Why Does My Drivetrain Draw More Current Than Expected?

You sized the motor, checked the datasheet current, and wired a supply that looked comfortable. On the bench the current clamp says otherwise. The rail browns out, the driver folds back, or the fuse teaches you physics the hard way.

High current is a symptom. The work is finding which assumption broke. This post is a practical triage for drivetrains that draw more than you expected.

Remember what current is answering

For a DC or brushless motor driven from a voltage rail, current tracks how hard the motor is working against load and how the electronics try to hold speed or torque. In rough terms, torque lines up with current (via Kt, with losses), and electrical power is voltage times current. If something increases required torque, reduces effective voltage, or worsens efficiency, current rises.

So "more current than expected" usually means one of:

1. The mechanical load is higher than you modeled
2. The electrical side delivers less useful voltage / more loss than you assumed
3. The controller is commanding something greedier than the steady-state calcs
4. The measurement is not what you think it is

Work through those in order before you blame the motor SKU.

Mechanical causes (the usual culprits)

Friction and binding. Seals, bushings, cable drag, misaligned rails, and hot grease all add torque. A dry run with the load disconnected (or the mechanism free) is the fastest split: if unloaded current is already high, look mechanical before you redesign the power stage.

Wrong reflected inertia or accel profile. Short, aggressive moves need large acceleration torque. Firmware that ramps hard will spike current even if steady-state cruise looked fine on a spreadsheet.

Backdrive, holding, and stalls. End stops, jammed leadscrews, and holding against gravity without a brake turn the motor into a heater. Stall current is not a rare edge case in real mechanisms; it is often the design load.

Gearbox and transmission loss. Datasheet motor current ignores a 70% efficient planetary at cold start. Multiply load torque by 1/efficiency before you compare to Kt. Belt slip and poor mesh waste power as heat and extra current.

Electrical causes that look like "the motor is greedy"

Supply sag. Cable resistance, connector drops, and weak converters lower the voltage at the driver. The controller then pushes more duty (or the motor simply slows and you compensate with higher command) to hold the same speed. Measure voltage at the driver input under load, not only at the PSU terminals.

Driver and winding losses. FET RDS(on), sense resistors, long motor leads, and hot copper all eat volts. Current rises to deliver the same mechanical power when effective voltage shrinks.

Wrong assumption about continuous vs peak. The "expected" number was often no-load current or a continuous rating, while the test was a peak-heavy duty cycle. Align the stopwatch with the datasheet mode you quoted.

PWM and measurement bandwidth. A cheap meter on a chopped motor line can under/over read depending on filter. Compare RMS-aware measurement or driver telemetry with a known shunt if the numbers feel magical.

Control loop causes

Closed loops hide mechanical problems as electrical ones.

• A position or speed loop with aggressive gains chatters around a setpoint and wastes current.
• Integral windup into a stop commands full effort against a hard wall.
• Current limits that are set too high (or never hit) let the mechanism demand whatever the supply can give.
• Open-loop duty that "worked on the bench unloaded" fights real friction once assembled.

If you can run the same motion open-loop at a fixed duty and current drops into a sensible band, the plant or the tuning is the story, not a mysterious motor constant.

A triage sequence that saves evenings

1. Log voltage and current together at the driver during the failure move. Sag tells on the supply; high V and high I points at torque demand.
2. Decouple the load. Free shaft or disconnect the mechanism. Unloaded current near datasheet? Load path. Still high? Wiring, driver, motor damage, or bogus telemetry.
3. Compare stall / locked-rotor carefully. Brief locked current vs datasheet stall (same voltage). Far above datasheet suggests measurement error or a different motor than you think. Far below with motion stall often means the controller is current-limiting or the supply collapsed.
4. Check temperature. Rising continuous current as the unit warms often tracks copper resistance and thicker grease, not a new bug every hour.
5. Replay the duty in a model. Put motor, supply ESR, gear efficiency, and load torque in one graph. Sweep friction and voltage drop. If the model only matches the bench when friction is 2x your spreadsheet, you found the bug in the assumptions.

That last step is cheap insurance. A small system simulation will not replace a torque sensor, but it will show which parameter has to be wrong to explain the clamp reading. Hardwave is built for that kind of Python system graph: motor + rail + load in one place so you can change efficiency or cable resistance and re-run instead of guessing which BOM line to swap.

What "expected" should have included

Before the next design, bake these into the expectation so the surprise shrinks:

• Load torque with friction and efficiency, not ideal CAD force
• Voltage at the driver under load, not open-circuit PSU rating
• Duty cycle RMS, not a single friendly operating point
• Controller current limit and foldback behavior
• Stall and end-stop as first-class cases

Drivetrains that draw more current than expected are rarely cursed. They are usually honest. The mechanism asks for more torque than the model admitted, or the electrical path delivers less voltage than the spreadsheet claimed. Separate those stories with a few measurements and a model you trust, then change the part of the design that is actually lying.