Key Takeaways
- Magnet free construction buys SRMs heat tolerance, high speed capability and freedom from rare earth sourcing risk, and it costs more current for the same shaft torque.
- Torque ripple and acoustic noise are the deciding constraints, and both are set by pole geometry, firing angles and current shaping rather than by the controller alone.
- A measured flux linkage map is the model an SRM drive rests on, so the machine, the converter and the control logic have to be studied as one system.
Switched reluctance motors trade magnets and rotor windings for a plain steel rotor, and that single choice sets up every advantage and every headache they carry.
The stator holds concentrated coils, the rotor holds nothing but laminated iron, and torque appears as the rotor is pulled toward the position of least reluctance. That makes the machine cheap, heat tolerant and safe to spin fast. It also makes the magnetics deeply nonlinear, since torque depends on how flux linkage varies with current and rotor angle at once.
Traction is where the tradeoff gets argued hardest. Electric cars reached 25% of new car sales globally in 2025, and each reflects a decision about magnet supply, cost and thermal margin. Because the equations resist an analytical solution, simulation is the honest way to study torque production, current shaping and noise before committing to hardware. You’ll learn more from a well parameterized model in a week than from a first prototype in a quarter.
What a switched reluctance motor actually is
A switched reluctance motor makes torque by pulling a salient steel rotor toward alignment with an energized stator phase. There are no magnets, no rotor bars and no rotor windings. Each phase is energized in sequence according to rotor position, so the machine will not turn without position feedback and a purpose built converter.
A common 8/6 layout gives four stator phases and six rotor poles, so each phase conducts over a narrow window and returns its stored energy to the DC link. Phase current looks like a train of shaped pulses rather than a sine wave, so what you know about sinusoidal machines transfers only loosely.
The converter belongs to the machine rather than beside it. An asymmetric half bridge with two switches and two diodes per phase is the usual choice, since it applies positive, zero and negative voltage to each phase independently. Fault tolerance follows, because phases share no magnetic path and one open phase costs torque without stopping the shaft.
Why magnet free construction gives SRMs a durability edge
The advantage list starts with what the rotor doesn’t have. No rare earth magnets means no exposure to magnet pricing, no demagnetization risk as winding temperature climbs, and no uncontrolled back EMF if the converter faults. The rotor is a stack of laminations, so it tolerates high speed and hot ambient conditions well.
Mining hoists, steel mill auxiliaries and industrial fans have used SRM drives for that reason. A rotor with nothing to debond or lose flux survives dust, vibration and thermal cycling that shorten the life of a magnet machine.
Efficiency matters more than the parts list over a service life. Around 25% of electric motor system electricity use could be saved cost-effectively, so the control strategy behind a pump or a fan keeps paying back for years. Peak efficiency at the rated point often trails a comparable permanent magnet machine, so the win appears in duty cycles that spend most hours away from it.
“Noise is structural more than electrical, so damping it means reworking pole geometry, stator stiffness or the current pulse shape.”
Torque ripple and acoustic noise are the real constraints
Torque ripple and acoustic noise are the two reasons SRMs lost most traction applications to permanent magnet machines. Torque arrives in strokes as each phase takes over from the last, and radial force pulling the rotor toward a stator pole deforms the stator ring, which then rings at its own natural frequencies.
Commutation is where the ripple concentrates. An outgoing phase still carries current that produces falling torque while the incoming phase has yet to build flux, and the gap shows up as a dip in shaft torque once per stroke. Torque sharing functions reduce it by splitting the reference between the two active phases, though voltage sets the limit. At high speed the DC link will not push current into a rising inductance fast enough to follow the profile, so the ripple returns.
Noise is structural more than electrical, so damping it means reworking pole geometry, stator stiffness or the current pulse shape. Skewing, wider conduction angles and slower switching edges each help, and each costs you torque density or efficiency somewhere else.
The flux linkage characterization your control loop depends on
Every practical SRM control scheme rests on a flux linkage map, which records flux linkage as a function of phase current and rotor angle. The magnetic circuit saturates hard near alignment, so inductance is never a constant you can look up. Torque comes from the derivative of co-energy, which makes that map the model itself.
A usable characterization sweep has to capture a specific set of things.
- Flux linkage at current steps from zero to well past rated, so saturation is measured
- Rotor angle resolution fine enough to resolve the aligned and unaligned extremes
- Static torque at the same points, which independently checks the derivative you compute
- Winding resistance taken warm, since a cold value understates copper loss
- Mutual coupling between phases when conduction windows overlap, which single phase tests miss
Where the map comes from decides how far you can trust the answer. Finite element analysis gives you the full surface before a machine exists, at the cost of mesh and material assumptions. Bench measurement gives you the truth for one built machine, at the cost of a locked rotor rig and patient logging.
“Torque comes from the derivative of co-energy, which makes that map the model itself.”
SRM versus PMSM for electric vehicle traction drives
The main difference between an SRM and a PMSM is where the flux comes from. A PMSM carries excitation in its rotor magnets, so torque per amp is high and control is well behaved. An SRM builds all of its flux from stator current, which costs more current for the same torque and buys freedom from magnets and supply risk.
| Comparison point | What it means for an SRM traction drive |
| Rotor construction | Plain laminated steel survives high speed and heat with nothing to debond or demagnetize. |
| Torque per amp | More current is needed for the same shaft torque, so the converter and cabling grow. |
| Control complexity | Position feedback and a nonlinear flux map are required before the drive runs at all. |
| Acoustic behaviour | Radial forces excite stator resonances, so noise treatment starts at the geometry stage. |
| Supply exposure | Zero rare earth content removes a sourcing and pricing risk that follows magnet machines. |
Cost is where the argument usually lands, and it’s less one sided than a parts list suggests. Magnet free construction saves money in the machine and spends it again in the converter, the position sensor and the calibration work. A high volume program absorbs that calibration once, while a low volume industrial drive often can’t justify the same work.
Simulating the machine and its drive as one system
Model the machine, the converter and the controller in one place, because SRM behaviour comes out of their interaction. A flux linkage lookup feeds the machine model, an asymmetric half bridge applies the voltage, and the controller picks firing angles from rotor position. Pulling them apart hides the effects you care about.
Picture a hill start inside a traction study, where torque reference is high, speed is near zero, and current chopping does all the work. Push the same model to highway speed and control authority shifts to the firing angles, since there’s no longer room to chop within a stroke.
Solver settings decide how faithfully that transition renders. Phase current moves fast at commutation, so a fixed step comfortable for the mechanical dynamics will smear the switching edges and understate ripple. Running the study inside MATLAB and Simulink with SPS keeps the machine, the converter and the control logic in one model, with editable equations you can open when a result looks wrong. For research groups that model doubles as the record, since a colleague who reads your parameters and re-runs the sweep reproduces your figures.
What disciplined SRM modelling earns you before hardware
SRMs reward teams who do the magnetics work early and punish the ones who treat control as a tuning exercise at the end. The machine is simple to build and hard to drive well, so the value sits in a model that tells you what torque, ripple and noise a given geometry and firing schedule will actually produce.
Nobody walks away from an SRM because the rotor is wrong. They walk away because the calibration effort and the acoustic work didn’t fit the schedule, and finding that out before the first lamination order is worth more than any single efficiency figure. Working through the flux maps, the current profiles and the converter behaviour in SPS ahead of any metal turns those questions into results you can defend in a design review or a paper. The teams who get good SRM outcomes are rarely the ones with the best hardware. They are the ones who understood the magnetics first.
