This was a good video about why less is more:
The above is Gamer Muscles experience and since he has a professional and "proper" English accent it must be true and so you know he know his stuff. lol. Joking aside I think he is quite accurate.
Motion reproduction is a pluralistic problem, meaning there are many software, algorithmic, electrical, mechanical and even tuning aspects involved to get it just right. Will try to summarize as much as I can.
Q: Why less motion is good?
A: Less motion is good for "car racing" because there are typically two general types of motion you are trying to replicate. REAL queues like heave, high frequency suspension, pitch and roll. These are real effects because the motion platform can replicate them exactly as in the virtual environment. Otherwise all other motion queues are EFFECTS that a 3-DOF has to 'fake' , like using roll for sway and pitch for surge, to sell us the experience and get us immersed. Sway and surge since are both EFFECT layers and not REAL layers, in a 3-DOF actuator system because they are trying to do something they can't. So for example a motion algorithm can make a heave calculation that is TRULY representative of what is happening in the software. It is a direct relationship. So 4" of vehicle heave in iRacing for example is 0.4" movement in a the actuator. EFFECTS however have no direct relationship. And so they have to be painted, canned or artificial, and some "human" has to not only translate from simulated to real but also what that simulated should feel like. So a sway moment when turning a Miata is 0.5G and we can represent that with a 5-degree roll rotation, and an F1 car is 1.5G so lets represent that with a 15-degree rotation. Now does rotating the motion chassis more, indicate more G's? No, it feels awkward, and looks something like this, or what I often refer to as the most popular motion simulator video on the internet, with 7 million views:
Actually rotating/rolling the chassis too much or too quickly (speed/acceleration and jerk moments are another long topic) can actually effect your steering. Often I turn down my roll to a low number, specially with formula cars, as the sudden jerk moment that is used to replicate lateral force or sway moves my body and involuntarily the wheel and makes for unwanted steering inputs.
So our assumptions are that more travel will give you more effect, and it does sometimes, but often times, the effect is often canned or algorithmically tied to something like sustained cornering G that just cannot be fully replicated. Drivers soon realize that leaning the chassis more to one side, to have the effect of experiencing more G is fun but not accurate to the real world. In the real world you would feel the G in your seat through a lateral/side force and not via a roll. Now in flight sims, this effect is no longer an issue because its is not an effect but a more real representation. You DO have high degrees of roll and pitch in a flight sim. Hence why people always say, you don't need 6" or more unless you are doing flight.
So, the above is just scratching the surface when it comes to motion systems design and implementation. Like I said motion is a pluralistic problem with no real solutions, just compromises. In the end, like all engineering problems, you have to ask yourself what is your goal, what do you want to achieve, then spend majority of the time redefining the problem, really understanding it from the ground up and applying what you think is the best solution to meet your needs.
Here are other considerations:
1) algorithms - how well are the simulated forces replicated by the physical actuators, do all the layers compete for stroke, like dynamic allocation OR does each layer have a defined amount of stroke (another one of my pet peeves with some motion systems).
2) signal processing - Are the algorithms running in a non-real time system like Windows or are they running in a hard real-time embedded controller chipset?
3) controllers - is the system running Arduinos, or advanced Texas Instrument PRU's with deterministic, real-time processing, direct access to I/O and with ultra low latency. This difference may be in the micro-seconds but it all adds up in the end.
4) motors - are the motors able to consistently reproduce the necessary motion, over time, and with various chassis and driver total weight/mass. Are they tested by a third party so that their power claims are verified. Lot's of fake motor graphs out there.
5) electrical - are you just using a run of the mill switching power supply from Meanwell which is often not intended for motion applications, or bulky EMI producing servos? What kind of regenerative resistors do you have and are they spec'ed for the total chassis weight/mass.
6) seat/chassis - and maybe most import, is the chassis solid enough and is the seat mounted rigid enough so that you do not lose out on all the technology above by having cheap chinese seat brackets that bend and absorb most of the motions. This is something I realized late, and would almost always recommend a rigid, fixed back and fixed mount seat, Some proper seat mounts from companies like Tillett go for $400+ but are CNC'ed and not stamped and bent from steel plates.
https://www.torqued.io/tillett-eb-f2-solid-aluminum-race-seat-mounting-bracket
7) driver location - does the sim software take into account the driver location on the motion platform and the distance of the actuators from one another to establish the best motion plane possible? meaning some people install their actuators more lengthwise than widthwise. If this number is not properly entered and calculated by the software than your degrees of roll will change with different actuator chassis mounting locations. So if an actuator is mounted 20" apart left to right but 40" from front to back (assuming a 4-post setup) then you will have a lot more ROLL then pitch in your system. Yes you can adjust that with sliders, but you are really guessing and the trig is not that complicated to calculate. ;-)
Lastly engine vibration:
Probably the best engine vibration (EV) reproduction is SimHub right now that I tried, just using the in-game sound as signal reference BUT dayton audio/buttkickers are low mass transducers, meaning they move a very small mass to try to give you a large effect, and again most important, mounting of transducers matters a lot. Mounting to aluminum, wood, composites etc, gives a different sensations each time. It's also messy with wiring and so forth but anyways the main point is that its a small mass diaphragm that moves around. Engine vibration with say 4-actuators or even a single actuator is much better in that it moves the whole chassis. The engine vibration translates throughout the whole chassis, like it would in a real car, its pretty cool. In some old V8's in iRacing like the LateModels (Oval dirt) and or the Trophy Trucks, you can not only feel the whole chassis move to the beat of the engine but also have it rotate the chassis as a longitudinally mounted engine would in a real car.
Effects are also common in EV and I think this is again something you only pick up after a few years of using motion systems. You will notice that some systems, just make a sine wave that "sounds" like a V8, I4, H6 etc... and just change the amplitude and frequency of the wave over time, this is what most people refer to as haptic, which is weird to me, but hey its marketing. Anyways, this works great at first, until iRacing releases a open wheel car that revs to 19,500 rpm. lol.
Hope that helps and stirs up this great conversation.
*** Above are my professional opinions and not financial advice ***
