Axial flux motor DC bus voltage selection

For each required duty point, record the lowest voltage at the inverter DC terminals after source sag and distribution losses. Separately record the largest voltage expected during charging, braking or a source transition. Include state of charge, source temperature, ageing assumptions and measurement location in the project record.

The table is an original budget template. No cell chemistry or series count is implied by “48 V” or “96 V.” Values remain open until the source and inverter owners agree the actual limits. A maximum component rating is not a target operating setpoint.

For a motoring steady-state screen, IDC ≈ Pshaft / (VDC × ηmotor × ηinverter). This assumes the stated efficiencies at that operating point, excludes auxiliaries and uses average DC quantities. A complete source budget must add auxiliary loads and account for actual loss maps, source ripple and transient behaviour.

Hypothetical comparison: delivering 10 kW at the shaft with an assumed combined motor-and-inverter efficiency of 0.90 needs approximately 231.5 A at 48 V or 115.7 A at 96 V. At loaded values of 40 V and 80 V, the same assumptions give 277.8 A and 138.9 A. These are arithmetic scenarios, not ratings for our low-voltage programme.

DC input current follows the input power balance. Motor phase current follows torque production, winding constants, speed and control. A battery current calculation therefore cannot be copied into a phase RMS current field. Near zero speed, high holding torque can coexist with low mechanical power and significant winding heating.

EMRAX documents winding-dependent torque constants in Nm per phase RMS ampere. That illustrates why a constant must carry its convention. For a balanced sine wave, phase peak equals sqrt(2) times phase RMS; this waveform identity is not a DC-to-phase conversion. Delta versus star and controller reporting conventions need explicit agreement.

Texas Instruments BPRA073, section 5.3, explains the ideal space-vector modulation voltage boundary. For a balanced sinusoidal fundamental from an ideal two-level inverter in linear SVPWM, the maximum phase-to-neutral peak is VDC / sqrt(3). Converting conventions gives line-to-line RMS VLL ≈ VDC / sqrt(2). These are fundamental values, not PWM switch-node peaks.

At a hypothetical 40 V loaded bus, the ideal ceiling is about 28.3 V line-to-line RMS. Reserving an illustrative 10% of that voltage leaves about 25.5 V for a preliminary comparison. Ten percent is our example assumption, not a recommended universal margin; real dead time, switching drops, current regulation and modulation strategy must be reviewed with the drive supplier.

Compare that available voltage with the required terminal voltage at load, not only open-circuit back EMF. Resistance and inductive terms are vectors and cannot generally be added as scalar magnitudes. Request the winding parameters, temperature, electrical frequency and motor-drive operating calculation. Field weakening is a separate validated operating mode with current, thermal and magnet limits.

Kollmorgen explains that deceleration can return energy to a drive bus and cause overvoltage if the system cannot absorb it. A charged battery, a source that cannot sink current or a protection disconnect can change the available energy path. Confirm the source and drive behaviour together rather than selecting the inverter only from nominal voltage.

For the procurement record, identify normal operating ceiling, warning and trip settings, component limits and credible transients separately. A braking resistor or other energy path must be compatible with the actual drive and application; this guide does not size one. Regenerative braking also does not establish a complete vehicle or machine stopping function.

For unchanged transmitted DC power and the same conductor resistance, doubling bus voltage halves average DC current and quarters the corresponding DC I²R loss. This arithmetic applies to that distribution path. It does not prove that a rewound motor has one-quarter of the winding loss or that its continuous output doubles.

Two hypothetical paths need different evidence. An existing 48 V machine with fixed protection and cable interfaces should first resolve its loaded voltage and current bottleneck. A new 96 V architecture can be compared at system level, including source, inverter, insulation, connectors, protection and service requirements. Neither is automatically cheaper or suitable for every application.

Keep unknowns explicit in the worksheet. The motor supplier needs the actual duty points and motor-side current convention; the drive supplier needs the winding and feedback interface; the source owner needs motoring and regenerative limits. Assign someone to reconcile the complete record.

A feasibility reply should identify which point is constrained by bus voltage, current, speed or thermal capability and which inputs still need measurement. A generic “48 V compatible” reply is insufficient for releasing a system interface.

Reviewed on 21 October 2026. The primary documents support winding/current conventions, ideal modulation and bus-energy behaviour. All numerical bus and current examples are our hypothetical arithmetic. They are not battery designs, motor ratings, tested efficiency or electrical safety approvals.

This is a cross-platform interface guide, not another 48 V or 96 V product page. The next gate requires actual source limits, drive documentation, winding data and an agreed test of the demanded operating points. The examples exclude auxiliaries and dynamic source behaviour unless explicitly stated.