…KFC225 one day approaching the LOC in NAV mode. This is very unusual anyway…
It’s pretty usual if you fly STAR prior to it and have roll steering.
There is a good article by J Liscomb on the Mystique of Autopilot Accidents. It was reprinted in the American Bonanza Society magazine
https://www.bonanza.org/magazine/archive/2000/december-2000/the-mystique-of-autopilot-accidents/
It was originally published in ISASI the accident investigators association.
The important item is the following:
Another characteristic of this system is that the pitch servo will not mechanically disengage the clutch mechanism where there is a load on the interface between the pitch servo and the clutch. This is true even if the autopilot is disengaged electrically.
The disconnect functions may not have a fail safe unless there has been an unload on the affected controls, which makes a nice symmetry with UPRT practice.
That article requires a login.
I can see this could happen with some autopilot servos. I don’t think it is possible to happen with the King servos, specifically those on the KFC225. You can get clutch problems though if the clutch has not been tested for many years. It is a preflight item but probably most people don’t do it.
I sat through this and was appalled. Not so much at the tragedy, but that with that level of automation, that level and capability of aircraft, that level of inability to iinterpret any elements of the system, that level of sad death of innocents, is truly shocking.
I do not think they could even switch it off to hand fly the thing. I had a very similar experience many years ago in a Cherokee 6, overloaded with auto gizmo, and the owner did not have the first clue how to operate it.
@BeechBaby very sobering, the other pilot was the non licensed son. The pilot had medical issues and under pain relief medication so arguably incapacitated. The estate was emptied out by the subsequent lawsuits. There may have been autopilot issues which may emerge in the final report.
The servo issues described in the PDF above may be specific to the KFC200 (the original article was in year 2000) and I don’t believe the KFC225 is capable of this since overpower slipping clutch is located outside the servo, in the servo mount.
I watched most of the video. It merely tells us that you can have $5M and know nothing about the plane you bought, and fool six passengers to fly with you.
Honeywell produced a service memo years ago regarding servo maintenance and in most cases it gets ignored.
King_servo_SM292_pdf
Yes; the only time I have seen the servo clutch torque checked annually was on TBMs. The customers there paid probably 5k for that little job. Not to mention the maintenance-induced damage done by the work.
It’s a huge job actually in the context of an Annual, plus almost nobody in Europe has the test rig, plus you are supposed to be preflight testing the clutch slip (I do).
BeechBaby wrote:
I do not think they could even switch it off to hand fly the thing.
Once the “Trim Interrupt” switch is activated, all four (4) trim motors and the AFCS are disconnected
and the runaway is stopped (refer to POH Sections 3 and 7-3).
https://www.pilatus-aircraft.com/data/tech_pub/PC-12%20SIL-003.pdf
This is system specific but the “problem with servos” is that with DC motors the gearbox ratio is huge – of the order of 1000:1 to 10000:1. That means that removing motor power does nothing. The only way to overpower the servo (via a yoke or stick) is via a slipping clutch located at the servo output.
On the King servos, there is also a solenoid-operated clutch which, when open, makes the servo output run freely. That is the normal state of the servo when the autopilot is not used. AIUI, the assertion made above is that this mechanism will not release (by removal of solenoid power) if there is torque on the servo output at the time. That would be a truly crap design! But a Bonanza, pre year 2000, could have only King or STEC servos, no?
Garmin servos use stepping motors and those don’t need anywhere near that much of a reduction ratio, so a slipping clutch may not be needed. I would still be amazed if certification permitted a reliance on that, however. And they still do need a solenoid clutch, because a stepper needs a fair bit of torque to drive it via the shaft.