Source: 2019 Q4 Beartracks, Russ Erb
This article is a requested response to Jared Yates’ article in the Second Quarter 2019 Beartracks, where he wrote about his dissatisfaction with his autopilot installation and its subsequent removal. His article left me with the concern that many would read it and come to the conclusion that an autopilot cannot be successfully installed in a Bearhawk. I say it can.
Bearhawk #164 “Three Sigma” has had an autopilot in it since it was first flown over 10 years ago, and I have been very satisfied with it. It is a great workload reducer in cruise, and especially for instrument flight. It did take some development work, though, and does come with limitations. If you understand the limitations and can live within them, then it will greatly increase the value and usefulness of your airplane.
Expectation Management and Adverse Yaw
Knowing Jared and having read his article, I think the difficulties that he had were twofold. The first difficulty was high expectations, brought on by daily exposure to highly capable autopilots, with various climb and descent modes. This can be addressed simply by reducing your expectations. My autopilot will follow a track, follow a GPS flight plan course, and hold altitude. That’s it. Even without all of the vertical modes (I hand-fly those), these capabilities are still well worth the installation.
The second difficulty arises from dealing with adverse yaw. I did find a simple and workable solution to this. As Jared wrote “The airplane just flies along uncoordinated, not getting any closer to the desired heading or bank.” I think what he was seeing arose from relaxed directional stability. Shortly after completing my Phase I testing, I invited a Cessna 180 pilot who is also a graduate of the USAF Test Pilot School to fly in my Bearhawk. One of the first things he commented on was that the directional stability was very low. He went on to say that this did not surprise him, since all of the airplanes that looked similar to the Bearhawk (mostly from the 1930s era) also had low directional stability. If I look at a side view of the Bearhawk, it does strike me that the vertical tail is rather small. Compare it to what you see on a Cessna 180. It is even smaller if you imagine that the rudder is not there. More on that in a minute.
In the early years of flying Three Sigma, I noticed that a little bit of turbulence would cause the ball (inclinometer) to swing from side to side, and as it did, the autopilot would bank this way and that to try to stay on the commanded track. If the turbulence was too bad (and that didn’t take very much), I would have to disengage the autopilot and hand fly the airplane. This sort of defeated the point of having an autopilot.

Increasing Directional Stability
On the way home from Oshkosh 2011, I stumbled onto a partial solution to the adverse yaw, which was to simply to hold my feet on the rudder pedals and prevent them from moving. If I tried to actively center the ball in response to the autopilot inputs I would rapidly find myself in a Pilot Induced Oscillation (PIO). Just holding the pedals still, however, made a big improvement.
Why did this help? The amount of adverse yaw seen is caused by the relative sizes of the yawing moment caused by deflecting the ailerons and the amount of directional stability. Just to review, adverse yaw is the name for when you put in left stick to roll left, and the airplane rolls to the left but the nose swings to the right, causing a sideslip. This is caused because the downward moving aileron (right) causes more lift and thus more induced drag. The upward moving aileron (left) causes less lift and thus less induced drag. The difference in the induced drag pulls the nose to the right. This is why your flight instructor taught you that you had to use coordinating rudder when rolling in or out of turns, to compensate for adverse yaw.
If the directional stability is strong, such as in a Cessna 180, the adverse yaw moment from the ailerons is overwhelmed by the large restoring moment of the high directional stability. Thus, if we could increase the directional stability we could reduce the impact of the adverse yaw.
The key issue here is the difference between rudder free and rudder fixed stability. In rudder free stability, your feet are off the rudder pedals, and the rudder is free to flop to one side or the other as sideslip increases. When the rudder flops, it isn’t providing much stabilizing moment. In fact, while not absolutely accurate, it is useful to think of it as a flopping rudder provides no stabilizing moment, so if your feet are not on the pedals, it is as though the rudder isn’t there. Take another look at the side view of the Bearhawk and imagine the rudder is gone. There are hardly any feathers on that arrow to keep it pointed in the right direction!
If the rudder, however, is fixed, then it remains aligned with the vertical fin, and all of the area of the rudder is now available to help stabilize the airplane directionally, which then reduces the effect of adverse yaw. This is what happened when I held rudder pedals.
To be able to use the autopilot without having to hold the rudder pedals, I needed a way to increase the rudder free stability. Adding more vertical surface, such as a ventral fin, was not a good solution as that would reduce the amount of sideslip that could be commanded when I wanted it. A better solution was to find a way to make the rudder less “floppy”.
The rudder pedals are attached to the fuselage frame with springs. Many people think that these springs are there simply to keep the rudder pedals from falling toward the pilot. While that is true, they also provide a method to stiffen up the rudder.
