1. For downloading SimTools plugins you need a Download Package. Get it with virtual coins that you receive for forum activity or Buy Download Package - We have a zero Spam tolerance so read our forum rules first.

    Buy Now a Download Plan!
  2. Do not try to cheat our system and do not post an unnecessary amount of useless posts only to earn credits here. We have a zero spam tolerance policy and this will cause a ban of your user account. Otherwise we wish you a pleasant stay here! Read the forum rules
  3. We have a few rules which you need to read and accept before posting anything here! Following these rules will keep the forum clean and your stay pleasant. Do not follow these rules can lead to permanent exclusion from this website: Read the forum rules.
    Are you a company? Read our company rules

[theory] Simulating G-forces with motion sim

Discussion in 'DIY Motion Simulator Projects' started by raidho36, Nov 22, 2014.

  1. raidho36

    raidho36 Member

    Joined:
    Nov 18, 2014
    Messages:
    49
    Balance:
    798Coins
    Ratings:
    +14 / 6 / -0
    It is notorious that actually simulating G-forces with motion sim is challenging and besides small handful of very specific cases, simulation could never be fully accurate. But to some extent it's possible to do it and it doesn't takes 6DOF super-expensive huge rigs.

    There are few types of common inexpensive builds. Swivel-chair-like are typically having pivot point way below the chair and are typically used to only simulate tilt. They can additionally simulate small horizontal accelerations by tilting. We shall call this "acceleration via earth gravity" type of G-force simulation. As long as angular speeds are small it actually works well. But suppose you want to simulate surge of accelerating car, including abrupt acceleration drops upon gear change. In this case, the chair would have to rapidly tilt. But since pivot point is below centre of the mass, this tilting creates a force in an opposite direction. This would be especially striking discrepancy between expected feedback and resulting feedback when you shift up - you would expect the chair to take off your back, but instead it pushes you into it even harder. It makes up for intense experience and is cheap, too, but it's very inaccurate in terms of representing forces. Perceived acceleration cuvre looks like this:
    swivelchair.png
    Another common type of inexpensive sim is gymbal-like system, called "joyrider", usually 2DOF. Major difference forces-wise is that centre of mass lies close to pivot point. This doesn't creates counter-acceleration (at the centre point) during rotation, so the response curve is a lot smoother. You can also get away with a lot weaker motors if you manage to get centre of the mass really close to the pivot point. Note that perveived acceleration line is slightly curved towards greater acceleration (angle) during rotation at the same angular velocity, this is because tilt-induced sense of acceleration is a sine function of the tilt angle:
    joyrider.png
    Some sims use traction loss simulation, which is by nature the same thing as with 6DOF motion simulators that actually use linear acceleration, only made in cheaper fashion, and it includes yawwing but from standpoint of acceleration it doesn't pays any role. This paragraph discusses both. Physically accelerating the seat linearry is of course the best way to simulate linear forces. But travel magnitude is very limited. So it can nicely simulate short bursts of forces but when it comes to continious force it has to wash it out and stop the rig, which causes a bit of reverse acceleration, usually also returning then to the central position. Its acceleration curve looks like this:
    sixdof.png
    And then there's combination of gravity-induced sense of acceleration and actual linear acceleration. This is usually only found in expensive 4-6DOF motion sims, but it's also acheivable in pendulum-type "joyrider" which is a lot favorable option to an enthusiast next to $50k equipment. The only difference between pendulum-like and gymbal-like joyriders is that former deploys centre of mass heavily set off the pivot point, so that the frame acts as pendulum. And by contrast with regular joyrider, that build requires powerful motors, by formula of:

    torque required * gear ratio = full mass with driver (kg) * arm length (m) * desired acceleration (m/s²)

