Thinking about a Third Autopilot Servo

[svyolo]

Yeah, a "Yaw Damper", or yaw damper function, is for bigger, heavier planes where the plane has so much inertia that it will occillate in yaw if displaced in yaw. Nowadays most of those bigger and heavier airplanes have swept wings. On an older plane, the "Yaw Damper" was a separate device. Nowadays it is just a software function.

But yaw is yaw, no matter where it came from. The EFIS systems are miniature versions of the INS based instrument and navigation systems on aircraft for the last 35 years. They have gyros and accelerometers that can measure pitch and pitch rate, roll and roll rate, and yaw and yaw rate. It doesn't matter what caused them.

Using those to run the integrated autopilot function built-in to the EFIS in some ways is easy. Commanding a roll or pitch to climb or turn probably is that hard. Expecting it to perfectly coordinate a turn with a rudder servo? I am not sure how well it will work. Depending on the airplane, it might "lead" too much, or "lag" Thinking about it, I think that to get it to work really well, it might be necessary to have adjustable gains on the rudder and maybe the ailerons. There are far more RV's that any other Exp aircraft, so if I was an EFIS manufacturer, I would tune it to work well on an RV.

Like I wrote in a previous post on this thread, VAF had a writeup about RV guys turning the yaw autopilot function on right after takeoff, and allowing the EFIS to coordinate turns the rest of the flight. Would it work across a lot of different roll inputs? Again, not sure. I bet if the gains are adjustable, that you would have a good chance of getting it to work well at 1 roll rate, which is that commanded by the roll function of the autopilot.

I just ordered a Pitot and remote magnetometer from Dynon. I am going to hold off on the rest until I am almost ready to fly, as that is what everyone recommends with the newer EFIS's. They just keep getting better, and cheaper.

[Rollie]

I finally got to test my autopilot this week. It had been wired incorrectly by the avionics shop and I finally got a different shop to fix their mistakes. I have all Garmin stuff in my Patrol and once you go through some ground setup and testing its time to do in flight testing and tuning. This mostly consists of engaging the autopilot, putting in a manual input and seeing how many oscillations the autopilot makes to correct back to what it was doing and then adjusting the gain. You do this for each servo then you test all the modes of the autopilot and if you set the gain properly, they should all function well. If not, go back and start over with the gain adjustments. It was simple to set up and works well. We finished the flight by watching as the autopilot flew an LPV approach (like an ILS but GPS based) down to 200 feet on glide slope and on course.

I only have a pitch and a roll servo, mostly because I couldn't figure out how to go about making a servo control the rudder. I'm pretty spoiled, I've had yaw dampers in the things I fly at work for years and they are working even when you are not coupled to autopilot which means that shortly after takeoff my feet go to the floor, even when I'm hand flying, and stay there until I'm on final. Now with no yaw servo/damper, I need to keep my feet on the pedals even when coupled. Its no big deal when hand flying, its natural to apply rudder as you apply aileron, but at least for this first flight, it felt very unnatural to only control the rudder while the AP did the rest. Even though you know its about to turn and you can feel and see it getting out of trim, I had a hard time putting the right amount of rudder in, either putting in too much or too little. I'm sure I'll get used to this quickly but for now it has me thinking about adding another servo.

[rodsmith]

Good info Rollie, thanks for sharing.

[Bdflies]

Since the beginning of this thread, I’ve become more aware of yaw, induced by the autopilot. Just this week, actually, I made note of my inputs on the rudder pedals. My flying experience is just the opposite of Rollie's. I’ve very little experience flying with a yaw damper and a lot of hours in some pretty unstable planes. If I'm at the controls, my feet are on the pedals. It’s completely natural to me. During this week's flight, I caught myself uncounsciencly applying rudder pressure. I chuckled when I realized what I was doing. For a while, I took my feet off the pedals. Didn’t last long, though. Not that the plane was slewing about, mind you. The feet just gravitated back to the pedals.
Reading Rollie's post, I was struck by how subjective the yaw issue can be. Sounds like a good subject for a short video!
I've mentioned that my gain is set pretty low, so the autopilot eases in roll control. I'll check the numbers and post them.

