Also I have to say that I've come across NO instructor (or examiner) who has either suggested this overpower test (or pulled me up when I didn't do it). Do the instructors on this forum teach it?
I instruct (since 1992) as per training syllabus of the FTO and AFM of the various aircraft types. As I stated more than once on this thread, I have not read or heard of this autopilot-overpower-test until yesterday. And unless it will not be included in either training syllabus or AFM, I can't and won't teach it.
...they almost got killed once when test flying a Metroliner. They experienced a trim runaway...
I have flown the Metroliner (luckily only briefly) some years ago. It only has electric trim (for reasons known to Mr. Swearingen alone) and heavily relies on it to function properly because it has large variations in trim during normal operation. Therefore trim malfunctions (runaway, mechanical failure, electrical failure) take up a big portion of the theoretical and practical part of the class rating course. I don't remember much of the Metroliner now, but I still can show you blindfolded how to disconnect the electrical trim by various means. Therefore it puzzles me a little that these people "almost got killed" by a runaway trim. But then the trim of a Metroliner works completely different from that of most SEPs. Like in almost every jet airliner, it moves the large fixed part of the horizontal stabiliser through a vertical jackscrew and therefore has much more power than the small moving section controlled by pilot (and autopilot - only installed in every second Metroliner anyway).
As an instructor, I do make a point of covering autopilot subjects when checking pilots out on new aircraft/types, especially when giving diff. training on EFIS / glass cockpit because of the amount of integration. I make sure limitations are covered (airspeed, altitudes, flaps), also all the ways A/P and el.trim can be disconnected/interrupted, and I make sure that the pilots knows the principle of the autopilot (rate based or attitude based, GPSS etc). I have not consistently emphasized the overpower check so far, mainly because I have never read an accident report that suggested that a pre-flight auto-pilot overpowering check would have prevented the accident. But I might reconsider that based on the failure mode Peter mentioned in the initial post.
Safety issues with GA-pilots and autopilots seem mainly to concern vertical modes: trim runaways, the pilot "helping" the autopilot inducing A/P counter-trimming, stalls during climb, descent into terrain; and also undetected disconnections (in IMC/darkness). Those are the issues I try to adress, especially to pilots flying IFR, and especially with two-axis autopilots.
I remember reading somewhere that in the US, it was found that while single-axis autopilots generally improved safety in GA aircraft, autopilots with vertical control (elevator w/ or w/o trim) generally was a challenge.
I never do, nor have I ever done an autopilot overpower test.
I have an STEC and also have to admit never having done this test. Whilst on this subject what are peoples views please on the a/p being used in turbulence. I appreciate the term turbulence is subjective but comments appreciated.
As an electronics designer, what amazed me most is that the King autopilots, and most other avionics we fly with, do not contain a watchdog.
Why should a Century IIB have a watchdog? 8-)
But seriously, watchdogs are no panacea, I've seen too many firmwares crashing in a way they just kept resetting the watchdog but ceased to work otherwise.
But seriously, watchdogs are no panacea, I've seen too many firmwares crashing in a way they just kept resetting the watchdog but ceased to work otherwise.
Yes but that's just crap software design. The problem is that the autopilot software could potentially get crashed by something outside the designer's control. For example the KFC225 does get crashed (or somehow massively affected) by external RF. If you read about the 3 or 4 failures over the same spot in France, you can see the need for making the thing a bit more robust.
Whilst on this subject what are peoples views please on the a/p being used in turbulence
If it works, use it. My KFC225 actually does a very good job; the King autopilots that use a KI256 for pitch/roll reference are generally pretty good. But one has to watch it because they will generally disconnect if the pitch/roll exceed certain values e.g. 20 degrees of roll (that's really bad turbulence).
I never do, nor have I ever done an autopilot overpower test.
Model?
For example the KFC225 does get crashed (or somehow massively affected) by external RF.
If so, wouldn't the real solution be to make it EMI tolerant?
Yeah, but "how much" EMI?
Also you can't control how some chimp does the wiring, for example.
The particular EMI cases I saw are almost certainly SAM missile sites, which will be tracking you with a very narrow beam of probably a few hundred watts.
I appreciate the term turbulence is subjective but comments appreciated.
There is an objective definition for everything. Here is a good description of the variuos levels of turbulence, googling a little more will also lead to it's refenrence and the values of the associated g loads (1.3 1.6 and 2.0 IIRC for light, medium and severe):
Whilst on this subject what are peoples views please on the a/p being used in turbulence.
I'm afraid I'm getting a little boring here by constantly repeating: "Operate it according to the AFM". I don't know how many types of autopilot I have encountered so far, but all were different in that respect. Some manufacturers don't trust their product much and want turned off, others don't trust the operators of their product and therefore like it to stay on. Or on in a specific mode while flying at a given speed (as in "my" Citation).
But seriously, watchdogs are no panacea, I've seen too many firmwares crashing in a way they just kept resetting the watchdog but ceased to work otherwise.
Properly designed software written in the right way and throughly tested and modelled should not do this. In the olden days there was plenty of assembly language programming done and some of it was not exactly the best or most robust. I expect I was responsible for some dodgy stuff in my youth. These days the tools and techniques available are far superior.
Yeah, but "how much" EMI?
These days automotive electronics routinely withstands field strengths of 100 volts per metre and more. If GA avionics products withstood this level I doubt you would ever have a problem. A few years ago 15 v/m was the best around and to be honest cars were't going wrong all the time, even with the older analogue mobiles and CB radios which produced far higher field strengths.