    where gear ratio must be low enough to allow for high accelerations at high rpm so that there isn't acceleration drops due to velocity limit. With gear ratio 1:20, mass of 100 kg, arm length of 0.5 meters, and desired acceleration of slightly above Earth's gravity, you would want at least 2000 rpm motor with 25 Nm of torque. A 4.5 horsepower winch motor would suffice, but then 0.5 meter arm length is probably too short, and that's a matter of consideraiton: short arm significantly reduces arm length to the head and accordingly reduces vestibulatory perceived acceleration, and pendulum swings faster, but considerably less torque is required and there's less centrifugal parasitic force (since centripetal acceleration is ω²*r so obviously having smaller radius helps). Combining linear acceleration with tilt in order to acheive continious sense of acceleration imposes additional variable: velocity. And it's by nature a parasitic variable because due to travel limitations, this velocity has to be dampened - by creating counter-acceleration. So there's a challenge of damping this parasytic velocity without damaging overall experience. You can come up with multitude of damping functions, simpliest of which is reducing acceleration rate proportionally to current velocity, so that as the rig builds up the speed, small degree of counter-acceleration is automatically aplied. Acceleration graph for combination of the two approaches looks as following (with pendulum specifically but 6DOF is no much different since it has to perform all the same things):
    pendulum.png
    That approach of course has drawbacks as well, as you can see the force has significant drop in the middle of rig travel because since the rig has to apply linear acceleration, the seat builds up the speed, and then the rig has to damp it, which creates some counter-acceleraiton, i.e. perceived acceleration drop. But as it was mentioned, for a motion simulator rig it's practically impossible to simulate all the forces perfectly correctly, you'd have to settle for "better than nothing", or in this case "better than other options".

    So that's regarding how can (or can not) simulate G-force with motion simulator rigs. Note that for all of above figures, the chair G-force is not actually directly equals to car G-force, but it represents it in the most accurate way it's acheivable. It is implied that forces are way under 1G, because simulating force of over 1G requires using centrifuge, but even then it also imposes great deal of other challenges without resoving which it's simply doesn't improves anything, cranking up the rig price by order of magnitude.

    Hope that helps. Sorry if I'm inventing the wheels here.
    • Like Like x 4
    Last edited: Nov 22, 2014
  2. Ads Master

    Ads Master

    Balance:
    Coins
    Ratings:
    +0 / 0 / -0
  3. bsft

    bsft

    Balance:
    Coins
    Ratings:
    +0 / 0 / -0
    interesting info there.
    keep-calm-and-open-up-another-can-of-worms.png
    More or less its been discussed before
    I have spent time behind the wheel of a street based race vehicle and time in the passenger seat of rally and circuit cars.
    Yes, what we do is simulation, not actually try to recreate real forces.
    we use mind tricks of correctly set profiles in motion to suit each persons rig.
    Granted I only have boring 2 DOF pivot, but several race car drivers and go kart professional drivers think I have got the "simulation" feeling about spot on. Going over jumps in game , the pitch of the seat makes you "prepare" yourself for the impact. Going over in a roll, there is a split second when you think you are going to do the same. Including that feeling of being "forced" back into the seat when you stand on the juice, and tensing slightly in your arms when you are about to hit the brakes.
    Anticipating that wall hit, you tend to tense up.
    I will now bring several others into this as they have different rigs and will comment about how theirs work
    @eaorobbie , you are the king we learned from.
    @BlazinH , you have a full lift frame
    @Pit and @Nick Moxley , you have 2 DOF with traction loss
    @noorbeast , you have 2 DOF with rift
  4. raidho36

    raidho36 Member

    Joined:
    Nov 18, 2014
    Messages:
    49
    Balance:
    798Coins
    Ratings:
    +14 / 6 / -0
    @bsft Thanks for the input. It's always helps to hear opinion of experienced people. You are right, it takes somewhat significant forces to kick in before you can really tell striking difference between simple simulation and real life-like experience. If you only simulate 0.1G worth of acceleration then there's probably hardly any telling in difference at all.

    I've been riding a 401 motion sim a bunch of times and that thing is really badass (and costs to match), although only 4DOF (pitch, roll, yaw and heave) it does great job simulating not only tilt but also forces. I've been driving some kind of formula car on it, can't remember - it was way before I got into sim racing, but the title I beleive was rFactor2. Granted, it can only tilt so much so I think it's somewhat 0.3G worth of acceleration is possible to simulate with it, but it was going to fullest extent. It's ability to drive itself vertically has great impact on how forces through tilting are perceived. It can simulate bump on the front, on the back, on the side, etc., plus when it tilts to simulate acceleration it rotates it in a way that pivot point is close to centre of mass of the driver, and it makes great deal of difference compared to rigs with pivot under the seat.
    • Informative Informative x 1
    Last edited: Nov 22, 2014
  5. Pit

    Pit - - - - - - - - - - - - - - - - Gold Contributor

    Joined:
    Oct 2, 2013
    Messages:
    3,029
    Location:
    Switzerland
    Balance:
    29,901Coins
    Ratings:
    +3,046 / 30 / -0
    My Motion Simulator:
    DC motor, Arduino, 6DOF
    Very interesting discussion here! @raidho36 meets the ground, but what do we blend out is the capability of our imagination (not only fooling the brain) and our learning ability to adapt the "new" impressions. Many of our sims can simulate 1:1 the shaking of the body in a car, but of course cannot simulate g-forces. Even the Toyota simulator has at some point an end and will be a simulator and nothing else.

