Regarding positive feedback oscillations

by | |

Q. The aileron (rudder or elevator) of my model is buzzing like crazy. It's going back and forth so fast it's nearly a blur. What's wrong?

Insert BEFORE video - EF elevator

- Extreme Flight elevator experiences positive feedback oscillations

A. Yup, seen it before. Happens to any servo, any brand. Usually it's because of the control surface being so large it flexes (you only think it's rigid, it's not, not really).

Fortunately, there's an easy fix . . . use a bit of Velcro (the fuzzy side) like in this image. No, you don't have to get fancy and color coordinate it like in this example, but it does look nicer!

- A bit of fuzzy Velcro provides a slight bit of drag to damp oscillations.

Anyway, what you're experiencing isn't to do with the servo but with the control surface not being as rigid as you might think. Basically, everything is a spring, even an anvil. What's more, I'll bet a milkshake if you disconnect the linkage, the servo doesn't do it any longer. Point being, a little bit of drag at neutral keeps it from ever really getting going.

And to repeat, it happens mostly with large flexible control surfaces of a type usually seen on 3D models, but can also happen with scale models, too.

Insert BEFORE video - Boeing rudder

- Giant scale Boeing PT-17 rudder experiences positive feedback oscillations

With the 3D models, even small models as small as 48" but ranging up to and including 35% and 42% models because the elevator control surface, for example, may be as much as 60% of the entire horizontal surface area! And not only is it huge but it's definitely not as stiff as you may suppose.

Point being, while it feels like it's rigid, it's not really in the grand scheme of things. Meaning, when it flexes and returns to neutral (springs back to shape, actually) this mechanical motion bumps the pot (potentiometer), which nudges it out of position and the servo electronics sense this as unwanted movement and thus, command a return to center. And this command is issued quickly (with our servos on the order of 350 times per second (350Hz) but even slower servos with refresh cycles of 125 or 250Hz will do this.

Anyway, this all happens so fast the control surface flexes again, and when it springs back into shape, this once again bumps the pot again, which starts the cycle allover again. Rinse, repeat! And note, this happens with any brand of servos and it's not a defect. It's mechanical feedback caused by the springy-ness of the control surface feeding back into the potentiometer.

In short, what you're seeing is referred to as a positive feedback oscillation. By the way, positive feedback takes a system away from the state of stability, while negative feedback gives a system more stability. What's more, we see this electronically too. An example you may be more familiar with are oscillations induced by gain, e.g. with gyroscopes. The positive feedback occurs when running the gain too high, and thus, lowering the gain stops it. In the case of a mechanical oscillation, adding a little bit of drag using the fuzzy side of a piece of Velcro is the mechanical equivalent of lowering the gain. It adds 'just' enough drag to preclude the oscillation from ever getting going.

Insert AFTER video - EF elevator

- Fuzzy Velcro stops Extreme Flight elevator positive feedback oscillations


Anyway, and quite honestly, it's our view we're quite fortunate this little bit of damping works because it removes the stress of not know what's going on as well as that caused by the incessant nattering of the surface. At heart, what the fuzzy side of a piece of Velcro provides is a slight bit of drag, a mechanical change in feedback, and this makes it stop. You could 'lightly' touch the control surface with your fingertip and achieve the same thing, but Velcro is easier.

Note; this doesn't happen in the air because the forces of air flowing over the control surfaces keep it from ever beginning. And by the way, this is strictly a static event. Point being, it's nothing to fret overly much about because it's entirely due to the size of the control surfaces. This was unheard of in the days when a large elevator was 1-1/2" wide while today, 8-12 inches wide is not all that unusual. But we know it bothers you (it bothers us as well when it happens to one of our models). Thus, we also resort to adding a bit of feedback via the popular Velcro dot trick.

Try it, it works! There's nothing wrong with the servo. One last tip. The drag doesn't need to be added at the control surface. For example, a model using pushrods to control the surface are sorted with a bit of drag within the fuselage at the pushrod, too. Easy peasy!

Insert AFTER video - Boeing rudder linkage

- Giant scale rudder no longer experiences positive feedback oscillations

--
John

This entry was posted in no categories.