Beat the Heat: Design and control of a Tunable Magnet Actuator for Reduced Thermal Dissipation in Microchips

October 1, 2026

Introduction

Heightened by investment in AI, the ever-increasing demand for more complex and efficient chips is pushing existing manufacturing technologies to their physical limits. In particular, the design of electromagnetic actuators, which are integral to most high-precision motion stages, faces an impasse: actuation force and precision cannot both be increased.

Higher forces require higher power, while the required precision demands low, stable, and uniform temperatures – a combination that results in impractically large actuators and cooling solutions. The problem is most acute when a force must be held for long periods, for example, to carry a payload against gravity or to maintain an alignment offset: a conventional actuator then dissipates heat continuously, even though it does no mechanical work. To overcome this limitation, the industry needs a new kind of actuator that produces high non-contact forces with lower power dissipation.

A schematic of the magnetic flux paths of a tunable magnet actuator (TMA); on the left is a sustained magnetised TMA; on the right is a current i_RA in a coil; the two flux paths contribute to a force F on the central mover

Tunable Magnet Actuator

The Tunable Magnet Actuator (TMA) eliminates the power dissipation required to maintain sustained force offsets. While nonlinearity is traditionally avoided in the name of precision, the TMA deliberately incorporates it, leveraging magnetic hysteresis to maintain a force in the absence of continuous input. This force is ‘tuned’ by short pulses that alter the remanent magnetisation of a ‘tunable’ magnet (TM) in the core of the actuator. In effect, the magnet has a memory: the state it settles in after a pulse depends not only on the pulse, but on the history of all previous pulses.

However, the introduction of additional nonlinearity complicates modelling. Removing the sustained coil current, and with it the Joule heating, comes at the cost of magnetic energy losses each time the magnet is re-tuned, due to its high hysteresis. For dynamic applications, we thus integrate a second coil that drives magnetic flux through a separate path with lower reluctance and lower hysteresis, producing a force proportional to its current. By optimally using either path – changing the remanent force with a current pulse, or applying a continuous current-dependent force – we aim to maximise the overall thermal efficiency of the actuator.

Neither mode is always the better choice. A pulse costs a fixed amount of energy up front, whereas the coil keeps dissipating energy for as long as the force is held. In the current prototype, for example, a force change that corresponds to 100 mT of remanent magnetisation is already cheaper to provide with the TM if it must be held for longer than about 0.1 s, whereas a 30 mT change only pays off after about 1 s.

measurement data of the tunable magnetic actuator

Measurement data from the control of the remanent magnetisation state of the TMA against the induced field H and over time.

Problem Statement

For a particular application with a set of design restrictions, we would like to understand how the actuator should be operated, and ultimately designed, to minimise its total energy losses. More specifically:

  1. Modelling. Can we develop a time-dependent electromagnetic model of the actuator that predicts the total energy losses when voltages are applied to either coil (the one magnetising the TM, or the one bypassing it). What is the simplest reduced-order model that is still accurate enough to be used for optimisation and control?
  2. Remanent-state control. Design a controller for efficient set point control of the remanent magnetisation of the TM. Which voltage pulse reaches a desired remanent state with minimal energy loss, and how can this be done reliably when the state of the magnet is only measured at prior remanent states?
  3. Choosing between the two modes. Given a force reference over time, when should a change in force be made by re-tuning the TM, and when should it be supplied by the second coil? What is the energy-optimal strategy when the reference is known in advance, and what is a good strategy when it is not, so that decisions must be made online?
  4. Design (extension). How do the dimensions and composition of the steel cores, magnets and coils affect the answers above, and which design minimises the losses within the given restrictions?

By the end of the week, we hope for a validated model of the energy losses, a strategy for tuning the remanent state and for dividing the force between the two modes, and, time permitting, recommendations for the design of a reluctance actuator with a tunable magnet.

The problem offers an opportunity to look beyond the traditional design principles that underpin modern magnetic short-stroke motion stages, and to develop a toolkit for improving energy efficiency through designed nonlinearity. Beyond magnetic systems, participants are introduced to an emerging group of systems that leverage hysteresis to exhibit tunable remanent states.