    What I would associate real car driving vs. simulation: (I am driving a real sports car, so I may experience what bumps on track could cause):

    - body "shaking" caused by bumps, potholes, curbs, textures of the track is practically 1:1, provided that the sim is perfectly adjusted and interrelated (no weak motors, bad settings ex.), ergo feedback from the shock absorber is very well.
    - g-forces could to be "realized" by "pressing" something to the body ("fooling the brain"), but will never reached (ex. chair G-force). Perhaps a 2DOF + TL + chair G-force would be an augmentation of immersion close to the reality.
    - the heaviness of a car (and IMO a very clear feedback of the sim) is coming from the traction loss. A very good example is the Mercedes SLS GT3 in AC. You feel the ponderous car as it would be real, if (!) FFB of the wheel and traction loss interrelate. This is the most funny discovery not known before (for me).
    - 3D do the rest...

    My conclusion: the whole scientific discussion with all arguments is true and great and has right, but the "whoos" and "wows" and "huus" and "creeeeeks" can nobody explain, if some of my friends are testing out the simulator, most of them after short time are stopping having a white face and are going to ask for the toilet (same on track I have seen after 20km Nordschleife people defooding) ...so I may ask myself, if the sim would not be "real", it does not matter really :) FUN is guaranteed.

    PS: from CXC website c&p:

    ...the driver does experience high sustained g-forces at times, but more often the primary sensations experienced in a racecar are unrelenting rapid-transition g-forces, one after another - accelerate, decelerate, turn this way, that way - and these simultaneously combined with more subtle sensations of ever-changing track surfaces, bumps, curbs, even contact with other cars.
    ...
    What's more, that perception of these loads relied more on speed than length - meaning short, quick movements of no more than four or five inches were the most effective way to trick the brain. The most realistic results came when we moved only the seat, thereby significantly reducing mass and enabling very fast acceleration in all directions.
    • Informative Informative x 2
    • Like Like x 1
    Last edited: Nov 22, 2014
  6. bsft

    bsft

    Balance:
    Coins
    Ratings:
    +0 / 0 / -0
    I take my dual racers to car shows and parties. I just get a "thats awesome!, I can really feel how the car moves."
    6 professional go kart racers and one former speedway racer used them recently.
    They think the re-creation of motion is excellent. Even in Dirt Showdown.
    Thats enough evidence for me to know I am on track.
    • Like Like x 1
    • Agree Agree x 1
  7. bsft

    bsft

    Balance:
    Coins
    Ratings:
    +0 / 0 / -0
    Banana? can I have that in plain language please....
  8. Pit

    Pit - - - - - - - - - - - - - - - - Gold Contributor

    Joined:
    Oct 2, 2013
    Messages:
    3,029
    Location:
    Switzerland
    Balance:
    29,901Coins
    Ratings:
    +3,046 / 30 / -0
    My Motion Simulator:
    DC motor, Arduino, 6DOF
    banana?
  9. BlazinH

    BlazinH Well-Known Member

    Joined:
    Oct 19, 2013
    Messages:
    2,168
    Location:
    Oklahoma City, USA
    Balance:
    16,045Coins
    Ratings:
    +1,833 / 32 / -1
    bsft no comprende @Pit. ;) Or maybe he just smoked one to many of them today! :D
    • Funny Funny x 1
    Last edited: Nov 23, 2014
  10. bsft

    bsft

    Balance:
    Coins
    Ratings:
    +0 / 0 / -0
    ......cough..........
    banana.jpg
    Actually its more information than I can comprehend there @Pit ....... mainly the last part from the CXC website. Too many details for me, but it makes others happy, right @raidho36 ?
    • Funny Funny x 1
    • Friendly Friendly x 1
  11. BlazinH

