Solving the Servo Jitters
CRRC Home ] Up ] Join CRRC ] Calendar ] Getting Started ] Forecast ] Search ] Contact Us ]

 

What's New
About CRRC
Articles & Tips Index
CRRC EMail List
Flying Sites
Instructor Program
Links
Newsletter
Pictures
Weather

Home > Articles & Tips Index > Flying > Solving the Servo Jitters

[Courtesy of Tom Hoopes 1999, oakley "at" xmission.com - January 1999]

Do you get weak in the knees when you are about to launch your dream creation and see one or more of the surfaces glitching wildly? Have you had to resort to having someone else launch your plane because the glitching seems worse when you are next to the plane? With the advent of micro servos, it has become increasing more popular to embed servos in the wing or tail to eliminate long or complicated mechanical linkages. High-performance designs demand such configurations which may use up to four servos in the wing alone.

Electrical connections from the receiver to a wing- or boom-mounted servo can conveniently replace equivalent mechanical linkages but a new set of problems may appear in the form of electrical noise. To understand this phenomenon, you must first know a little theory on servo operation. Modern servos require three wires for operation which are: power (+4.8 volts), ground, and signal. Although the power and ground leads are self-explanatory, the signal lead may not be as evident.

All servos whether connected to AM, FM, or PCM receivers are positioned by issuing a positive pulse from the receiver to the servo whose width determines the servo arm/wheel position. Typically, a pulse width of 1.5 milliseconds (that's 1 1/2 thousandth of a second) will center the servo, a pulse width of 0.8 milliseconds will rotate the servo to one extreme and a pulse of 2.2 milliseconds will rotate the servo to the other extreme. The resolution or granularity of servo movement can be dependent on the make and type of servo. Servos also have a characteristic called "dead band" which is the amount the width of the pulse may change before the servo will actually try to move to the new position.

It is desirable to have a certain amount of dead band or else the servo will always be attempting to align itself with very minute changes in the pulse width which can result in buzzing, chattering, and higher than normal current consumption.

When servos are located in the wing or tail boom, they must be located further away from the receiver which generally means 12" to 36" extensions. These extensions seem innocent and simple enough that little thought is ever devoted to them, but these extensions can be the source of servo jitters and potential loss of control.

Extension wires can cause two problems:

  • First the wires must be of sufficient gauge to carry adequate current to the servos, especially nearing a stalled condition
  • Second, the an extension may act as a crude antenna that can pickup RF (the Radio Frequency signal that is emanating from your antenna or the pilot's transmitter that may be next to you) and feed it into your servos. Both of these situations can cause servo glitches and jitters.

Now that we have identified the problem, how do we solve it? Let's attack the first problem.

How much current will a servo draw when stalled? Well, that depends again on the servo, but it can exceed 100 milliamps. I would recommend using 26-gauge wire as a minimum with 24 gauge preferred. For reasons that are too lengthy to explain in this article, another preferable quality to have in your extension cable is a continuous twist throughout the cable length. This can be done quite easily.

  • Cut three wires approximately 1 1/2 times longer than what you require.
  • Hold the ends of the three wires together and wrap a piece of tape securely around all three wires.
  • Anchor the taped end of the three wires to a door knob or chair leg and insert the other three ends into the chuck of a variable-speed drill.
  • Carefully run the drill until a tight twist is formed in the cable. It is normal for the cable to untwist slightly when it is released. Heating the wires slightly with a heat gun will help set the twist.
  • As a reminder, use colors that make sense (i.e., red = +4.8, white = signal, black = ground). This should help avoid a costly wiring mistake.

The solution to the second problem may not be quite as clear. Logically, we need some way of eliminating or filtering out the high-frequency RF noise from the signal line but still allowing the positioning pulse to pass without being altered or else our servo positioning will be destroyed. We can do this by using a passive component - a capacitor. In this application, the capacitor will appear as a direct short or "shunt" to the high-frequency noise but look like an open circuit to the low frequency (remember our 1.5 millisecond pulse?).

Not any capacitor will do! It must be of the proper value. A little math and collecting of empirical data (testing different values) indicates that a small monolithic or ceramic disk capacitor of 150 pf to .001 uf (pf=picofarad, uf=microfarad) will do the trick. Solder the capacitor as close to the servo as possible and connect it from the signal lead to the ground lead. Remember to use heat-shrink tubing to avoid short circuits.

Feeding the servo leads through a ferrite bead causes the servo lead to act as a small RF choke at high (RF) frequencies. In other words, this is another way to attenuate the superimposed RF noise as done with a capacitor above. I believe Critter Bits sells a set of "clamp on" ferrites. Other sources include Amidon Associates, Digi-Key, J.W. Miller, etc.

The capacitors are generally cheaper and easier to acquire. At your closest Radio Shack, look for P/N 272-125 (470pf) or P/N 272-126 (.001uf). They'll run about $0.25 each. Remember, solder the capacitor as close as possible to the servo and place across the signal and ground lead.

I have never had a case of the jitters that I couldn't solve by following the guidelines that I have presented. Good luck!

Tom Hoopes, oakley "at" xmission.com


CRRC Home ] Up ] Join CRRC ] Calendar ] Getting Started ] Forecast ] Search ] Contact Us ]

Disclaimer