You might have found the one that decided to be a part of the first section of the bathtub curve
For my thesis I tested SG-90 (the cheapest servos you can get) and if you don’t overload them (about 40% of the datasheet torque ). Most of them worked for a reasonable amount of time 50+ hours (order of magnitude or maybe two over expected lifetime of a plane made with such cheap components) but two failed within 5 hours of the test start.
PS the plan is to test some better servos but that would be impractical for writing thesis as they are expected to take longer to break.
Now try and hold the servo by the hand while the motor is running at full power, and try manually to tilt it rapidly! There you would get full gyroscopic effect and torque felt on the servo
I do want to make a device where I can safely move it by hand. Shouldn’t be too difficult but during my current measuring test, there was no significant change with the motor running.
You can make a simple mechanical tilt mechanism with a lever, i.e. a stick with a hole in the middle bolted to something solid, but which you can pivot, on one side would go the servo/motor, on the other you just move it by hand. Safe
I did take a cursory look at a few of the servos and mechanically they looked fine. I would suspect something related to motor. The biggest issue is that servos often stopped working temporarily.
I can only imagine the size of the brushes in an S90 motor. Probably the rapid reversal of a servo motor gives the brushes a hard time compared to a one-way motor.
Im surprised that its only hi-end servos that use brushless motors. I would have expected to see them in mid-range at least. My new savox is brushless as is the popular tilt servo Heewing T2. It must be the way ahead.
These have been great. Gdw designed them to be tilt servos. No play and very strong. They even have a metal motor mount. My vtol weight is almost 4kg with 6kg thrust per motor.
). Most of them worked for a reasonable amount of time 50+ hours (order of magnitude or maybe two over expected lifetime of a plane made with such cheap components) but two failed within 5 hours of the test start.