[Headline: Unravelling the mystery behind flybarless gyros]
For this month’s column, I talked with Team Thunder Tiger and Team Futaba pilot, the king of low head speed and old-school 3-D, Gary Wright, about flybarless gyros. I want to thank Gary for taking the time to share his knowledge with MA.
Chris Mulcahy: What is a flybar?
Gary Wright: A flybar is a device whose purpose is to create unsolvable vibrations and make helicopter flying as difficult as possible. Seriously, a flybar is a type of mechanical rate gyro, whose purpose is to enhance stability and controllability of a helicopter.
It is a spinning bar that wants to fly in the same plane all the time, thus giving us a crude reference point. It generally connects to the blade grips in such a way that it imparts a pitch change in the blades in an attempt to restore stable flight due to its gyroscopic reference plane.
There are two common types of control for a model helicopter. I’ll simply refer to them as Bell and Hiller because we’ve [likely] all heard the term “Bell-Hiller mixing arm” on flybar-equipped helis, so the terms are familiar.
In a “Bell” control system, input from the swashplate is transferred directly to the blade (like our modern flybarless helicopters). In a “Hiller” system, control is transferred to the flybar and not directly to the blades. The flybar, in turn, controls the blade pitch.
Bell-type systems exhibit great initial responsiveness, but the control response tends to decay as the helicopter accelerates about an axis. This is because the initial angle of attack increases with a cyclic pitch input, but the angle of attack decays as the helicopter starts rotating about that axis. Hiller-type systems lag a bit in initial response, but they exhibit more consistency in rate … [in other words] they’re a bit slow to get started but are good at maintaining a given rate of rotation once accomplished.
Because of the input to the aerodynamic devices (commonly paddles on our flybars, but can be cup, ring, or other shapes), they tend to force the flybar (our mechanical gyro) into a new plane of rotation in relation to the blades, thus assisting in maintaining rate of rotation.
As you’ll see on most models with a flybar, there is a mixer arm that allows input both directly from the swashplate and from the flybar—“Bell-Hiller mixing arm.” This results in a somewhat crude approximation of a PID [proportional integral derivative] control loop.
CM: Now that we know what a flybar does, can you explain what a flybarless gyro does?
GW: A flybarless gyro does the same thing as the flybar, hence the terms virtual, or electronic, flybar. The reference, however, is an electronic sensor rather than the mechanical sensor we had with the flybar, plus integration of a more advanced control loop, commonly a PID loop (I don’t know of a flybarless gyro that doesn’t use PID, but there could of course be some that use other control algorithms).
The flybarless gyro gives us somewhat of a closed loop system rather than an open loop system, thus control and stability can be more precise. To expound a bit, an open loop system would be like an automobile cruise control, consisting only of a throttle lock. On level ground it would be somewhat acceptable. However, it would accelerate downhill and accelerate uphill because there is no reference from which to compare and the control loop is “open.”
The cruise controls we have today are closed loop systems. They have a reference (a sensor) that feeds back information about the speed that is being experienced and can adjust the throttle position to maintain that speed via some algorithm (a P/I-type system to maintain consistency of rate). Flybarless gyros give us this type of capability, but with far faster responsiveness than a mechanical flybar.
CM: There are plenty of gain settings to get a grip with. What exactly is PID, and how does it affect our helis?
GW: PID is an acronym for proportional integral derivative. It is an equation for a closed-loop control system that is elegant and can result in not only stability, but consistency in control rates, and tuning ability of acceleration/deceleration rates. In our control loops the gyro sensors sense a rate of rotation.
Compare it to the rate being commanded by the pilot, then compute an error if those rates are different, and command movement of the servos based on that error. If all they did was sense a motion and feed in an amount of correction, they would be “P” control loops—rate gyros—and would simply damp (no, not dampen—they’re not getting wet) unwanted movements.
With the PID algorithm they compare the experienced rate with the commanded rate, and compute the amount of correction; however, they are also comparing the rate to many iterations of sampling over time and integrating that information in the computations to attempt to maintain consistency, and they are also utilizing the derivative to predict the future needs, thus assisting in applying a more exact correction.
