We pilots routinely kill ourselves. Not only is this a bummer for the people intimately involved in the wreck, its also bad for business. Cirrus Aircraft defined itself as a trend-setter in combatting this with certified aircraft when the first SR20s rolled off the line with whole-airframe parachutes in 1999. Better than just surviving a catastrophe at the expense of the airplane, however, would be recovering back to controlled flight. Cirrus again led the charge in light GA with “LVL” button as part of the Garmin Perspective avionics suite (a modified G1000) in 2008. More sophisticated than a simple rip-cord solution, the LVL (blue) button leverages the tumble-proof digital “gyros” to pitch and/or roll the aircraft back to level flight without overstressing anything. Avidyne came up with a similar feature called “Straight and Level” in its DFC 90/100 autopilots. This makes the feature retrofitable to aircraft formerly equipped with the STEC 55X. Both Garmin and Avidyne have protection against going too fast or too slow-in some cases even when the autopilot is technically turned off. This is taking us into uncharted territory: these autopilot systems can actually disobey the pilot for that pilots own good. All Airbus and “Im sorry, Dave. Im afraid I cant do that” pot shots aside, this begs the questions of whether these systems really work, whether they are worth the expenses, and, more philosophically, is this a direction we want to go? Right now the only contenders in small, certified aircraft is the Cirrus Perspective system and the Avidyne DFC-series, so we’ll focus on those.
Smart autopilot Tricks
Step one is clarifying what the systems do and don’t do. Weve tried both the Perspective and the DFCs one-button recovery to level flight. They work as advertised. The systems are even smart enough to unload the wings before attempting to roll level and pull up. Unofficial (but we’ll say reliable) sources have told us the DFC has even demonstrated recoveries from inverted dives. We expect the Perspective would do the same.
Both systems also have under- and overspeed protection when the autopilot is engaged. If the pilot commands a climb without adding sufficient power, such as during a botched missed approach, and the aircraft will slow to a specific speed and then pitch down just enough to hold that speed. Meanwhile, beeps, flashing text, and audible callouts will warn “Underspeed Protection Active.” Restoring power will restore the climb and disable the mode. Again, we tried it and it works.
The Avidyne system uses a similar tactic of the pilot commands a descent that crosses Vne: It pitches up. The Perspective system is different. It only offers overspeed protection with the Electronic Stability and Protection (ESP) thats part of the $39,900 Perspective 700 Select package. ESP was also retrofit on many Perspective SR22s sold before the function was offered.
ESP runs full-time, even with the autopilot off. If you overspeed the airplane, either by commanding too much descent on autopilot or by manually pitching down, the autopilot servos push back on the stick to slow the plane from 200 knots to 190 knots. Yes, the autopilot could be pitching down because you told it to descend, and then pitching back up not to overspeed-in an endless loop until the pilot fixes the problem. Good thing it isn’t smart enough to need psychotherapy.
Its only 10 pounds of force, so you could easily override it. In fact, when we tried a manual dive in an ESP Cirrus, we found the push subtle given the higher stick forces at 200 knots. The combination of the push and the visual and auditory warnings should get a pilots attention, however.
ESP also limits pitch and roll. Pitching to 17.7 degrees nose up or 16 degrees nose down will earn you that same 10-pound pushback toward level flight. Rolling past 45 degrees of bank will generate a five- to seven-pound push on the stick back toward about 30 degrees of bank along with only a subtle visual cue (see sidebar, page 9). An ESP aircraft (Garmin has the system running on some other G1000-equipped aircraft) cant get into a steep spiral. It may spiral and descend, but the bank will keep oscillating between 45 and 30 degrees. ESP can be temporarily disabled at any time if the pilot chooses.
The evolving term for this kind of thing is “envelope protection.” The autopilot steps in with warnings or actual action when the pilot flies outside of a prescribed performance envelope.
High-flying Cirrus aircraft take this autopilot-as-overseer a leap further with Hypoxia Check. If a pilot cruising at 15,000 feet doesnt push any buttons for 30 minutes (virtually impossible for a Cirrus pilot with a pulse), the system throws up an alert (which, of course, entices the pilot to push a button). If the pilot doesnt respond, the autopilot will descend the airplane, first to 14,000 feet and then to 12,500 feet. The time before a warning tapers from 30 minutes at 15,000 feet down to five minutes by 25,000.
Weve come a long way from a Nav-o-matic wing leveler.
