1. Do not share user accounts! Any account that is shared by another person will be blocked and closed. This means: we will close not only the account that is shared, but also the main account of the user who uses another person's account. We have the ability to detect account sharing, so please do not try to cheat the system. This action will take place on 04/18/2023. Read all forum rules.
    Dismiss Notice
  2. 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!
  3. 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
  4. 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

Building innovative Stewart platform

Discussion in 'New users start here - FAQ' started by Danny Miller, Jul 10, 2024 at 09:23.

  1. Danny Miller

    Danny Miller New Member

    Joined:
    Tuesday
    Messages:
    4
    Balance:
    24Coins
    Ratings:
    +0 / 0 / -0
    My Motion Simulator:
    6DOF
    I'm a newbie at this, but that's conditional- I'm an engineer and experienced maker. I've been inventing innovative stuff left and right this year, with a lot of it being functional mechanical designs.

    I've got some ideas that would take the Stewart platform to a new level. So far it checks out on paper (well, Fusion 360 modeling, and force calcs).

    Still, I'd like a sanity check. The notable points:
    1. 800 mm dia base (well, triangle with the corners on the 800 circle)
    2. linear actuators should do 1000 mm/s with 9.6kgf nominal. Peak torque is about 150% of that.
    3. actuators are about 750 mm retracted and can extend +500mm
    4. Oculus Quest so no screen array.

    I listed a

    Here's the kicker-
    Balljoints are limited to about 17mm of roll/pitch, I couldn't see a good way to go further. So I came up with what's looking like a MUCH better design and I'm targeting 40 deg of roll/pitch which sounds like a wild ride. It's still classified as a Stewart but it's quite different.

    Now 9.6 kgf isn't that much, but I plan to use 3x air cylinders to push up on each corner and offset the weight. They're all 3 connected together so roll/pitch just moves air from one cylinder to the other. They are also connected to a small air tank which is >10x the cumulative volume of air from mid-cylinder to the stops. Reason being, if you had no extra volume but the pistons, then heaving down lower means the pressure increases, a 500 mm travel cylinder would be doubling at even half the travel 125mm) to the bottom (250mm) which would be far too much force for the actuators to fight and it wouldn't go lower, and also the pressure falls while heaving upward, again, overloading the motors.

    But the addition of a >10x shared reservoir means the system's only going to increase pressure by 5% when heaved to the bottom, or decreases 5% when they're all 3 fully extended The compressor does NOT run once the system is going, it's just providing the springing force. You could adjust pressure to offset the cockpit and rider weight during initial setup. So, ideally, the motors aren't loading with any torque at all at rest, and when they do move, the 9.6kgf*6*cos(30)=~49.88kgf for motion. Again, the whole platform is being held up by connected cylinders so the platform with rider is essentially floating in zero g so it could do about 70% of a G of acceleration.

    I haven't seen much about pneumatics, and I've been working the prob on paper for a couple of weeks.

    One of the central questions I had- the best cylinder I saw for this would be a 20mm dia at about 100 psi. The smaller dia, higher pressure is less air to have to move quickly in and out, and less tank volume.

    This raised on problem- the 20mm cylinder is only a 1/8 NPT port, and I measured a push-connect fitting as only 4.37mm ID inside of that. I wanted more, but I found an online engineering calculator and it seemed to be saying that the force is only going to drop by about 2% when extended at 100mm/s, as the air rushing in from a 100psi tank as the cylinder is allowed to extended that quickly means the pressure will drop inside the cylinder and that forces the motors to bear weight and limit its performance.

    But, like I say, the calc I found online seems to say you'd only get <5% loss in lift force when extending at 1000mm/s sec a similar effect while retracting as the cylinder pressure increases above the rider's weight.

