PID controllers / autopilot stability

Post by **Garry Russell** » 26 Oct 2006, 23:32

speedbird591 wrote:
Garry Russell wrote::k:

Garry
Garry's just worried whether it's in the correct forum

Or indeed the correct website :roll:

Garry

speedbird591 · Post by **speedbird591** » 26 Oct 2006, 23:40

TobyV wrote:If the likes of DM and RP understand it, you have nothing to worry about

Yes - it's like religion. As long as the priests understand it, the rest just need faith

Ian

TobyV · Post by **TobyV** » 26 Oct 2006, 23:47

Garry Russell wrote:
speedbird591 wrote:
Garry Russell wrote::k:

Garry
Garry's just worried whether it's in the correct forum
Or indeed the correct website :roll:

Garry

To use DM's example of the second graph, it might be equavilent to you asking your aircraft's a/p to hold 8,000ft whist in the climb but when you reach it it continues climing to 9,000ft, then realises its way over its altitude and decends rapidly to 7,000ft then ascends rapidly to 8,500ft, then descends to 7,500ft, then ascends to 8,250ft etc etc. Understanding what the affect of these variables in the aircaft.cfg file will help to alleviate this problem

Edit - Theres a Wikipedia article on the subject here

Post by **MALTBY D** » 27 Oct 2006, 00:56

Interesting about the Tinmouse settings Ian.

I copied those into the Trident & ran my autoland tester saved flight with it.
It did indeed capture the localizer beam cleaner & I was convinced this was the magic ingredient.
However the autoland test has a 36kt side wind & it didn't like that at all, eventually drifting so far off line that it nearly took the roof off The Airport pub (Manchester).
Ok, 36kt is a serious side wind & more than the real Trident was allowed to land with, but the Tinmouse settings failed at 24kt too, which was allowed.

From what I can remember, the inability to correct the drift will be because the nav_integrator_boundary is only 0.5 (2.5 on the Trident)
That makes the Trident more coarse, but gives it the ability to cope better with a rough ride.
It would be a pity, but I suspect that may be all it is. Those derivative settings on their own didn't seem to help the Trident any.

DM

Post by **DaveB** » 27 Oct 2006, 01:00

I'm sure it shouldn't be but this is one of the most amusing threads I've read in a long time (ignorance is bliss)

Put like that Toby (in laymans English).. that makes total sense. Fortunately, neither DM's or Ricks models behaive? in this way. There are other forces (within flightsim) that can make an aircraft slow to gain an ILS and anyone who bought the Alphasim Valiant will know that they missed the point totally

ATB

DaveB :tab:

Post by **Garry Russell** » 27 Oct 2006, 01:52

I do love it when engineers explain things...to make them so much clearer.......... of course :think: :k: :redface:

http://www.micom.net/oops/USAF%20Training.wav

Garry

speedbird591 · Post by **speedbird591** » 27 Oct 2006, 08:54

I think I must be getting more cleverer than wot I thought I was. I read Toby's stuff again (and again and again) and I think I'm ready to do the test!

As for Garry's brilliant missile guidance explanation: when I finished laughing, I realised that that made perfect sense too. You just program the missile with all the co-ordinates of everywhere in the World that you don't want it to be, leaving one little Osama-bin-Laden-sized space unaccounted for. And being the obedient little missile that it is, it goes straight there because it's not allowed to be anywhere else.

OK - I'm not sure that's the way I would have done it, but perhaps that's why I have to spend Saturday nights arguing with drunks about train tickets

Ian

DanKH · Post by **DanKH** » 27 Oct 2006, 10:02

Brilliant little article there Toby, some good considerations there to do if and when you design aircrafts....but as you say yourself, it all depends on Microsoft coding the "other things" correctly.... well by the looks it may seem so, or at least they have come close to it ;-)

Aerodynamics and electronics has so many variables that it takes more than a "simple" software developer to code a good simulator....

In my humble opinion i think that Microsoft is on the right track by defining FS as a platform instead of a full-scale simulator. More of these insights may be good to have in mind, when and if Microsoft decides to go the whole way, and making a FS-version for developers, like they have done Visual Studio....Hey they might even incorporate "FS-Sharp" as a new programming language into the Visual Studio ;-)

Post by **Garry Russell** » 27 Oct 2006, 11:40

Very true

Tobys article makes perfect sense if you read it and study it

I just couldn't resist the temptaion.

It's simply a case of getting on the logic train and it all falls into place

Garry

AlexP · Post by **AlexP** » 27 Oct 2006, 11:46

Garry Wrote

It's simply a case of getting on the logic train and it all falls into place

Trains!?! I thought this all related to aircraft. Now I'm really confused!!! :madhead:

Alex