Linux, Avionics, and spins

[Rich Braun]

Jerry wrote:
> But still has the older instrumentation with needles. That does mean that
> they may be able to determine airspeed and other information showing on the
> instruments if the salt water did not erase the traces.

If avionics had gone the direction of automotive electronics over the
last five or ten years, there would be data recorders tucked into every
plane built in that amount of time which keep fairly long history of
a variety of environmental conditions, in flash memory where it could
be retrieved after a crash.

Starting with the 1997 or 1998 model year, some automakers have been building
such gear into their designs.  It's privacy-rights issue for some, so there
haven't been any decisions yet on who gets access to the info beyond the
automakers for purposes of crash-analysis research.  For planes, though,
the privacy issue is pretty much moot--pilots and manufacturers alike would
want this data.

As with cars, of course, instruments would still have needles.  Carmakers
years ago tried the all-digital instrument panel, and most consumers
rejected it.  Pilots are an even more conservative lot, so I wouldn't
expect many visible changes in the instrumentation one sees on the console.

The toys-and-trinkets items I'm talking about are primarily in functions
under the hood.  Imagine having a radio which could actually get you the
exact info you need on approach, without all the distractions and say-agains.
Doing a lot of the same old thing but with digital technology would make
most things work better.  Until it could be done on a large enough scale,
though, it'll wind up being expensive and not as bullet-proof as needed.
So most of the prototype stuff out there has essentially been junk,
worthy of the usual criticisms you hear about unreliability or vulnerability
to power glitches or whatnot.

> Being an aviator
> from the old school, I would prefer to have a pressure airspeed indicator,
> an alcohol inclined plane ball (eg needle and ball), a pure barometric
> altimiter, and a barometric rate of descent indicator. While these
> instruments often lie, they are reliable and predictable. And, not to
> forget, a real magnetic compass. I have experienced a complete in-flight
> electrical failure (at night over water).

Having all those things is very useful, I'll agree.  But I think current
state of the art could do much better with power backup and fault tolerance
than anything which has been introduced into aviation; automobiles have
become far more reliable lately and you don't hear about complete electrical
failures any more.

> Light planes do not normally have a dual set
> of instruments, so the older mechanical instruments are better. The newer
> digital instruments, while much more accurate, can fail.   

Indeed.  You've helped make my point, which is that the new stuff could be
made far better if it got some decent engineering attention and if it
were produced in enough volumes to bring costs in line.

I've been working with Internet technology for quite some time, and have
seen it go from glitching several times a day to a carrier-class environment
where outages are rare.  Fault tolerance follows the volume curve.

-rich
-
Subcription/unsubscription/info requests: send e-mail with
"subscribe", "unsubscribe", or "info" on the first line of the
message body to discuss-request at blu.org (Subject line is ignored).