From another NTSB report, this related to a fatal accident when they attempted a turn back after a simulated (!) EFATO. The quote gives some more info on the stick shaker/pusher. Apparently the A/P disconnects if it experiences -0.3G or 1.6G.
The airplane’s Pilot Owner’s Handbook states the following: that the normal takeoff in the airplane is accomplished with 15 degrees of flaps, rotates at 81 knots, and climbs out at best angle (Vx) of 110 knots or best rate (Vy) of 120 knots. Takeoff flaps take 17 to 20 seconds to retract; this reduces dramatic pitch changes during climb out. The airplane’s wing level stall speed (defined by pusher activation) changes during a 15 degree flap retraction from 76 knots to 91 knots. If the airplane is banked 45 degrees, with flaps up, the stall speed increases approximately 19 knots, to 110 Knots indicated.
AFAIK the KFC325 is just a normal old “classic” autopilot, identical to the KFC225 (i.e. the usual modes AP PIT HDG NAV VS and it has a Hold on all of those, but no IAS hold AFAIK) but it does have one extra feature: it has an airspeed input and uses that to modify some control loop parameters. It was used on the pre-G1000 TBMs.
bq.
The KFC325 in the TBM does have a IAS-hold and works perfect.
Peter, I don’t disagree that your scenario would put a lot of pressure on a pilot naturally. Pilots can get overwhelmed. But single pilot IFR is a very serious business and if you are going to do it in real weather you owe it to yourself and more importantly your passengers to have the knowledge and training to cope.
I have a highly automated aircraft and could literally do gear up, a/p on until short final (although I don’t). For too many pilots an autopilot disconnect or failure becomes an emergency when it shouldn’t be. No matter how much systems knowledge you have there are always why-is-it-doing-that moments. Training is the key.
There is an analogy in motor racing. The hardest thing in racing is working on improving your own driving. It takes time and effort. So instead many people spend money on their cars, fresh tyres every race etc as it is easier. But that only materially improves your performance if you are limited by the car. Most amateur racing drivers are limited by their own skills.
I think flying is the same. We are generally the weakest link in our aircraft.
How often do we see in Internet forums after an accident report “XYZ aircraft crashed during approach to airport” speculation about multiple low probability simultaneous systems failures which could have led to the accident. When the report comes out, it was pilot error through a stall/spin. It is human nature to try to blame the machine but I think we have to be honest that if we have an accident it will almost certainly be our fault. So we have to work hard to avoid making those mistakes as much as possible.
I’ve done most of my flying in actual IMC without an autopilot and lots of it enroute.
At UK low level, so have I, and I can hand fly for ever (not sure I would be much good for anything too exciting post-landing though) but let’s forget the ideal scenario where everything is going well and the sun is shining, and transplant yourself into a plane where (hypothetically of course) you have 4 screaming /shouting kids, a wife who is scared/crying and putting pressure on you to get out of there (passengers universally hate turbulence in IMC; they put up with it in an “aluminium tube” only because they cannot walk up to the cockpit door and bang on it while screaming “get me out of here!”), “you” are behind the plane because you have not flown it much, you (like most pilots) don’t have much aircraft systems knowledge and easily get into the “WTF is it doing now” situation, ATC are giving you instructions for which you are not prepared because you have not set things up well ahead of time, and then the autopilot fails and without any warning chucks an out of trim aircraft back at you to hand fly.
If you (and Jason) think the above is unrealistic, speak to one of the accident investigators who listened to the ATC tapes where the GA pilot’s finger froze on the PTT button during the period before the impact.
Also in clean configuration the un-accelerated Vs may be a bit higher, say75 knots?
Yes – good point, forgot that. Probably 80-85kt (without looking it up).
AFAIK the KFC325 is just a normal old “classic” autopilot, identical to the KFC225 (i.e. the usual modes AP PIT HDG NAV VS and it has a Hold on all of those, but no IAS hold AFAIK) but it does have one extra feature: it has an airspeed input and uses that to modify some control loop parameters. It was used on the pre-G1000 TBMs.