    <5% sounds much lower than I would have thought with a 1/8 NPT orifice with a 4.37mm ID port on it. Anybody able to do a better calc? 20mm ID cylinder 500mm travel, ~100 psi, at 1000mm/s

    This has changed the design scenario a bit- the rod end joints can expect to be roughly equally in tension and compression peak loads. But always limited by the strength of the motor torque peak.
  2. noorbeast

    noorbeast VR Tassie Devil Staff Member Moderator Race Director

    Joined:
    Jul 13, 2014
    Messages:
    20,742
    Occupation:
    Innovative tech specialist for NGOs
    Location:
    St Helens, Tasmania, Australia
    Balance:
    146,304Coins
    Ratings:
    +10,813 / 52 / -2
    My Motion Simulator:
    3DOF, DC motor, JRK
    I have read through your verbal description, but would appreciate some dimensioned sketches of what you have in mind.

    Pneumatic assistance sounds interesting, but at the same time needs precision control, particularly with respect to VR use.
  3. Klaus Schmidinger

    Klaus Schmidinger Member

    Joined:
    Oct 3, 2022
    Messages:
    46
    Balance:
    346Coins
    Ratings:
    +11 / 0 / -0
    My Motion Simulator:
    AC motor, Motion platform, 6DOF
    A long time ago I also tried to build a hexapod with pneumatic cylinders (http://www.tvdr.de/sim/1) but gave up because the whole thing wasn't "stiff" enough. Looking forward to learn more about your approach!
  4. Danny Miller

    Danny Miller New Member

    Joined:
    Tuesday
    Messages:
    4
    Balance:
    24Coins
    Ratings:
    +0 / 0 / -0
    My Motion Simulator:
    6DOF
    But it doesn't use precision control. Once it's set to equal the platform and rider's weight, it's just closed unless there's leakage.

    A tripod of linked cylinders is intentionally unstable. If the motors are unpowered, and capable of backdriving, then you might be able to balance in the center but move off center with no linear struts to stabilize you and you'll put more weight on one, and it was only *equal* to your weight before. The piston will move down and due to the shared air volume with equal pressure the other two will move up.

    Vertically, we're trying to keep it from being stable. If there was no equalizing tank, if they were pumped up until you floated in the middle, you actually would be stuck there, albeit with a bouncy cushion. When all 6 motors try to pull you down, pressure will increase and again it's all shared pressure. So it will resist getting pulled down. If you heave up, pressure will decrease and thus no longer equal to your weight, so it will be like a big bungee cord pulling you into the center which isn't as good as it sounds, the motor's tuning would probably have to be limited overall since it has to fight your weight in both cases.

    With the buffer tank (and all the air line's ID) volume being >> than the 3x cylinder displacements, then all 3 cylinders going from midpoint to the bottom of travel on a heave means only a minimal increase in system pressure and the linear actuators have it really easy.

    Basically we seek to be neutrally buoyant, like in scuba diving. Except in scuba diving, they want the BC vest to basically keep you upright by keeping more mass below than above the center of buoyancy. We don't want that inherent stability nor instability in any DOF, but it won't be perfect.

    Of course it's not entirely balanced. Your combined platform and body CG should end up above the center of the imaginary triangle formed by the 3 joints. What actually happens is hard to say exactly without delving into the inverse kinematics solution. If you roll 25 deg to the right and hold it there, if your CG is still centered on the platform, you're still mostly balanced but not quite because the cylinders have new theta angles. But I only specified 25 deg of roll, I don't know where it is- left or right of the center of the base? The IK solution probably offsets the CG to the left or right. Offsetting loads your weight more on one side than the other, so the linear actuators will need to compensate by pushing or pulling with linear force to hold position. Still much better than holding your entire weight up.

    The bottom line is the precision is entirely up to the linear actuator's driven distance created by the IK math, as in any other solution. They will hold you fixed in all 6 DOF and not be "spongy" at all. There are no valves and no control over the system once it's set up for the rider's weight, although I may restart the compressor into the shared volume if pressure starts leaking out. And that would just be to get back to target pressure.
  5. PeterW

