A servo moving on the bench is an encouraging start for a small mechanism. Once a linkage, an enclosure or a moving load is attached, the important questions change. Can the supply support it? Does the motion stay within what the mechanism allows? Does the commanded position match the movement you actually need?
These are familiar early problems. An Arduino forum thread about a board losing power when a servo was connected turned toward wiring and supply connections, and Arduino's servo troubleshooting guidance separates signal configuration from adequate servo power. The thread is one case, not proof that every servo problem has the same cause, but the split between signal and power is the right place to start.
On our bench we drove two small hobby servos from a Raspberry Pi Pico on a Graft carrier board, through an I2C servo driver module, with the servos on their own supply. The program was built in BIPES blocks, which generate MicroPython. This note covers what that setup showed and what it leaves for the next test.
Follow the command path
The controller talks to a PCA9685 driver over I2C0, with SDA on GPIO4 and SCL on GPIO5. We installed the supporting library and configured the servo interface for 50 Hz with a pulse range of 500 to 2500 microseconds.
Those pulse endpoints are settings in this example, not universal mechanical limits for every servo. The driver converts a requested angle into a pulse width using its configured mapping, so "90 degrees" means "the pulse the mapping assigns to 90", not a measured shaft angle. A different servo, or the same servo inside a linkage, needs its own limits established before it is asked to travel to an endpoint. Driving a linkage into a hard stop at the end of the pulse range loads the gearbox and the supply at the same time.
The test program commands channel 0 to 45 degrees and channel 1 to 90 degrees, waits two seconds, then commands both channels back to zero. The commands go out one after the other. Both servos moved, but we did not record shaft-angle measurements, synchronized trajectories, torque or repeatability under load. Sequential commands with a pause between them also say nothing about coordinated motion, where two axes need to arrive together.
Keep power and movement in the same test
In our setup the servos had their own supply, separate from the controller's logic power. A signal connection tells the servo where to go; it does not provide the energy for the motor. The servo's current draw rises when it starts, reverses or stalls against a load, and that is when a shared or undersized supply sags and resets the controller.
So supply behavior and the signal reference have to be checked in the complete circuit: the servo supply, the controller supply and a common ground between them where the driver's design calls for one. An unloaded servo that moved once on the bench does not establish any of that.
For a mechanism, the next test we would run is:
- set a restricted travel range that reaches the needed positions without forcing the linkage into a stop;
- watch the servo supply voltage during starts, reversals and holding, with the real load attached;
- record the positions actually reached, against a fixed reference mark, over repeated moves from both directions.
A video of the bench run is a demonstration, not a substitute for those measurements on the intended assembly.
Decide what "positioned" means for the product
A pointer moving to a rough visual location has very different requirements from a fixture that must return to the same place within a tight tolerance. Define the acceptable position error and the load before choosing a control approach. A hobby servo's internal feedback reports nothing back to your controller, so the firmware only knows what it commanded, not where the output went. If your product needs to know the real position, plan for an external reference or a different actuator.
If the requirement is simply on-off movement, such as opening a flap or running a fan, a switched motor or solenoid may be the more direct choice.
Write the answers down before the enclosure and the mounting points are fixed: travel range, load, required accuracy, how often it moves, and what should happen at power-up. Those numbers decide the servo size, the supply and the mechanical stops far more than a first bench movement does.