If you study the PID algorithm, (it won’t be difficult to find on the Internet), you’ll see how it utilizes the three gains: proportional, to know the amplitude of corrective action that’s needed; integral, to maintain consistency of rate over time; and derivative, to refine the stops.
Simply stated, if it’s not stable, your P gain may be wrong. If it’s not consistent, like a tail whipping when doing pirouettes in forward flight, your I gain is not right. If it bounces or rebounds on stops, or doesn’t stop precisely, your D gain is probably incorrect.
When the proportions of I and D are correct, you can just adjust the proportional gain (the “normal” gyro gain we’ve been familiar with for years), because it controls the overall amount of correction and although each one can interact with the others, just think of the proportional gain as the master.
CM: What is your typical procedure for setting up a new flybarless heli?
GW: I tend to like to use a system or procedure to adjust something rather than the “hunt-and-peck” method of experimentation. I’ve developed a system by which I can set up any flybarless controller rather quickly, as long as I can determine what naming convention they use for each of the functions.
For instance, all have P, I, and D gains; however, they may refer to them as stability, consistency, response, P, I, D, “Bell and Hiller,” etc. It can be confusing as they don’t all use the PID standard for naming.
Plug everything in correctly, and if using a type of single input (for example, SBus for Futaba), ensure your channel mappings are correct. Aileron input for the transmitter must be aileron input on the device, for example.
Check that all the “wiggly bits” wiggle in the correct direction. Most of our machines use three servos for the swashplate, so there are three inputs and two selections for each: forward and reverse on the servos, so three to the second power means there are only nine combinations.
I simplify it by checking one servo with collective pitch, reverse if needed, then the second servo, then the third. When collective works correctly, I simply reverse aileron or elevator in the transmitter if needed.
Select normal or reverse for tail servo and set endpoints for no binding. Check collective-pitch range, and adjust for what you want. My reference is 16°of collective [pitch] in each direction (I run some pretty low head speeds, requiring a lot of pitch). I then can reduce it in the pitch curve for each flight mode, so the higher flight modes are normally 12°-13°.
Check cyclic range: each gyro asks for a different reference range of pitch to work properly. For instance, the Futaba CGY750 asks for 9°-10° (I use 10) and the Ace RC GT5 asks for 8°. I then check for binding at the extremes of collective, plus aileron/elevator.
To get the right collective and cyclic ranges and achieve no binding, you may have to play with servo arm lengths a bit on some units. I generally start with 60% P gain (transmitter controlled on most units). Check gyro correction directions for tail/elevator/aileron and pirouette compensation.
Fly and start adjusting the P gain until just below the area where it wobbles on a given axis. Further tuning of the D gain for stopping without bouncing is then done.
Last, I’ll tweak the I gain if the helicopter is not consistent from hover to fast forward flight. This is accomplished by checking things in hover, such as pirouettes and hovering tumbles and rolls, then doing the same in fast forward flight.
For cyclic, I start with hovering rolls at full stick deflection, with the dual rate cut down to a comfortable rate, like 50-60%, continuing to roll into forward flight, [and] accelerating to the highest speed. The roll rate should not change. If it does, I tweak the I gain to achieve consistency.
CM: Any words of advice for newer pilots learning about flybarless setups?
GW: If you are adjusting and tuning and simply can’t get all the wobbles out and it’s very difficult to get I and D set properly, reduce the P gain. It shouldn’t always be the goal to get gains as high as possible. You want it to hold well, be consistent, and stop precisely without bouncing. If you achieve that, there really isn’t much need to increase gains further.
Don’t get discouraged with the process. Contrary to popular belief, the process is the same with all units. The terminology may be different, and the reference ranges for pitches may be different. There are devices that all do the same thing, [and are] offered from different manufacturers. You wouldn’t drive a car differently from a different manufacturer simply because the knobs/levers are in different locations.[dingbat]
International Radio Controlled Helicopter Society (IRCHA)