Question Research/Planning for a 6DOF Build.

Hi All!

Long time lurker who has been planning a build for far too long and am hoping I could get some input/advice so I can finally pull the trigger and put my dreams into a reality.

After seeing the price of the 6DOF MotionSystems.eu platform (20k+ EUR), I realised DIY was going to be my only option. I was also considering the DOFReality 6DOF unit but appreciate I'd be mainly paying for the convenience and a relatively quick and easy setup(?). Not to bash what they're doing at all, but I feel like the specs potentially leave a bit to be desired and I can get more out of a cusom build(?) as I don't necessarily mind the extra work and tweaking that will be necessary in setting up my own solution.

I'm especially inspired by Peacemaker Motions's setup but also Departed Reality's and Christopher Knauf's approach (all on Youtube but I've seen their build posts here also).

For a bit of context, I'm a mixed simmer (Racing and Flight mainly, but don't want to limit myself, so would like the ability to try out whatever else may randomly tickle the tastebuds such as trains, boats or whatever else I can experience with motion) and I exclusively play in VR. This pretty much solidifies the choice for the rig being 6DOF.

My first purchase was a Next Level Racing simrig (GTTrack), although this was not really with motion simulation in mind. I was aware they sell a motion platform bolt on, however was always of the thinking that I would just build and bolt it onto my own DIY motion build when the time came so I could continue to enjoy my racing and flying with my existing setup. I also appreciate that may complicate things due to the dimensions and weight of the existing rig and am not against the idea of ditching/modifying the existing rig if I have to. I understand it may not be as simple as bolting on also, so I won't force it to work if I simply need to start from the ground up.

I have a few (or many) questions which I've been struggling to find answers to and would grately appreciate your expertise. I'm so glad a community like this exists!

Motors motion type:

Threaded rods vs arms

I've noticed there are mainly 2 types of movement used for the actuators, arms/levers and threaded rods. What are the pros/cons of each approach?

Arms:

• I've noticed the arms can have a smaller footprint but does this also limit the range of motion?

• How does the length of the arm affect the overall range of motion?

• Is a longer arm "weaker" and slower?

• Does this in any way affect the motor choice, i.e. not needing a motor as 'powerful' as a threaded rod linear actuator approach?

Threaded Rod:

• Is the rod length essentially the length of motion/travel?

• How does this affect the dimensions of the Stewart/hexapod Platform? i.e. does the footprint of the platform affect the amount/range of motion possible or is there like a 'golden rule'?

• What is the significance of the thread pitch i.e. does translate to slower/faster motion or increased/decreased 'power' requirement?

• Is the pitch the same as the lead size?

P.S. I know there are also belt systems/approaches but would prefer a direct force for increased precision/reliability - belts have backlash, lag, wear over time, etc... I just feel the other 2 approaches would be better for longevity and accuracy.

Motors:

• I've seen a large variation in motors used, however from my research it seems the 80/90ST Servo Motors are the defacto - however I appreciate this may also be very dated information?

• Are there newer/modern alternatives that are perhaps smaller/more power effecient/more powerful, etc? I see SFX-100, ODrive and other such words thrown around but must admit, I don't quite understand what their differences are or whether they're just variations of the 80/90ST motors?

• I'm assuming the motor determines the compatible drivers/controllers? I.E. If I'm to use a Thanos Controller, it is only compatible with certain motors?

Torque:

• How does the motor torque affect the motion? Is this about the speed/strength of the motion or does it more appropriately translate to weight/load limit?

• Does one motion type require more torque over the other (arm vs threaded rod)?

• How do I know the requirements in order to be able to handle a certain load?

• Does this affect the speed of motion or is it more about 'strength' (weight/load limit)?

Motion:

• How much motion is too much motion? I understand that the software cleverly translates motion not into direct movement of the platform, but instead to the force the motion in sim would exert (if that makes sense). With that said, if for example I'm 20 degrees pitch up in flight sim, would I need the same or more motion to also compensate for the additional G factor?

• Does the size of the rig determine the available range of motion/required motor strength?

Part Strength:

• I do have a 3d printer but question the reliability/strength of some parts as quite a lot of load of force will be exerted on certain parts. Is this a concern?

Apologies for the absolute wall of text but I really want to get this right and fully understand what I'm doing. That's all the questions for now as I don't want to get ahead of myself and subsequent questions (if any) will depend on the answers here.

I did ask ChatGPT for some input and it gave me the following response which may save some time/effort for responses, please let me know if there are any additional considerations to the answer or anything simply wrong.

Thank you for your time and patience, hope this is not 'too much'. (links removed to mentioned products/parties as it was flagged as potential spam), please let me know if there is anything I should amend in the post/any rules I'm breaking, really sorry if that's the case as is not my intention.

