01 / BUS
Bus current is not phase current
The tool calculates average input current as electrical bus power divided by nominal voltage. Use motor and torque-speed data to size phase current and inverter devices.
Estimate 24V bus current, then review GaN reference data and design limits for your OEM servo board.
Published · Updated
What the public references establish—and what they do not.
01 / BUS
The tool calculates average input current as electrical bus power divided by nominal voltage. Use motor and torque-speed data to size phase current and inverter devices.
02 / FREQUENCY
TI reports TIDA-00909 testing up to 100kHz and peak efficiency up to 98.5% at 100kHz. That is a result for one 48V reference design, not a general servo-board rating.
03 / INTEGRATION
The EPC9146 shows a compact 81 × 75mm, 400W evaluation board, but EPC marks it obsolete. Its size and peak efficiency do not predict a sealed actuator’s temperature or fit.
Source pages reviewed October 11, 2026. Published evaluation results are context, not a performance promise for a custom board.
| Source / context | Published evidence | Boundary |
|---|---|---|
| TI TIDA-00909 reference design 48V/10A high-frequency, three-phase GaN inverter. | TI specifies a 12–60V input range, 7A RMS / 10A peak output, testing up to 100kHz PWM, and peak efficiency up to 98.5% at 100kHz. | These figures describe the TI reference design and its test conditions. Peak efficiency is not a continuous-efficiency guarantee for another board. |
| EPC9146 product page 400W three-phase motor-drive demonstration board. | EPC lists an 81 × 75mm board and efficiency above 98.4% at 400W output. | EPC marks the board obsolete and recommends another reference for new designs. Its dimensions and efficiency do not establish fit or performance for a 24V OEM board. |
| EPC9146 Quick Start Guide Thermal setup notes for bench evaluation. | The guide describes room-ambient bench operation with natural or forced convection and explains that a heatsink can increase device current capacity. | The evaluation setup is not a thermal rating for a sealed actuator, potted assembly, or different PCB stack-up. |
| EPC AN030 switching-loss note Method for estimating hard-switching power loss. | The note separates switching-loss terms and explains that loss depends on device and operating conditions; it also discusses reverse-recovery loss in silicon MOSFETs versus GaN FETs. | A device-level calculation still needs the actual device, bus voltage, current, switching frequency, gate timing, and thermal path. |
Resolve these with the target motor, layout, enclosure, and production supply chain.
Reference-board current and efficiency figures cannot be transferred directly to a confined joint. Copper area, PCB stack-up, ambient temperature, duty cycle, airflow, and the mechanical heat path all change junction temperature.
Raising PWM frequency changes switching loss. Fast switching edges can also couple into current-sense and encoder paths; the result depends on the power loop, layout, cabling, and interface.
The EPC9146 is marked obsolete by EPC. A reference board is useful for understanding an architecture, but it is not evidence that its parts remain suitable or available for production.
A bare-PCB outline omits the real integration envelope: connectors, keep-outs, heat spreader, cable bend radius, assembly access, and enclosure interfaces. GaN device cost alone does not determine system cost.
If you have worst-case switch loss and board-matched thermal data, use the 24V GaN servo control-board thermal estimator to screen a first-order junction-temperature estimate.
Share the motor data, phase-current targets, bus range, cooling path, and board outline. The engineering team can review the OEM requirements; this estimate alone does not qualify a design.
Practical questions for sizing and qualifying an OEM design.
No. It divides electrical power drawn from the nominal 24V DC bus by 24V to estimate average bus current. Phase RMS and peak current require motor and operating-point data such as torque, speed, winding characteristics, and control strategy.
No. A 12–60V input range shows the reference can operate across that bus range under its own test conditions. It does not validate the current, thermal path, EMI, firmware, or mechanical fit of a different board.
No. GaN can reduce some switching-loss terms, but conduction loss, switching frequency, layout, passives, cooling, and operating point still set board temperature and size. Compare complete designs at the same requirements.
Include bus range, continuous and peak phase current, torque-speed envelope, duty cycle, PWM target, motor electrical data, feedback interface, enclosure and ambient limits, outline constraints, quantity, and validation requirements.
EPC currently marks EPC9146 obsolete and recommends another reference for new designs. Treat its published dimensions and performance as historical evaluation data, then confirm current parts and availability before selecting a production architecture.
Share the DC bus range, continuous and peak phase current, motor speed and torque, feedback interface, PWM target, cooling path, and mechanical envelope so the engineering review can assess the real constraints.