Where I used to be hangared, they serviced TBMs with the 325 (EFIS-40 TBMs i.e. pre-G1000) and Socata stipulated (don’t think it was an AD but probably was an SB and this company always did the whole MM 
) that the AP servos are removed at every Annual and the slipping clutch torque is checked and adjusted to spec. This was widely regarded as a stupid job because the pilot should check that clutch on every preflight (as on the 225) and it took 2 man-days and cost four figures. I saw it done many times.
The 325 servos are a curious design in that the clutch is (IIRC) inside them, whereas the 225 servos have the clutch in the mounting bracket.
IME, on most aircraft with these King APs, this slipping clutch (which is an essential safety feature in case of an autopilot going haywire but not disconnecting via the solenoid clutches inside the servos) is seized solid and the pilots never do the preflight check to make sure they can overpower the AP. I doubt that was a factor in this accident though.
The VS of a PC12 is about 65kt, so should the stick shaker come on at 109kt?
While speed and configuration are parameters, angle of attack/boundary layer separation via stall vane (s) or an AoA system may be the main parameter triggering the stick shaker/pusher? Also in clean configuration the un-accelerated Vs may be a bit higher, say75 knots? It would have been interesting if there had been more comment on the design and activation parameters of the PC12 system? Activation at 109 Knots may suggest acceleration either due to gust/turbulence or the A/P or pilot had pulled some G – the pilot being possibly the more likely actor to pull G? Although being near convective activity turbulence would also be a candidate for triggering the stick shaker? Alternatively the stick shaker mode may have had quite conservative activation parameters if the aircraft is in clean configuration?
Very interesting accident scenario, however I think it somewhat jumped too quickly to the conclusion that it was pilot in experience, which self evidently is a factor – a bit more engineering discussion on the automatic flight control aspects, the stick shaker/pusher system, and the interface with the A/P would have been useful.
Agreed — I’ve done most of my flying in actual IMC without an autopilot and lots of it enroute.
Ultimately, this accident happened apparently because the pilot could not take over the flight manually, for whatever reason (most likely IMHO because he was not current on type and got caught off guard, in IMC) but there are very few sky-gods around and autopilot failures do not help.
Peter, you do love this sky God term and i am not sure it adds anything to these discussions. I just don’t accept that hand flying an aircraft IMC after an autopilot failure involves super-pilot skills. It is normal flying if you are an instrument rated pilot. If you can’t do it you should not be flying in IMC.
Josh, it wasn’t a Mustang, it was a CJ2 in the incident that damaged the airframe and returned to Leeds. We operate a similar aircraft and allowed the AAIB to take some measurements on ours. For interest I tried a stall at climb power and FL430 (in the sim). It wasn’t difficult but a lot of height is lost in the recovery. Maybe the sim isn’t accurate that high
I don’t buy the idea of stick shaker or pusher activation at over 100kts in this sad accident, but have never flown the type myself so wouldn’t know for sure
The VS of a PC12 is about 65kt, so should the stick shaker come on at 109kt?
The stall warner should come on well before the stall, and that really wakes you up – again, unless you are really busy in which case you might filter it out.
From the report, the AP disconnected at an IAS of 109kts. Those with experience on type suggest normal climb speed around 150-160 with that being within the stall warning range. There’s no hard proof, but it seems a reasonable inference.
Climbing in VS mode will get you in a lot of aircraft, even those with supposed reversionary modes to prevent you stalling. It is also believed to be the cause of the Mustang incident I mentioned. Plenty of older cheap autopilots (including those fitted to many bizjets) display an unpleasant oscillation in speed hold mode, and thus VS is the preferred climb and descent mode for comfort. It requires focus flying two crew, fairly easy to see how a single pilot with potential distractions and little instrument experience made the initial error.