    PeterW alias Wickie

    Joined:
    Oct 21, 2018
    Messages:
    204
    Occupation:
    Dipl. Ing. Mb (FH)
    Location:
    Germany
    Balance:
    1,619Coins
    Ratings:
    +365 / 3 / -0
    My Motion Simulator:
    6DOF
    Hi Danny,
    I also thought a lot about weight compensation with compessed air zylinder that is mounted parallel to a motor-driven linear actuators. But I did it a little different with my first prototype-actuator that I built some years ago. I didnt use an extra air-zylinder, but combined both Systems (air-spring and motor-driven ballscrew) inside the cylinder. This worked great, but as I saw that the motors allone (without the weight compensation) with 5mm pitch ballscrews pushed easily the weight (30 kg each actuator) , I abandoned this idea.
    But later and after using my 6DOF-simulator a lot, I wanted ballscrews with higher pitch (10mm; 16mm; someday 20mm?) which means less noise, faster and much better feeling for flight sims. With this high pitch-ballscrews AND limited motor-power (or limited PSU), weight compensation seems a good way for smooth motion, again.
    But as you know (for sure), there are big motors available that are able to move also a 20mm pitch ballscrews easily. Long story short: whether weight compensation makes sense or not, depends on the usecase (racing with smaller strokes or flight sim with huge strokes), the ballscrew-pitch, the power-supply and the motor-size you intend to use.
    Another comment I want to give: you wrote, you want 1000mm/s actuator-speed:eek: . Beware ! This speed can really be dangerous and break your neck! Imo you will not need more than 500mm/s which is still crazy fast when your move your maybe 150 kg platform. And you will need super strong fixtures between your simulator and the floor! Specially with your small footprint of a 800mm triangle.

    Quite interesting project!!:popcorn
    I wish you good luck and great success with your project.
    • Agree Agree x 1
  6. Danny Miller

    Danny Miller New Member

    Joined:
    Tuesday
    Messages:
    4
    Balance:
    24Coins
    Ratings:
    +0 / 0 / -0
    My Motion Simulator:
    6DOF
    Ballscrews? Where we're going, we don't need ballscrews...

    OK, what acceleration is good/excellent/too much?

    I'm doing the math and it could be close to 6m/s^2, about 0.6g, with the first pass of the design concept. That's a pretty rough estimate. There's some things I can change to trade off max v for force and accel but it's a limited number of options that would keep it simple/cheap/backlash-free.

    OK, calcs- if I got 6m/s^2, it would need 0.166 sec to reach 1000mm/s. 0.166 sec of 6 m/s accel will use 82mm of travel. And another 82mm to brake. Earlier I thought it wouldn't even have enough travel to get to that speed but no, this looks like it could. Now "should" is a good question.

    Actually, how does the velocity matter? It sets how long it can sustain accel, not accel itself, that sets the neck-breaking factor. Well I prefer to view it as a kidney-stone-passing factor, but whatever,
  7. noorbeast

    noorbeast VR Tassie Devil Staff Member Moderator Race Director

    Joined:
    Jul 13, 2014
    Messages:
    20,742
    Occupation:
    Innovative tech specialist for NGOs
    Location:
    St Helens, Tasmania, Australia
    Balance:
    146,304Coins
    Ratings:
    +10,813 / 52 / -2
    My Motion Simulator:
    3DOF, DC motor, JRK
    There are physiological and psychological limitation/thresholds, motion simulation often exploits those processing weaknesses, for example with washout filters.

    Keep in mind it is not just about speed, but also mass and inertia, both of the rig and the user.
  8. cfischer

    cfischer Active Member Gold Contributor

    Joined:
    Sep 7, 2015
    Messages:
    357
    Location:
    Colorado
    Balance:
    2,595Coins
    Ratings:
    +251 / 1 / -0
    Food for thought.

    Check out Iris Dynamics
    .

    as well as Kinemaniacs



    both use pneumatics and direct drive motors to get high response rate motion.

    Also think about delivering 1 to 1 rotation data from a sim to a driver in pitch roll and yaw with 1 to 1 high frequency data in surge sway and heave. IE the pitch of a rally car going over a bump or jump can be delivered to the pitch of a motion sim exactly 1 to 1. That can set requirements for your motion. You will see large range of motion for pitch and roll (±30°-40°), continuous yaw with very high accelerations, and a few inches of surge sway and heave travel are important for the highest quality kidney-stone-passing factor.

    All low frequency data should be communicated through pressure based systems that can be sustained as long as needed. That means g belt, g seat, g helmet, and more.