The original springs I installed on my rudder pedals had a spring constant of 2 lb/in. That means it took a force of two pounds to stretch them one inch. I searched for the strongest springs I could find that were the proper length at the shortest required extension and were able to stretch to the required maximum length. The springs I found had a spring constant of 5 lb/in. This stiffened the rudder noticeably and greatly reduced sideslip (ball swings) in turbulence, allowing autopilot usage with no feet required under most normal conditions. This passively improved directional stability, providing what I believe Jared was trying to do actively with the third servo.

My Autopilot Installation
I had always planned for an autopilot, having seen the workload reduction possible while flying in a friend’s Cessna 180 from California to Florida for Sun n’ Fun 1999. At the time, only certified autopilots were available, so I designed a servo mount for the right wing to use an S-Tec servo with a capstan to connect to the aileron cable. Between the time I designed that mount and when I was ready to install an autopilot, many other options became available, and these were significantly less expensive. I selected the TruTrak ADI Pilot II autopilot (don’t consider this a recommendation, because while TruTrak will still support the ADI Pilot II, it is no longer offered for sale). Fortunately, I was able to adapt my existing wing servo mount to the TruTrak servo. While my panel includes two Dynon D-10A EFIS units, I did not use the Dynon autopilot because it was not released until after my airplane was already flying with the TruTrak unit.
I used capstans on the aileron cable and elevator cable because that was what I was familiar with. Later Cal Brubaker worked with TruTrak to design an installation using pushrods, shown on Bearhawk CD. This was essentially the installation that Jared used. If I was building now, I would use the pushrod system as it is simpler and more compact.
Tuning The Autopilot
After installation, TruTrak provided instructions to adjust several values controlling the system, to include the gains (sensitivity) of the lateral and vertical response. Since this was the first Bearhawk installation I knew of, there was no guidance on setting the gains. Therefore, I arbitrarily set some numbers, which turned out to be way too high. The results were good in smooth air, but it became very twitchy and unusable in any amount of turbulence.
The TruTrak representatives gave the best answer they could on adjusting gains. They agreed from my description of how the airplane was flying that the gain was too high. They recommended that I reduce the gain until the “autopilot flies like you do”. I reduced the gains twice, keeping records of what they had been and what they were now. One day while I was flying with another pilot, I engaged the autopilot while I was sorting out another issue. We bounced around a little bit in turbulence, but I wasn’t concerned because at some point I was thinking that the other pilot was flying. When I realized that autopilot was engaged, I realized gains were reasonably set.
Limitations
As I mentioned earlier, my autopilot installation will hold a specified track, turn to a track, follow a GPS flight plan course, and will hold the current altitude. While it will make small adjustments to altitude (up or down 90 feet max at a time), the autopilot has no actual vertical modes. Which is to say, it is only useable in the cruise segment of flight. I’m okay with that. That’s all I ever expected it to do.
A few years ago, I noticed a weird but perfectly reasonable response of the autopilot. While flying through sink (downward moving air), the autopilot would command a pitch up to stop the loss of altitude. As the speed slowed, it would command more pitch up. As the angle of attack on the propeller disk increased, the P-factor increased and would cause the nose to yaw to the left. This would start a yaw rate to the left, which the autopilot would respond to by banking right. Thus, I’m flying along and find myself slow on airspeed in a weird nose up sideslip to the right. The solution was simply to disengage the autopilot, hand fly out of the sink, and then re-engage the autopilot.
When I started training for an instrument rating, we tried getting the autopilot to fly the approaches, but when we slowed to approach speed (using 75 KIAS), the higher angle of attack and P-factor from the prop became a problem. The P-factor would yaw us to the left, at which point the autopilot would bank to the right. The reduction of lift would cause the nose to drop and swing back to the right. The loss of altitude would cause the autopilot to pitch the nose up, and the cycle started again. So here we were in some sort of autopilot-induced Dutch Roll-like motion. I tried to tame it by actively using the rudder to keep the ball centered, but that turned out to be a higher workload than simply hand flying the airplane. Thus, I accepted that I have to hand fly the final portion of the instrument approach.
Benefits of an Autopilot
The major benefit of the autopilot is reduced workload in cruise. If you are only flying a half hour to your favorite breakfast stop, this is no big deal. If you have another qualified pilot, you can reduce your workload by having the other pilot fly. On the other hand, if you are the only qualified pilot on board, flying five legs over two days for 15 hours of flight time to get to Oshkosh (yes, I’ve done that), then having an autopilot can significantly increase the probability of a successful outcome.
For instrument flying, especially single pilot IFR, the autopilot can be a huge workload reducer. Yes, mine won’t fly the approach legs, but it will do an amazing job of flying the enroute legs. Even in training, I will engage the autopilot at cruise conditions between practice approaches while I am programming the GPS, setting up the approach, getting ATIS, and generally getting my act together. These are very difficult to do if I simultaneously have to fly the airplane, but are very simple if I can hand the airplane over to another pilot, in this case “George”.