rotating MGU-K bearing, compatible with a fully assembled power unit and capable of supporting anticipatory diagnostics within advanced Test & Measurement workflows. 

Wireless telemetry

Bringing wireless telemetry inside a sealed, rotating electric motor is not a measurement challenge—it is an electromagnetic challenge.

Rotation, metallic enclosure, inverter noise, and millimetric integration constraints create one of the harshest environments in which any wireless link must operate. Unless we first engineer a passive, rotation-invariant coupling mechanism, no sensing architecture can function reliably. 

RAIN RFID technology as an enabler

Industry has historically adopted RAIN RFID Technology in the UHF band (860-960 MHz) at large scale for identification, tracking, and logistics. The same physical principles now enable battery-free sensing, thanks to UHF tags equipped with on-chip sensors that report physical variables such as temperature. 

For the MGU-K application, the operating principle of the RAIN system is straightforward: 

• The reader (active device) remains entirely outside the motor; 

• A fixed interrogating antenna, embedded in the MGU-K housing, injects the RF field generated by the reader inside the cavity; 

• The bearing-mounted sensor tag (passive device) requires no battery, harvesting a few microwatts of RF energy to operate its sensing circuitry; 

• The tag transmits the measured value by modulating the field it reflects to the reader (backscatter). 

Powering, sensing, and communication occur through a single passive interaction—no heat generation, no wires, and no electronics exposed to the internal environment. 

The proposed telemetry architecture is compliant with the EPC Gen2 / ISO 18000--63 standard, ensuring interoperability with industrial UHF readers and seamless integration into the Dewesoft Test and Measurement ecosystem. 

Electromagnetic design 

To make this architecture work inside the MGU-K, two passive components were co-designed as a single, tightly coupled near-field system: 

1. A battery-free RAIN Sensor, consisting of a capacitive dipole antenna connected to the Magnus S3 RAIN IC from Axzon. The Magnus S3 includes an on-chip temperature sensor (±0.5 °C typical accuracy, −30 °C to +120 °C range), which is fully compatible with the thermal envelope of the MGU-K bearing. 

2. An interrogating antenna, embedded inside the MGU-K housing and connected to an external UHF RAIN RFID reader through a coaxial feedthrough. We likewise implemented this antenna as a C-shaped dipole, capacitively coupled to the RAIN sensor antenna. 

Their electromagnetic behavior was optimized jointly by means of Full-wave EM simulations (CST Solver) so that:

• The coupling region remains strictly near-field and E-field dominant (minimizing susceptibility to inverter-generated magnetic noise), 

• The link budget remains stable across all shaft angles. 

• Metal proximity does not collapse or detune the interaction. 

The outcome is a validated, fully passive telemetry layer that can be co-located with the bearing and integrated inside the motor, providing the technical foundation—and the first proven example—of fully passive, in-motor temperature telemetry at the bearing level. Figure 4 shows the resulting prototypes of the RAIN sensors and the reader antenna. 

System architecture

 The telemetry layer already produces true physical measurements: each interrogation cycle yields a decoded temperature sample generated directly by the passive sensor on the rotating bearing. We synchronized these measurements into a single data stream in DewesoftX. 

Our work represents one of the first operational demonstrations in which a third-party HW - here a RAIN reader - is fully integrated into the Dewesoft ecosystem through openDAQ, allowing a passive in-motor telemetry source to behave exactly like a native Dewesoft acquisition device. 

We organized the system architecture into three functional layers, each responsible for a distinct stage of the measurement chain—from physical acquisition, to abstraction, to final integration into the Test & Measurement environment, as schematized in Figure 5. 

Physical layer 

At the physical layer, the system performs wireless measurement and data extraction: 

• The interrogating antenna inside the housing provides the RF excitation. 

• The passive RAIN RFID tag performs precise temperature readings on the rotating bearing. 

• The UHF RFID reader retrieves the sensor payload via backscatter. 

The reader executes an embedded SW layer (HEXA®) that manages Gen2 air-interface transactions, queries the Magnus S3 sensor register, decodes the payload, and performs lightweight on-edge pre-processing. 

The output of this layer is therefore a stream of clean, application-level temperature samples, accompanied by sensor identifiers and acquisition metadata—not raw RF signals. These samples form the physical measurement data that feed the next stage. 

openDAQ bridge 

The openDAQ bridge receives the reader’s UDP measurement stream and exposes the entire RFID subsystem as a standard openDAQ device. Its purpose is not to reinterpret the data, but to express them within Dewesoft’s measurement architecture: 

• Each passive sensor becomes a well-defined measurement channel. 

