What is inside a roll servo?

[kevinh]

Hi ya'll,

I've worked a fair amount on drone software & hardware and I'm kinda interested in experimenting with rolling my own roll servo (carefully/on-a-bench, with even more peer review before putting it in a plane).

I'm aware of the shear pin and arm which is added outside of the trutrak, dynon etc... servo. But I'm curious about the innards. Have any of ya'll ever looked inside? Is it a stock stepper motor (ideally - can you read the part number/vendor?) or do they actually include custom gearing inside? I'm less interested in their stepper motor controller (I'd be plopping in a CANBUS based stepper controller).

Kevin

[rbibb]

Basically reduction gearing to increase torque, stepper drive circuitry.

[WingsOnWheels]

Don't most have some sort of torque feedback? Or maybe that is determined by rate vs current draw. Most of the EFIS systems will tell you when to trim the plane, so there must be something.

[kevinh]

Originally Posted by WingsOnWheels

Don't most have some sort of torque feedback? Or maybe that is determined by rate vs current draw. Most of the EFIS systems will tell you when to trim the plane, so there must be something.

yep - most stepper controllers can measure current draw, so that's how that would be measured.

[JonJay]

Kevin - the old Blue Mountain products used an electro magnet of some type to engage and disengage the stepper motor from the servo arm. I never tore one apart to see how they rigged everything up but when disengaged, you could feel absolutely no drag on the controls. Like the rest of their stuff, cool design, when it worked.....

[Rainier Lamers]

Originally Posted by kevinh

yep - most stepper controllers can measure current draw, so that's how that would be measured.

Nope.
Current draw means nothing in this application.
You can do this with DC or AC motors but only while they are running.

Rainier
CEO MGL Avionics

[BobTurner]

The Trio servos use a clutch, so when off, the gears are disengaged. When on, a modest force on the stick will slip the clutch. If the clutch refuses to let go or slip, a nylon gear will strip (with sufficient force applied) to release the autopilot.

I am not sure, but I believe internally the servo is mounted on shock mounts which allow it to rock back and forth just a bit. This motion is measured, and used to drive the trim commands.

[Rainier Lamers]

Originally Posted by BobTurner

The Trio servos use a clutch, so when off, the gears are disengaged. When on, a modest force on the stick will slip the clutch. If the clutch refuses to let go or slip, a nylon gear will strip (with sufficient force applied) to release the autopilot.

I am not sure, but I believe internally the servo is mounted on shock mounts which allow it to rock back and forth just a bit. This motion is measured, and used to drive the trim commands.

Nope.

Rainier
CEO MGL Avionics

[Rainier Lamers]

OK, I'll answer.

Method 1:
Mount servo motor on a platform allowing slight movement - spring loaded. When it moves it closes a microswitch. Usually two fitted - one for each direction. Used with some older servos. Works pretty well.

Method 2:
A strain gauge is fitted to a strategic part (often part of the output gear). This can measure force quite accurately but is expensive. Works very well.

Method 3:
Does not use any special hardware at all. You integrate servo position over time. Any change relative to a first position is assumed to require trim. This is helped with a differentiator to get a good solution relatively quickly after engaging but engaging should happen with an aircraft that is trimmed.
Note: This does not work well with every aircraft but is used often. Can go pear shaped if you fly a lot in turbulence and averaged position starts to get meaningless.

Method 4:
Vibrate the output shaft a tiny amount by modulating the stepper motor current a bit. Measure the amplitude and shape of this vibration to derive force and direction. Requires accurate and fast contactless position sensing at high resolution.
That's what we do...

Method 5:
--- Place your idea here ---

Rainier
CEO MGL Avionics

[BobTurner]

What did you mean, "nope"? I meant method #1! Maybe poorly described.