(As seen through the eyes of a Feedback Control System Engineer)

Written by E. Bruce Baker

Over the last several years, there has been much discussion about the need for motion bases for aircraft simulators. Several times, an attempt has been made to prove or disprove the need for a motion base, and indeed, the need has been both proved and disproved. From all these studies and from personal experience, a few fundamental truths have emerged:

1. Air combat does not require and motion base
2. Hovercraft simulation (Helicopters VSTUL) requires a motion base
3. Map of the Earth flight, particularly under IFR conditions, using a system such as the AH-64 PNVS, requires a motion base
4. Manual terrain following (200’ altitude, 300 knots) requires a motion base
5. Carrier landings require a motion base
6. Realistic pilot response to wind gusts require a motion base

In general, any pilot task with requires frequent, rapid control inputs requires a motion base.

And analytical approach can be taken to understand exactly what a motion base is doing as far as the pilot is concerned. Figure 1 shows a block diagram of the pilot’s roll control loop for a real aircraft. The pilot gets cues from the aircraft which tells him what the aircraft is doing. The roll acceleration and velocity cues are primary through the “seat of the pants,” while the roll angle cues are entirely visual. Admittedly, the pilot gets some velocity cue from looking out the window, but under transient conditions, it is difficult for the pilot as he must derive velocity by mentally differentiating position. The pilot cannot get accelerations cues by looking out the window.

Figure 2 shows the pilot’s roll control loop for a simulator with motion. The roll acceleration and velocity feedback signals are still present although they have been modified by the two filters. It should be relatively obvious that the design of these filters is critical.

It has been determined experimentally that pilots normally close the roll control loop in the 1.0-1.5Hz region when they are doing a high work load control task. This means that the pilots are responding to inputs which are below 1.0-1.5Hz and are ignoring inputs above 1.5Hz. Id can be shown analytically that the roll control loop transient response will not change significantly provided the motion base filters do not appreciably change the phase of the acceleration and velocity signals in the vicinity of the cross over frequency (1.0-1.5Hz).

Figure 3 shows the phase shift for two different combinations of the motion base wash-out filters and actuator servo bandwidths. Both of these combinations were designed to provide zero hase shift at 1.0Hz. The objective is to provide a phase curve that is flat (0 degrees) in the vicinity of 1.0Hz. The objective is to provide a phase curve that is flat (0 degrees) in the vicinity of 1.0Hz. Increasing the actuator servo bandwidth and decreasing the wash-out filter frequency improves the flatness of the phase curve. The actuator servo bandwidth is normally limited by the motion base and cockpit structure. The wash-out frequency is limited by the actuator stroke since decreasing the wash-out frequency by a factor of 2 requires increasing the stroke by a factor of 4.

Figure 4 shows the pilot’s roll control loop for a simulator without motion. Note that the acceleration feedback is completely missing and any rate information must be derived by differentiating the roll position from the visual display. It can be shown analytically that unless the pilot drastically changes his compensation - i.e. the way he uses cues to control the aircraft - the roll control loop becomes unstable. Furthermore, the pilot’s gust response is completely different without motion than it is with motion.

This argument can be extended to include the other axes of the aircraft.

The following documents are actual documents from the simulation industry. The comments in the documents are from actual pilots after being in the simulator. Their comments are from before and after simulation tuning occurred. Please download the whole document here.