NavList:

Greg, you wrote:
"Rudder drag is a cost factor also. If the autopilot is working too hard and using too much rudder then the barrels per mile number will increase a bit. Optimal settings are probably automatic on todays modern units. Older autopilots had manual settings for rudder response and yaw allowance."

Computing cost has become so low that the traditional feedback problems can all be managed in exquisite detail in microprocessors. Of course, it's just more electronics to fail, but when it works, it REALLY works!

As an aside, it's worth mentioning that this very issue of rudder feedback loops in autopilot systems has indirectly given us one of the most common words in the modern communication lexicon: "cyberspace", a sometimes synonym for the online universe and the internet. This word cyberspace was originally coined and used in science fiction, but it was derived from the word "cybernetics" which was championed by Norbert Wiener starting in the 1940s as the scientific study of feedback systems, analog and digital as well as human social systems. And the word literally means "rudder" etymologically -- from Wikpedia: "κυβερνήτης (kybernētēs) 'steersman, governor, pilot, or rudder'." We all know the concept of a gentle hand on the rudder. You don't want to over-control, or you will find yourself fighting random noise at great expense. And in Wiener's social applications, government systems must be built to control with a "gentle hand on the rudder" for similar reasons. It's that very concept, rudder-control rendered in automated systems, that is at the heart of cybernetics.

You also wrote:
"GPS course made good seems to wander and lag the compass as I just found out crewing on a friends boat a few weeks ago. We were making anchorage in zero visibility due to fog. I was watching the radar and the helmsman was watching the GPS chart plotter. The helmsman was steering a zig zag course using the chart plotter course to steer."

From the conditions you describe, I would imagine you were moving very slowly, right? Was your helmsman watching a course made good? Or was it the course to a waypoint? If you're moving very slowly, the small random errors in the GPS position from one fix to the next can translate into noticeable changes in course even if you know your actual course is nearly straight. This is closely related to the original "cybernetics" issue above, and with the right algorithms, it can be managed. The GPS fixes are a random walk around your actual motion. Suppose for example you get a new GPS fix from your system every ten seconds (it's probably faster but let's assume some slower "polling" rate with averaging). Suppose you're puttering along in the fog at 2 knots, which is just about 1 meter per second. Now suppose the random noise in the GPS fixes is around 2 meters. If I travel ten meters in ten seconds, but my GPS fix bounces to the left by two meters, the course made good would be wrong by around 11 degrees. If I steer based on that course, I will end up over-controlling the rudder. This assumes the very simplest algorithm for handling the situation. Even older plotting systems should have some mechanism to account for this. But in simple software systems, it's un-necessary. In the simple smartphone app that I described, it will sometimes show me in motion at 1 knot in some random direction when I am quite motionless. Of course, this effect is reduced in direct proportion to the speed. The displayed course is quite steady in straight line motion for speeds above 30 mph.

-FER

----------------------------------------------------------------
NavList message boards and member settings: www.fer3.com/NavList
Members may optionally receive posts by email.
To cancel email delivery, send a message to NoMail[at]fer3.com
----------------------------------------------------------------