On Friday 18 July 2003 13:30, Jim Wilson wrote: > Lee Elliott <[EMAIL PROTECTED]> said: > > > > The autopilot altitude hold seems a bit messed up though since yesterday - > > quite bad porpoising - I checked that with the 747, a4, c-172 and a couple of > > mine. Terrain following still seems ok though. I doubt this is due to the > > replay stuff. > > > > Lee, > > It is suprising that you are finding problems with the altitude hold, that > really hasn't changed. You aren't using throttle control as well are you? > > In anycase with the 747, I just comitted some changes to the 747 autopilot > config. This addressed some problems that have been there a while. > > As for the others. I'm not seeing a problem with the c172p at all. The a4 > does seem to have a problem. > > One issue that I looked into has to do with the frame rate frequency of > update. At first I thought there would be an impact with the > elevator-adj-factor which essentially is a gain/output adjustment (lower > values increase elevator movement). There is also an integral-contribution > value that has the effect of drawing the AP closer to its target altitude. It > occurred to me that the integral value will probably offset any frame rate > related problem with the gain so long as they were balanced. > > If you adjust the elevator-adj-factor so that the porposing goes away (try > lowering it in 25% steps) and tweak the integral so that the aircraft hits the > target (increase in very small steps). You can make these adjustments during > flight using the property browser. I'd be interested in see what you come up > with for numbers. Note that the integral-contribution should generally be > between 0.01 and 0.002, otherwise it is probably is introducing its own > problems. Also try varying the throttle to see how the AP settings perform > (don't stall though :-)). > > Again, let me know what you get for numbers on one or two AC, and also note > the frame rate. > > Thanks, > > Jim
Hello Jim, Sorry, I missed this post at first. The porpoising is certainly related to the elev_adj_factor - the lower it is the worse the problem. It's hardly apparent with the c-172 or 747 but more so with the a4. With the YF-23 it's very obvious. Auto-throttle isn't the cause of the problem - it never used to be a problem and it happens with static throttle settings anyway. Reducing the elev_adj_factor doesn't fix the problem but just seems to reduce the frequency of the oscillation - have another look at the C-172 but fly it for several minutes in still air after you've reached the target alt - it'll still dive and climb around the target alt, albeit very slowly. If I used the same elev_adj_factor for the YF-23 as for the C-172 I think I'd probably get the same period of oscillation but the amplitude would be greater - as it is now, it's diving and climbing about 200ft either side of the target alt, with a period of about 2-3 seconds. Used to be flat and steady. I'll be honest and admit that I don't know what the integral-contribution does, so I don't mess with it. I think I've got to say that the bottom line is that this has not been a problem until now and as far as I could tell from a quick test, the terrain follow fucntion still seems to be as it was before. Apart from my ignorance, I'd not dispute the inclusion of new or improved features but they've got to be implemented in such a way that they don't break existing things. How difficult would it be to have both a 'current' autopilot and a development/experimental version, selectable via an option or, more preferably, through the property tree, so that it could be changed in flight? LeeE _______________________________________________ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
