Autopilots... failure modes

"Brenor Brophy" wrote
Here is a link to a very good article on autopilots on the Avionics West web
page. I think it answers a lot of your questions.

http://avionicswest.com/articles/kno..._autopilot.htm

That's not a bad article. I do have some issues with it.

First off, it ignores the most important difference between rate-based
and attitude-based autopilot - performance in turbulence in a slick
airplane. Attitude-based autopilots (those that use pickoffs on the
horizon gyro) work all the time. Rate-based autopilots (those that
use pickoffs on the turn coordinator) work in smooth air or with
light, draggy airplanes. In a heavier slicker airplane, especially a
twin, turbulence makes for a very uncomfortable ride because all the
corrections are too much too late. There are NO rate-based autopilots
in the transport category - they just wouldn't work.

The whole system failure issue is more complex that it seems. The
reality is that ALL autopilots built for GA use can roll you over in a
heartbeat - every one of them without exception. Vacuum failure is
only one mechanism for this - it causes the gyro to provide incorrect
bank information. Here are some others:

Bad connection. Really. A bad conenction to the attitude indicator
(AI) or turn coordinator (TC) means the system has no idea what the
bank angle or rate of turn is - but it thinks it does. Over you go.

Bad brushes/dirt. A TC with bad brushes can come to a halt - and not
flag out. An AI with dirt/water sucked in can stop spinning or go
nuts because of the pendulous vanes - and not flag out. Warning flags
on GA gyros are near-useless - they monitor only the power source, not
actual rotation.

Electronic failures - there is lots of circuitry required to read that
AI or TC. Any of it can fail. Over you go.

Bad computer. The modern autopilots are computer based. The sytem
crashes, and over you go.

The bottom line is that with any GA autopilot system, you as the pilot
must continuously and carefully monitor whichever gyro(s) the
autopilot is NOT using. The autopilot is only a way for you to take
your hands off the controls, not a way to stop the scan.

Michael

In article ,
(Michael) writes:

The bottom line is that with any GA autopilot system, you as the pilot
must continuously and carefully monitor whichever gyro(s) the
autopilot is NOT using. The autopilot is only a way for you to take
your hands off the controls, not a way to stop the scan.


Hip, Hip, Hurray!!

Chuck

"Michael" wrote in message
om...

http://avionicswest.com/articles/kno..._autopilot.htm

That's not a bad article. I do have some issues with it.

First off, it ignores the most important difference between rate-based
and attitude-based autopilot - performance in turbulence in a slick
airplane. Attitude-based autopilots (those that use pickoffs on the
horizon gyro) work all the time. Rate-based autopilots (those that
use pickoffs on the turn coordinator) work in smooth air or with
light, draggy airplanes. In a heavier slicker airplane, especially a
twin, turbulence makes for a very uncomfortable ride because all the
corrections are too much too late. There are NO rate-based autopilots
in the transport category - they just wouldn't work.

You might want to take a gander at this article. I'm not sure what you mean
by slow corrections, but evidently S-TEC (taken with a grain of salt) has
optimized the corrections, as well as smoothed them to avoid overstressing
the aircraft in turbulence.

See Pages 6 & 7.

http://www.s-tec.com/pdf/autopilotbook.pdf


Matt
---------------------
Matthew W. Barrow
Site-Fill Homes, LLC.
Montrose, CO

Matt,

You might want to take a gander at this article. I'm not sure what you
mean
by slow corrections, but evidently S-TEC (taken with a grain of salt) has
optimized the corrections, as well as smoothed them to avoid overstressing
the aircraft in turbulence.

See Pages 6 & 7.

http://www.s-tec.com/pdf/autopilotbook.pdf

....but the S-Tec related Meggit line of products just offerd an autopilot
aimed at the multi/turboprop market, and it's attitude(position) based.

Seems the realized that a rate-based auto is not on par with the airplane
when installed in, e.g, a fast single-engine turboprop, just as Michael
said.

Regards

Kai