Bill

[svyolo]

I think if I was going to try to install a rudder servo it might be easier to take the two rudder pedal cables and have them actuate a bell crank. Then the rudder cables aft of the bell crank are a closed loop, and you could connect the servo to the bell crank. Where? Maybe under the rear seat floorboards, or just aft of the cargo bulkhead.

Or maybe operate the rudder pedals, or the torque tubes, directly?

[jaredyates]

The capstain/pulley system I created for the yaw servo seems to be working OK mechanically, but I have not been able to get the system dialed in, only because of a lack of support from Dynon. They have still not issued any instructions about what several of the settings do to the system, which is really surprising to me. This is now a system that they are installing in TC airplanes, where documentation is king, but yet they can't even tell us what several of the parameters are? I had decided after Oshkosh to go ahead and remove it all, but then right after that, unrelated circumstances led to a new discussion with Dynon tech support. That discussion sounded promising, like maybe there was going to finally be some documentation, but that was early August. A month later after not hearing anything I checked back and it sounds like it is just not a priority for them, and that the discussion from early August was not genuine. I've been promised a series of articles for the Beartracks about Bearhawk-specific rigging from one of the guys that knows the most about it, which will hopefully equip me to make the airplane fly even better, perhaps making the autopilot even less appealing. If I can't get the system to maneuver the airplane, I don't really need it for straight and level flight, so I don't think it would be worth the weight. Thankfully I had not yet gone to the trouble of installing the electric elevator trim. If someone with more tinkering patience or lower performance standards is planning to buy the parts and is interested in mine, we should discuss it. The right offer might push me off the fence to go ahead and remove it all. I'm probably better off selling the expensive parts to a RV builders, being that the system seems to work great in their planes.

[AZBearhawk272]

Well, I be that someone who’s scribing a missive on rigging a Bearhawk. From a symplistic viewpoint do not expect an autopilot to fly an airplane that cannot fly by itself. That said I flew 3K miles to and from OSH one day was 7 hrs solo and never wanting for an autopilot.
Let it be known, I have experience rigging aircraft, 0 experience with auto pilot! This should be interesting.

Stand by, first installment heading Jered's way
shortly.

[svyolo]

Jared;
Autopilots should fly the airplane a LOT more gently than hand flying. In the case of something like a BH, I think having the roll rate (gains) set very low will help reduce the adverse yaw in the first place. It might also help if you rigged the ailerons 1/2" up each, like Mark suggests, but you have probably already done that.

I am planning on Dynon/AFS. Do they have adjustable gains for all 3 axis?

I am going to fly 1st without AP servos, but do plan on adding them. I am already planning on how and where to mount them.

Just another thought. Planes with a lot of adverse yaw tend to have a lot of pro-verse roll. The BH has very little dihedral, so I doubt that would be the case in our instance. Have you ever thought of just hooking up the yaw servo, to the roll channel?

[svyolo]

It might sound strange, but the "roll" function is really to maintain a heading or course, not control roll. If the rudder alone can do it better than the ailerons alone, then it might function better.

[jaredyates]

It might sound strange, but the "roll" function is really to maintain a heading or course, not control roll. If the rudder alone can do it better than the ailerons alone, then it might function better.

I think their system is more advanced than that. There are lots of settings for various types of gains. For example, there are 7 different settings in pitch, 3 of which have no documentation at all. Several settings in yaw, but none other than the basic sensitivity and gain are documented at all. And those are only very vaguely documented. It's not that the system isn't smart enough, it's that they have released the features without adequate guidance of how they are supposed to work.

[rodsmith]

I installed 4 pulley brackets on the firewall tubing for possible future yaw servo. Total weight cost 4.6 oz. Would have been near impossible to add to a completed plane.

[schu]

What are the 7 settings?

[AZBearhawk272]

Have no idea what the autopilot P,I,D,R gain settings will be for an auto pilot setting but over the next month or so I will find and share.
Today I flew a nicely rigged Bearhawk that happens to fly hands off.
So with my feet flat on the floor, right and left turns were initiated, sustained, and removed
and the ball never moved more than halfway out the cage. I see no reason that an autopilot would have any difficulty tracking the plane to
a course.
Fortunately there is a Dynon autopilot on board that has never been implemented. I see a challenge coming on.

[svyolo]

What are the 7 settings?