• Units, identifiers, and channel properties according to the openDAQ model. 

• A time base suitable for Dewesoft acquisition. 

• The device becomes discoverable and accessible as any native acquisition source. 

Bidirectional communication allows Dewesoft actions—such as configuration parameters, reporting behavior, or diagnostic requests—to be propagated back to the RFID subsystem when required. 

DewesoftX data acquisition software

Once exposed via openDAQ, we import bearing-level temperature channels into DewesoftX as fully native signals. We can view, log, process, and—most importantly—synchronize them with electrical and mechanical quantities such as torque, speed, inverter currents, and cooling parameters. 

Measurements

We conducted an experimental campaign to evaluate whether direct temperature sensing at the bearing level, on the rotating MGU-K element, can reveal meaningful thermal behavior under realistic operating conditions. All tests were performed on a controlled yet fully representative power-unit dynamometer test bench, shown in Figure 7.

We exercised the MGU-K under on-track–equivalent operating profiles, including rapid load transitions, high-torque regimes, and fast thermal excursions. The rotating bearing was instrumented with a passive RAIN RFID temperature sensor and interrogated by the in-motor UHF reader antenna, which we connected via a coaxial feedthrough to a Kathrein RRU1440 reader positioned outside the housing. 

For benchmarking, we acquired two conventional temperature-sensing technologies in parallel: 

• A wired PT1000 mounted on a stationary structural component near the bearing; 

• A non-contact infrared (IR) sensor, used only when line-of-sight was available. 

Figure 7 shows the sensor placement for the test bench setup. 

We acquired and time-aligned all channels in DewesoftX through openDAQ-based integration. These included RFID, PT1000, IR, and motor variables such as speed, torque, and electrical power.

The test campaign included multiple repetitions of the same driving-cycle profile with different bearings. Each bearing was installed in the same MGU-K assembly and tested under identical mechanical, electrical, and thermal conditions.

During these repeated tests, one bearing showed a clearly abnormal thermal response, even though it operated under the same conditions as the others. This anomaly was not known in advance and

Case A: healthy bearing

For healthy bearings (Figure 8), the RFID-based temperature measurement closely matched the IR reference. It accurately captured both the absolute temperature and the rapid transients caused by load changes. In contrast, the PT1000 signal remained smooth and strongly filtered by heat conduction through the surrounding structure.

Interpretation: The RFID sensor measures the thermal state of the rotating element; the PT1000 measures only the structural thermal field. 

Case B: defective bearing

For the bearing that was later found to be defective (see Figure 9), the RFID-based measurement revealed immediate thermal deviations. These deviations appeared consistently across the same driving-cycle profile. The PT1000 trace, by contrast, remained largely unchanged compared to the healthy bearings and showed no diagnostic sign of the anomaly. The IR sensor confirmed the hotspot, but it required partial disassembly, which made it unsuitable for use in a closed motor or in-vehicle setup.

Interpretation: Only the RFID-based measurement provided the spatial selectivity and temporal responsiveness required to detect the early thermal signature of a compromised rotating bearing. 

Conclusions - enabling previously impossible measurements 

Our work shows that a passive in-motor telemetry layer can provide direct, synchronized access to bearing-level temperature inside a sealed electric machine. This measurement point was previously out of reach for conventional sensors.

With this direct access, engineers can detect early thermal deviations, improve electro-mechanical correlation, and validate electric vehicles (EVS) powertrain behavior under realistic operating conditions.

The approach is inherently extensible. The same architecture can evolve toward higher-temperature operation, enabling bearing-proximal measurements under extreme thermal loads. In parallel, the robustness demonstrated on a dynamometer test bench provides a credible foundation for on-vehicle deployment and for continuous bearing-level monitoring under real-world driving conditions. 

Beyond the specific MGU-K application developed with the F1 team, these results point to a broader trend in modern electric powertrains and industrial machinery. As power density increases, the main challenge is no longer sensor accuracy, but access to the measurement point itself.

Integrating passive RAIN telemetry into Dewesoft’s open Test and Measurement ecosystem offers a practical and scalable way to measure sealed, rotating, or otherwise inaccessible components that conventional technologies cannot easily reach.