Wow, these are really a lot of questions. I can't answer them all at once but step by step.

I think what you call "threaded rods" are usually referred to as "ball screws". They are widely used for high precision positioning applications like CNC machines. High precision and low backlash is not really important for sim rigs. So most DIY builders use the cheapest possible ball screws available. Basically all work but they have a disadvantage: the rolling balls make a rumbling sound especially when rotating fast. So a good advice is NOT to use the most commonly used pitch of 5mm but instead choose 10 or even 20mm pitch and add a reduction gear (timing belt) instead of coupling the motor directly to the screw. This way the screw spins slower making less noise. It also results in much less inertial load to the motor and less vibration due to runout/imbalance.

Lever arm actuators or rotational actuators (compared to linear actuators with ball screws) are generally easier to build because the don't require specially machined or expensive parts like linear bearings. They are also cheaper if (and only if) the (always required) reduction gearbox is already part of the motor. Many use wheelchair or windscreen wiper motors. If you have to buy seperate planetary gearboxes for large servo motors the price of the gearbox can be higher than the motor itself so the price benefit is gone.

In theory you could make the arm as long as you want. But longer arms need more torque so you have to select the gear reduction factor accordingly.

So before commiting to an actuator type, arm length or thread pitch my advice is to calculate the required motor power, first. Ball screws and rotational actuators using toothwheels both have good efficiency. So the required power depends only on the weight of the rig and the desired max speed. Worm gears have a lower efficiency and require more motor power for the same output.

Lifting 75kg at a climb rate of 1m/s requires one horsepower. So if rig and pilot weight 150kg and you want to move 0.5m/s it's the same 1HP. The movements of the rig are generally not streight up at constant speed but very complex. So as a rule of thump you can assume that only half of your motors (3 of 6) have to handle the full weight of the rig. So 300W per motor should theoretically be enough. Many use 750W motors which is plenty.

Next you need to decide how much stroke you need. More is always better but if you don't have a barn or hangar but have to put your rig in the living room or garage the ceiling height limits your max heave and pitch.

Thank you very much for your reply!

I'm mainly heading in the direction of linear actuators with ball screws. The main reason for this is a nice balance of speed, precision and motor 'friendliness'.

I'm looking at ballscrews with an 8mm lead for a nice balance of my mixed playstyle - 10mm/20mm as you have mentioned is fine and fast for racing, but I don't want to lose some of the finer detail I'd benefit from in flight sim for example or add complications of reducers/gearboxes, etc. You have stated the high precision and low backlash not being important for simrigs, but it's the accuracy that I'm actually desiring. Is it really that negligible? - I want to feel as much detailed feedback/motion as reasonably possible.

5mm is very detailed but not as 'fast' for when I'll need it in intense applications such as racing, this is how I've gotten to 8mm as a nice balance. With that said, I have not considered any reductions and this seems unnecessarily over complicated if the motor and ballscrews are appropriate in the first instance? Or is it more for the noise as you have mentioned? My preference is to couple directly to the motor although I don't have a specific reason for this currently other than purely from a simplicity and 'specs' standpoint. I don't want any additional complication/point of failure or to sacrafice resolution/detail and coupling directly seems the most effective and accurate option for that?

Said motors I'm leaning towards are the 80ST's with ~2.3-2.9nm of Torque (750w motor) which I've calculated should be fine for ~1k KG force/load at a 10mm lead, so 8mm will be slightly more(?) (overkill, but possible and playing it safe to have some overhead?). To clarify, my load will likely be about a quarter or half of that max.

But I also don't know about the others such as sfx-100/odrive setups and their pros/cons/considerations.

Am I on the right track so far or am I missing some important considerations?

Precision and "detail" as you say are no issue at all. If you have a 3D printer with stepper motors you see a big difference of microsteps vs. full step and the pitch also matters. But all modern servo drives have at least 1000 steps/revolution, many have 8000 or 10000. 8mm vs. 10mm pitch would mean 1µm vs 1.25µm for 8000 steps/rev. That's ridiculous, you won't feel any difference.

What's more important is the feeling, that means you want low delay and smooth motion. Backlash can be felt if it's more than 0.1mm as it creates small bumps on direction reversal. But due to the bias load of the weight you hardly ever run into it. Even with aggressive car racing you should never encounter negative gs except when crashing and then it doesn't matter.

Arm actuators are non-linear, of course. At both ends of the stroke the "effective pitch" (linear motion per rotation angle) decreases. This makes calculation and tuning a bit more complicated. But the software does everything for you, so no worries.

