Unstable autothrottle in FS2004 ?

[rcbarend 37]

With all of the default jets in FS2004 (737, 747, 777) I experiance the following "problem":When I fly on autopilot/autotrottle in a stable flight (e.g. at 5000 ft, 250 knots IAS, no flaps, no wind) the throttles are constantly fluctuating over 20-30% of the throttle range, in a 5-seconds cycle.Altitude is very stable, as set in the AP.And the speed constantly varies within a few knots.This throttle behaviour seems not very realistic.It looks like MS has changed the autothottle control algoritme in FS2004, because in FS2002 this worked perfectly.Does anyone else notice this ?Is there anyway this can be "tweaked" ?Rob

[Orlaam 306]

I've never had any problems with the autothrottle on the default 737, however I never really use the other two so can't comment on those.Not sure what could be going on. Does TOGA work fine?

[jcboliveira 1]

In the aircraft.cfg you will find autothrottle_available=1autothrottle_arming_required=1autothrottle_max_rpm = 90autothrottle_takeoff_ga=1gs_proportional_control=25.0gs_integrator_control=0.53gs_derivative_control=0.00gs_integrator_boundary=0.70gs_derivative_boundary=0.00Try to tweak the last 5 values and see if there is any change. Of course this isn't the method to do things. I'm beggining to see what is this but have to write an application to measure values from aircraft. At this point don't really know how the PID controllers affect aircraft.Jos

[rcbarend 37]

Hi Jose,After some further testing I discovered the following:This behaviour of the throttle when the AT is enabled (I'll call it oscillation) is affected by the setting:max_throttle_rate=0.03 // Default: 0.10In fact this parameter determines the rate of adaptations (= number of adaptatations) per second of the throttles. With the default value of 0.10 the adaptation rate is so fast, that it continuously under-/over compensates. Using a value of 0.03 gives a much better result, though I still wonder why it should "oscillate" when the targetspeed is reached.A disadvantage of decreasing this value, is that the AT also reacts much slower when it SHOULD react fast.Eg. observe when you use the AT/speedhold at takeoff (don't know if this is a correct procedure, but it just illustrates the effect).Anyway, MS definitely changed the control algoritme of the AT (just compare it with FS2002 !), and IMHO made it less realistic.Either intentionally or by mistake (bug :-) ).Although I must admit that it is quite difficult to make a fool-proof and robust algoritme to control speed (which I discovered when I made my implementation of the Taxispeed gauge).Regards, Rob