    BlazinH Well-Known Member

    Joined:
    Oct 19, 2013
    Messages:
    2,168
    Location:
    Oklahoma City, USA
    Balance:
    16,045Coins
    Ratings:
    +1,833 / 32 / -1
    @raidho36, Theory is a good place to start but eventually you just have to go for it and build something so you can try it for yourself and see if it immerses you as you intended. Or have you already done so?
    • Agree Agree x 3
  12. raidho36

    raidho36 Member

    Joined:
    Nov 18, 2014
    Messages:
    49
    Balance:
    798Coins
    Ratings:
    +14 / 6 / -0
    @BlazinH I'm thinking of building a sim, right now I'm digging into the topic, trying to cover it as broadly as it possible so that I don't miss anything critical. O.P. is my findings about simulating forces - majority of sims can simulate tilt and car jerking allright so that's pretty much out of quesiton. So far the viable option is the common swivel-chair type because it doens't takes much space in the room, time to build, and horsepower (and electricity bill length) to drive. But then I enjoy NASCAR oval racing no less than road racing, and oval does includes lengthy forces on regular basis - turns take between 1/5th and 2/3rds of the track (usually about a half) and there are continious centrifugal forces during turns, even though track banking reduces them somewhat. So I also thinking of ways to build a sim that would simulate forces somewhat, it would of course do it by tilting and I'm figuring out how do I do it nicely. Swiveling chair usually does the trick, but in OP I've highlighted some issues with it, and so alternatively I'm thinking of building a pendulum joyrider, but then I'd need something like 20 kW worth of power bare minimum to drive all that the way I want it.

    Theory is extremely important if you don't want to learn how to do things right by doing them wrong. And even then, you can only understand so much about things by surfacial observations without knowing what's going on under the hood.

    @Pit Yes, from that article, having to apply less force to gain momentum is of course helps, especially if you're limited on horsepower. It's also true that major sense of the car comes from abrupt short motions, not from continious forces. In this regard it's worth mentioning that vestibular apparatus doesn't detects angular velocity, only angular acceleration, so if the car rapidly rotates it's the same as if it was going straight - much like if it's going straight without accelerating it's the same as if it was perfectly still. Rapid jerking create a lot of acceleration and sense of forces immediately kicks in.

    I have to disagree though with point about fooling the brain and inability to explain: VR showed that having imperfect solutions that are "still fun" and learing to enjoy what you got doesn't work as good, that yes you do have to fool the brain, and in order to do it you do have to explain every little thing that makes it work. In order for VR HMD to fool your brain it takes low latency, high headtracking accuracy (1:1 obviously) and fine-calibrated optics. It has to do with physics and physics has perfect explanation for such things. Similarry, simulating motion on a motion sim has to do with physics, and therefore is perfectly explainable - you just have to find out exactly how it is explained. And in this regard, it's all boils down to a very simple thing: forces. Which is both an obvious thing and a striking revelation. Every motion of a car creates forces and if you can recreate it properly, you can simulate it properly, fooling your brain to beleiving that you're actually driving that car (wearing a VR HMD would also help of course). Bump is when the car creates force upwards, tilt is when gravity downforce is off perfectly vertical line (relative to your body), traction loss is when the car goes sideways, and sway, heave and surge is simply any non-zero forces added up on top of everything else. All forces are nicely represented as vectors, and then they add up together reducing to a single superposition force-vector that acts on the body. If your motion sim can simulate this force-vector, then it can simulate pretty much every type of force, every type of car motion. That still leaves simulating rotations an open question, but as opposed to forces which are usually big, rotational accelerations in a car (let alone an airplane) are typically very small. From my experience with 401 rig that did yaw, I could tell that this yaw specifically didn't do anything even when I spin the car, whereas tilting did created a lot of sense of forces, road texture and car tilt.
    Last edited: Nov 23, 2014
  13. Pit

    Pit - - - - - - - - - - - - - - - - Gold Contributor

    Joined:
    Oct 2, 2013
    Messages:
    3,029
    Location:
    Switzerland
    Balance:
    29,901Coins
    Ratings:
    +3,046 / 30 / -0
    My Motion Simulator:
    DC motor, Arduino, 6DOF
    I was exactly thinking about that, a joyrider would be the best choice for a nascar race car. But the 20kw are nonsense...a joyrider does not pull the whole weight, the pivots reduce the weight significantly. My rig is very heavy but traction loss do not need much power, the rig can be moved by one small finger...
    That's how it is :)

    IMO not only VR HMD. All (seat, wheel, pedals, visuals) should (or better have) to be working in real time. If not so motion sickness (nausea) is guaranteed.

