Guides / Design
Gimbal brackets: keeping two axes square when the part is mostly air
August 30, 2026
A gimbal bracket is a demanding part disguised as a small one. It carries a camera that has to point where the controller says it is pointing, through vibration, while weighing as little as possible.
Almost all of the difficulty is in relationships between features rather than in any single dimension.
What the part has to hold
Two bearing seats on one axis, coaxial. If the two seats are not concentric, the bearings fight each other. The axis binds, friction varies through rotation and the motor draws more current at some angles than others. On a brushless direct drive gimbal, that shows up as visible stepping in the footage.
Motor face square to the axis. The motor stator bolts to a face. If that face is not perpendicular to the bearing axis, the rotor runs eccentric and the axis wobbles.
Two axes square to each other. Roll to pitch, pitch to yaw. An error here is a cross coupling the controller has to correct continuously, and it never fully does.
Mass close to the axis. Every gram away from the rotation axis is inertia the motor has to accelerate. This fights the weight reduction instinct, because material near the axis is cheap in inertia terms and material at the ends is expensive.
Why one setup does the work
Tolerancing your way to squareness is expensive. Machining your way to it is not.
On a 5 axis machine the bracket is held once and the tool reaches both bearing seats and the motor face without releasing the part. Every feature is then referenced to the same origin. The squareness you get is the machine’s squareness, typically much better than anything achievable by flipping the part and relocating it on a fixture.
That is the whole argument for 5 axis on this part. It is not about exotic shapes, it is about relationships you would otherwise have to inspect and sort.
Sensible numbers to put on the drawing
| Feature | Typical call out |
|---|---|
| Bearing seat bore | H7, or a stated press interference |
| Coaxiality of the two seats | 0.02 mm |
| Motor face perpendicular to axis | 0.02 mm over the face |
| Axis to axis squareness | 0.05 degrees, stated as a geometric tolerance where possible |
| Bearing seat surface finish | Ra 1.6 or better |
| Everything else | general tolerance |
Give the datums properly. On a gimbal bracket, the primary datum is almost always the bearing axis, not a flat face on the outside of the part. If the drawing datums are the outside faces, the part will be inspected against features nobody cares about.
Stiffness beats weight here
A gimbal bracket with the material removed from the wrong places rings. The motor drives it, the structure flexes, the sensor reads the flex and the loop goes unstable at some frequency nobody predicted.
Practical rules:
- Keep the material in the direct path between the two bearing seats, and between a seat and its motor face.
- Take weight out of the arms and the ends, where it is also buying you inertia.
- Prefer a ribbed section over a thin plate of the same mass.
- Do not pocket both sides of the section that carries torque unless you have a reason.
The general weight rules in designing machined drone parts for weight apply, with the added rule that stiffness matters more here than on any other part on the aircraft.
Material and finish
6061-T6 for almost every gimbal bracket. It machines cleanly, it is stiff enough in a sensible section and it anodises evenly. 7075 shows up on brackets that also take landing loads or on very thin sections. Titanium is rare and usually unnecessary, because the part is stiffness limited rather than strength limited.
Finish is usually a type II anodise, black to keep reflections off the lens. Mask the bearing seats. A 10 micron coating in an H7 bore is enough to turn a transition fit into a press fit, and a press fit on a thin housing distorts the bore.
Assembly details worth deciding early
- Bearing retention. Shoulder, circlip groove or retaining compound. Each one changes the machining.
- Preload. A wave washer needs space. Design the pocket now rather than filing it later.
- Cable route. A slip ring or a cable path through a hollow shaft is a machining feature, not an afterthought. Add it before the first article.
- Balance adjustment. Slotted mounting for the camera plate gives the assembler somewhere to go. A fixed hole pattern assumes the camera never changes.
Before you send the RFQ
Send the model, the bearing part number, the motor part number and one line about the payload mass. That is enough for us to check the fits, ask about anything that looks like it will bind and price the part properly. Reply within 24 business hours, and 5 axis work is in house, so the squareness discussion is with the people cutting the part.
Frequently asked questions
How square do gimbal axes need to be?
Tighter than most people expect. An axis error of a tenth of a degree shows up as drift the controller has to fight continuously. Holding the two bearing seats in one setup gets you there without a tolerance chase, which is why 5 axis is the normal route.
What fit should a bearing seat have?
For a small ball bearing in aluminium, a light press or a close transition fit on the outer race, usually H7 on the housing bore. Loctite retaining compound is common on small gimbals and it does not remove the need to hold the bore.
Can a gimbal bracket be 3D printed instead?
For a first prototype, yes. For flight, printed brackets creep, they are far less stiff and the bearing seats do not hold their size. A machined bracket is the version that gives repeatable pointing.
Need this part quoted?
Send the model and the quantity. Price and lead time come back within 24 business hours.