A motor winding can carry substantial current while a supply or low side shunt reports little or none. During recirculation, the current may remain inside a bridge path that bypasses the measurement resistor. Choose sample timing from the actual conduction states rather than assuming every ADC reading represents winding current.
Draw current paths for each state
Identify drive, brake, coast, and relevant PWM states for the selected bridge. Trace the winding current through switches, diodes, supply, and shunt. The TI brief on motor current sensing locations explains how some sensing positions cannot observe current during particular recirculation states.
Separate phase current, supply current, and measured shunt current in the diagram. They may coincide during one interval and differ during another. Label the sign convention so a valid reverse current is not treated as an impossible value.
Mark usable acquisition intervals
Overlay the PWM timing with amplifier settling, blanking, and ADC acquisition. Identify an interval where the shunt carries the desired current and the analog output has settled. Account for the ADC’s acquisition duration, not only the trigger instant.
At extreme duty cycles, that interval can become short or disappear. Calculate the available time across the intended duty range and compare it with the measurement chain’s requirements. A sample position that works at 50 percent duty may not work near either endpoint.
Verify against an independent current view
Compare the sampled result with an appropriate winding current probe or other characterized method. Capture bridge outputs and ADC trigger timing in the same test when possible. This helps distinguish a missing conduction path from amplifier recovery or digital timing errors.
Repeat while current direction changes and during the relevant braking mode. Do not infer a zero winding current solely from a zero shunt reading when the path diagram shows that the shunt is bypassed.
Define behavior outside the valid window
Choose a supported strategy for invalid intervals, such as a different sensing location, adjusted modulation, or a documented reconstruction method. Mark unavailable or estimated samples appropriately in the control algorithm. Silently reusing an old value can conceal loss of observability.
Save the state diagrams, valid timing windows, analog settling evidence, and comparison measurements. The resulting interface should make clear when a reading is direct, when it is reconstructed, and when no trustworthy value is available. That distinction supports both stable current control and meaningful fault detection throughout the intended PWM operating range.
Move from reading to a build-ready project
Use the fabrication and assembly checklists to prepare drawings, production files and test requirements.
Open the DFM checklist