There is no doubt that the trim system is complex and results in many automation upset incidents. Although the trim mismanagement was not implicated in the EC155 incident, it was in Cougar 851! There is an interesting point in HC's discourse above:
If there is 40kts wind (not that uncommon!) no more follow up trim. Now you move the cyclic forward to rotate, airspeed increases so flap back requires the cyclic to be moved further forward. At say 80kts you decide to press GA and let go. The cyclic jumps back to its hover position, the aircraft rears up starts to lose speed. You then start to fight it and a Dancopter EC155 - like scenario ensues. If you operate the beep trim, you beep the IAS datum not the stick position or attitude datum. This is a dreadful way to operate.
because it appears to describe the sequence of events in the Cougar 851 incident (as we saw from the video replay at the conference): the pilot takes off from the platform using the published S92 Cat A profile but does not press the trim release trigger. As the aircraft passes through 60kts (the lower limit of the autopilot at that time was 50kts) the pilot selects Go Around (GA) and releases the controls. From that point the autopilot is programmed to accelerate to (the Vy of) 80kts, achieve a vertical speed target of 750ft/min and maintain heading. However, the trim system is of necessity (certification rules on hard-over tolerence) limited in authority and cannot overcome the trim forces that remained when the controls were released. The trim actuators
never overcome the back force and the aircraft decelerates through 50kts and the autopilot disconnects. It is likely that the deceleration and autopilot disconnect is not anticipated because of lack of knowledge/awareness and a very serious incident (resembling the EC155 incident) ensues.
(I'm sure someone will correct me if the following is not accurate)
We are presently in a situation where, in (at least) one of the latest generation aircraft there are trim releases on the cyclic and collective as well as 'coolie hat' beepers on both. These functions are 'overloaded' in the sense that they result in different actions according to the mode engaged (SAS, ATT, two-cue/three-axis coupled and three-cue/four-axis coupled); both coolie hat trims have longitudinal and lateral functions. As most who fly these aircraft will know, when the aircraft is in two-cue/three axis mode, the aircraft trajectory is controlled by pitch attitude (i.e. holding airspeed or vertical speed by changing pitch attitude) control over the active parameter is provided by the cyclic trim function (release or beeper trim). If the three-cue/four axis mode is engaged - in the simplest form the IAS and VS are both engaged (by the pilot or, under envelope protection, by the system) - control of the VS moves from the cyclic beeper to the collective beeper. Not exactly 'simples' is it!
If this does not describe the autopilot system in your current aircraft perhaps we should ask the question why - is there not a common certification criteria for all Aircraft Guidance Systems (AGS)? This brings us on to another topic covered at the conference and that is the absence of two very important rules for certification of displays, HMIs and AGS - XX.1302 and XX.1329 - and why? However, enough for now; we will no doubt move to these subjects later.
Jim