Jared, I wanted to ask the same thing. Is the support manual online? My recent light plane experience is nada. My experience with flight management systems is all of my adult life. I don't know your, or others' experience with autopilots. Maybe some have a lot, and some have none.

Modern (last 35 years) autopilots, flight directors, and flight management systems have many modes. They are generally broken into lateral (horizontal), and vertical.. You set a paramater, either lateral or vertical, and the FD and/or AP use pitch, roll, yaw control surfaces to control that parameter.

Older autopilots (light aircraft, jets older than 35-40 years), were much simpler. Older Boeings had a little, I don't know what to call it, "control", that allowed you to set a pitch attitude or input a roll command. Those are now "degrade" modes of more modern autopilots.

Newer systems are much better. You enter a heading ("Heading Select"), LNAV (lateral navigation) or it can track a LOC or some other parameter. Vertical is Vertical speed, VNAV (Vertical Nav which follows a path) or Level Change or Open Climb or descent, which is an idle, or max continuous power climb.

I don't know what Dynon or Garmin call these things, but I bet they are the same or similar to Airbus and Boeing.

I do have a big question for anyone who flies a BH. Say you are heading 180, and you want to turn to 270. Push on right rudder, can you turn to 270 fairly easily with only rudder? I already know that the BH had enough adverse yaw that just using the ailerons is not that effective. I am asking this because if the BH has enough Pro-verse roll, maybe it would work better to control the "lateral" autopilot channel with a rudder servo, than an aileron servo. Or maybe not. I did fly one plane (a tug for hang gliders) that I think they should have just left the ailerons off. And I built a dozen RC gliders with no ailerons. They had much more dihedral than the BH, so I am skeptical how well it would work.

Just asking.

[Bcone1381]

I think the 'settings' Jared is referring to are settings, as opposed to modes. I envision that the settings control the behavior of the autopilot servo movement when that servo is called into action.

For example, when the autopilot needs left aileron, what roll rate does autopilot desire, how fast will the servo move, and how far will the servo move to satisfy the left aileron command. Will the system monitor the progress of the roll as it makes the command to adjust the rate of roll so that a desire....stuff like that is what I imagined.

When comparing an RV-8 to a Four Place Bearhawk, the ideal settings for one aircraft would result in very poor performance if used in the other. Maybe too lethargic, too slow, too jerky, too snappy, overshooting a desired heading, wallowing back and forth in a rhythmic fashion...... WOW!! Why did it do that???? Whats it doing now! Hang-on Honey!!

[jaredyates]

There are manuals online, but the manuals don't address the settings. Here, for example, are the settings in the pitch menu:

Poorly documented:
SENSITIVITY
PITCH GAIN
G ERROR LIMIT and GAIN

Absolutely no documentation:
ALTITUDE GAIN
PULL RATE
VSI GAIN

Yaw settings (no documentation at al):
Yaw Sens:
Rudder Gain:
AY Gain:
Rudder Rate:

I've had a variety of totally unrelated answers about what sensitivity and gain are. One guy said that they were multipliers, or something wacky like that. The most plausible answer is that sensitivity is how closely it looks for accuracy in the measured value, while gain is how aggressively it strives to correct once the value is deemed to need correction.

Modes include the typical heading hold, altitude hold, vertical speed, IAS, an nav couplings.

[svyolo]

I believe sensitivity would be how much displacement, or rate of displacement, it takes before the autopilot commands a correction. Gain would be how much control displacement, or rate of displacement, the autopilot commands in response to a certain amount of of measured error.

As an example lets say pitch sensitivity is set at 50 feet, or some numerical value that corresponds to that. The plane can wander off 49 feet, with no correction from the autopilot. At 50 feet off altitude, the AP commands a correction. How big the correction is, or how fast it is applied, is gain.

If it were me, I would set all the sensitivity settings at a mid range setting, and set all the gains to a very low setting. I would slowly increase the gains until you get the response you want.

Too much sensitivity would be annoying, but not unsafe. Too much gain would be both unpleasant and potentially a bit bad. You'll know what it is because it will be an oscillation, and maybe one that gets bigger and bigger with each oscillation.

You might also want to see if you can get a look at Garmins documentation. I would bet it is very similar, with similar terminology. Hopefully explained better.