    Time for going on, don't waste time in theories, go forward and believe me if you will follow the hints, tricks, experience and suggestions of this community you will archive success. :thumbs
    • Like Like x 1
  14. raidho36

    raidho36 Member

    Joined:
    Nov 18, 2014
    Messages:
    49
    Balance:
    798Coins
    Ratings:
    +14 / 6 / -0
    @Pit thanks, I will! :) Although it's still over a week until I get home so I'm not really wasting time, more like putting it to constructive use.

    >But the 20kw are nonsense
    Yeah... But that's the math - 0.5m arm and 100kg of weight is almost 500 Nm of torque, and at 60 rpm of the rig frame is about 3kW but realistically I'd need more than that, and base figures will likely to be bigger, and then there's two such motors, plus heave drive. Joyrider in doesn't takes a whole lot of force yes, but I'm modifying it by taking the centre of masses far off the pivot point so that the rig acts as a pendulum.
  15. Pit

    Pit - - - - - - - - - - - - - - - - Gold Contributor

    Joined:
    Oct 2, 2013
    Messages:
    3,029
    Location:
    Switzerland
    Balance:
    29,901Coins
    Ratings:
    +3,046 / 30 / -0
    My Motion Simulator:
    DC motor, Arduino, 6DOF
    I comprehend. Obviously your are knowing what to do but IMHO you should be "warned" - to take the center of masses far off the pivot point could result in a very inertial mass. You will loose all the fast rapid-transition g-forces and by association the immersion (PS: not if using veeeeeery fast and strong motors, IMO surely can not be succeed using 12v). Perhaps @value1 (huhu :sos) can comment more details by his experience on this matter?
  16. raidho36

    raidho36 Member

    Joined:
    Nov 18, 2014
    Messages:
    49
    Balance:
    798Coins
    Ratings:
    +14 / 6 / -0
    @Pit
    > you should be "warned" - to take the center of masses far off the pivot point could result in a very inertial mass
    Yeah I know that already, the formulas and all. Veeery strong motors. Speed is not really a requirement, just raw power output which is indifferent from motor speed. I dunno about 12V but I think I've seen a 3.5 kW 24V winch motors on amazon. You can also use several weaker motors driving the same gear shaft, but that doesn't reduce total power consumption of course.
  17. Pit

    Pit - - - - - - - - - - - - - - - - Gold Contributor

    Joined:
    Oct 2, 2013
    Messages:
    3,029
    Location:
    Switzerland
    Balance:
    29,901Coins
    Ratings:
    +3,046 / 30 / -0
    My Motion Simulator:
    DC motor, Arduino, 6DOF
    I am curious and I hope you will get succeed :)
  18. RacingMat

    RacingMat Well-Known Member Gold Contributor

    Joined:
    Feb 22, 2013
    Messages:
    2,094
    Location:
    Marseille - FRANCE
    Balance:
    19,098Coins
    Ratings:
    +1,974 / 20 / -2
    My Motion Simulator:
    2DOF, DC motor, Arduino
    Hi!
    what kind of driver do you intend to use for such powerfull DC motors?
  19. Pit

    Pit - - - - - - - - - - - - - - - - Gold Contributor

    Joined:
    Oct 2, 2013
    Messages:
    3,029
    Location:
    Switzerland
    Balance:
    29,901Coins
    Ratings:
    +3,046 / 30 / -0
    My Motion Simulator:
    DC motor, Arduino, 6DOF
    :popcorn

    Sorry, but I has been forced to do so :)
  20. raidho36

    raidho36 Member

    Joined:
    Nov 18, 2014
    Messages:
    49
    Balance:
    798Coins
    Ratings:
    +14 / 6 / -0
    @RacingMat yep that's the problem. :D I do have trouble figuring out how to drive that much power. Looks like my only option is soldering a custom super-high-power driver. I'm still looking into ways of reducing required power without compromising too much on experience.
  21. Pit

    Pit - - - - - - - - - - - - - - - - Gold Contributor

    Joined:
    Oct 2, 2013
    Messages:
    3,029
    Location:
    Switzerland
    Balance:
    29,901Coins
    Ratings:
    +3,046 / 30 / -0
    My Motion Simulator:
    DC motor, Arduino, 6DOF
    ...and where to find so powerfull PSUs...if you will realize once upon a time this joyrider 12V would be a mess. A 12V motor would draw twice as much current from its 12V supply compared to the 24V motor on a 24V supply (there exist some server psu solutions 12V -> 24V at the well known rc forums)