Holes in the Safety Net
“The parachute gets all the attention, but Im convinced this system is just as revolutionary,” says Bill Cattley, Northeast Division Director for Cirrus, when talking about ESP. The jury is still out on whether thats true, but if it saves even one life, how could anyone say its a bad thing? Weve lost at least a few Cirrus owners to hypoxia alone.
Smart autopilots arent entirely idiot proof. The Perspectives System originally uncoupled the autopilot when the pilot initiated the missed approach. Now it stays coupled and commands a climb, relying on the underspeed protection to prevent a stall on the missed if the pilot doesnt add power.
We tried this and could see a pilot convinced he had started the missed and confused why the aircraft isn’t climbing. A confused mind has a way of misreading, or even ignoring, alerts. The warning says, “underspeed protection active,” not “add power, stupid.” Its not quite at the sardonic CFI level yet. That means it can only point out the problem, not the corrective action.
The same is true of preventing an accelerated stall in a base-to-final turn. We suspect 10 pounds of push and a visual-only warning on a PFD wont have any effect on a pilot pulling and banking while focused outside the cockpit on the runway hes overshooting.
This gets at some of the user interface issues with smart autopilots, whats the best way for them to intervene? Steve Jackob-son of Avidyne puts in this way, “There are really two conditions to worry about. Ive gotten myself distracted and need someone to slap me in the face, or Ive become a dog watching TV.” By dog and TV he means the pilot sees the right information, but just isn’t processing it correctly. Thats where “Add power, stupid” may be whats really needed.
Avidynes DFC 100 has envelope protection when the autopilot is in flight director mode that simply alerts the pilot. The DFC 90 retrofit version doesnt have this feature. Avidyne hasnt announced any upgrades to the DFC series for envelope protection when the unit is completely off, but we suspect they will to compete, if nothing else.
That brings up the potential for unexpected complications with this kind of technology. One of the carrots to get owners of STEC 55X autopilots to stomach the $10,000 upgrade to the DFC 90 was that the new indicated airspeed mode could be used for emergencies. We reported on this in our September 2010 review of the DFC 90. If the engine fails, the autopilot can hold the best glide speed of 88 knots while you deal with the emergency.
Except that we know of installs where that doesnt work. With servos that are technically within spec but less than optimal, the airspeed mode can overshoot and then get stuck in underspeed mode at about 80 knots, rather than 88. Weve seen this in both the SR22s we fly periodically that were upgraded to the DFC 90.
Avidyne tells us, “It would take one hand to count the number of open issues” with the DFC 90, but considering we saw this issue twice is two airplanes, we expect its more widely spread. Its possible DFC 90 pilots are out there assuming this feature will work and will only discover it doesnt when they actually try to use it. The AP
still wont let the airplane stall. But flying seven to eight knots below best glide with limited heading response isn’t what the upgrading owners signed on for.
CFI in the Machine
While there’s something vaguely Orwellian about an airplane taking matters into its own hands, the systems are really logical extensions of terrain-warning alerts and audio callouts, or red chevrons on the attitude display pointing back toward level flight. Weve flown them. They arent obtrusive in practice. Even the most conscientious pilot can become distracted or hypoxic, and there’s real room to save lives here (see sidebar, page 10).
But there is the more slippery and politically incorrect issue that smart systems like this lure people into personal flying who wouldnt-or shouldnt-be in that seat. Cirrus is arguably again the poster child here, and not in a flattering way. There are many conscientious, competent and safe Cirrus pilots. But there are also many low-time pilots relying on the technology to make their mission possible. We don’t see this as a bad thing per se, but we, and many others, opine its why the fatal accident rate for something with so much safety gear is not significantly different than aircraft without it.
The new startup screens in the Cirrus we used for the ESP tests drove this home. They prompt the pilot to choose personal minimums as an “infrequent flier, average pilot or elite aviator.” Each category has a symbol over it youve probably seen before: a green circle, a blue square and a black diamond.
The reasoning is to make the decisions simple. But in-flight decisions usually arent simple. Taking a 200-knot machine cross-country isn’t a weekend on the slopes.
While a one-step unusual attitude recovery and overall envelope protection is a fine thing, and arguably worth the expense, we wont get too excited until we see smart systems that aid and augment some higher-level thinking. Where are the smart systems warning us weve lined up with a runway thats too short for the current temps and pressure, or point out our destination just dropped below approach minimums and offers some alternatives?
We still think there’s something important about a pilot wanting to actually, you know, fly. Hopefully the future of smart systems will be some artificial intelligence that aimed at helping a pilot do that better-rather than just feeding the dog.