The speed you need depends on the distance of your pivot points where the actuators are attached to the rig. For small 2DOF or 3DOF rigs where the actuators are directly under the seat you don't need much speed so 5mm pitch is OK. But for a big rig where the joints are far out like this

(thanks @Klaus Schmidinger and @Dirty )
you want something like at least 300mm/s. With 5mm pitch this would require 3600RPM which is insane for a long spindle. For 10mm pitch it's 1800RPM which is reasonable but maybe still loud. 16 or 20mm pitch would be even better but requires a reduction gear.

2.5Nm with no reduction will give you around 1.3kN force at 10mm pitch for each actuator. At an angle of 60° (each pair of actuators forming an equilateral triangle) this results in 6.7kN total lift force. More than you'll ever need. The servo motors can also be overloaded for short periods of time by a factor of 2 or 3. So peak forces for acceleration and decelleration is also no problem.

So if you have the money I'd go for AC servos. I don't know much about the SFX or Odrives I have to admit. But those "universal" motor drivers usually need a lot more tuning and extra tricks for protection. So yes, you can save some money but a ready-to-go combination of drivers and motors is much more convenient.

BTW, the low friction of ball screws has a disadvantage: you need to account for the case of power failure. Worm gears are self locking and rotary actuators automatically provide a "soft stop" at the bottom. With ball screws and high pitch the rig almost free falls down in the case of a failure. So you'd need one of

1. motors with electromecanical brakes
2. hydraulic end stop dampers
3. brake resistors for the drivers
4. gas springs as counter weight balance

Really appreciate the feedback and patience!

To clarify I'm understanding correctly, in reality, the difference between 5, 8, 10, 16 and even 20mm lead rods is practically microscopic in relation to motion simulation?
So instead these really matter more for precision applications such as robotics and CNC for example?

I cannot find the specific steps for the 80ST-M02430 motors (and I'm guessing it may vary per manufacturer), but assuming worst case scenario in your example of 1,000 micro steps, it would still essentially be less than a 1mm difference between the rods? Meaning for the application of motion simulation, the human body is highly unlikely to tell the difference in 'detail' between them, meaning realistically, no decernable accuracy is lost by moving to a larger lead size?

In that case, if the lead doesn't practically matter, am I right to assume the lead choice would be more down to the desired lift force and/or desired speed of motion and spacing of each actuator?
Ofcourse this is all with the assumption that the motor is adequate for the choice of lead screw and the torque requirements that would accompany.

As for the power failure issue, I'm planning on using a UPS which I could probably set to send the park command in the event of a power failure. I did consider motors with brakes, but I don't know enough about the practicality of this application as I believe the controller would also need to be compatible with the brake function or some additional functionality would need to be implemented to cater for the brake?

Well, yes and no. First, a screw with 10mm pitch and direct drive would result in the same speed range and resolution as a screw with 20mm and 1:2 reduction gear. But the 10mm screw has to rotate twice as fast and has four times the inertia (that scales with speed squared). Inertia matters for long and heavy screws and it affects servo tuning. But for 16mm diameter and <1m it doesn't really matter. Lower speed means less wear for the dust wiperrs and less noise.

You don't feel position. In fact you can't tell any difference at all if you wear VR goggles. You feel acceleration. Human senses are quite inaccurate. I'd say +/-10% acceleration makes no difference for large strokes. But you can feel vibration with very little amplitudes.

Yes exactly. Make sure the motors are powerful enough (750W surely are). Then choose arm length or screw pitch according to your requirements and finally select the gear ratio so you have enough torque for the selected lever length or pitch.

An UPS can protect against an actual power failure. But there are other cases like software bugs that can cause a servo drive to quit (overload due to excessive acceleration). Motors with electromecanical brakes are the easiest. You can power the brake magnet from a relay connected to the "status OK" output of the drive. In any case of failure power to the brake is cut and it engages. But this is also the most expensive solution.

3D printed parts can be very handy. Have a look at @Dirty 's linear actuator design. He made quite clever use of 3D printed parts and even 3D-printed bearings out of special low friction plastic from IGUS. This is totally OK if you make sure that the tension load stays within the limits and a failure/breaking of the parts dont have catastrophic consequences. For example, if a bushing inside the actuator fails then the pushrod rubs against the housing but it doesn't fall apart.

However, if one of theese 3D printed joints fail the whole rig can flip over which could cause serious injuries.

Are there any resources for the dimensions/layout of the base/actuators and how the length of the rods relate to this and affect the motion/footprint?

You can use FlyPT Mover to visualize and calculate everything, it's the easiest way

Just add 6Dof linear hexapod and edit its dimensions, than run 3dviewer

It's been a while!

Just another quick question on the ballscrew, I've ordered some 1610 SFA C5 ballscrews, however noticed actually most people mention SFU ballscrews (my fault for not paying attention on the product page). Is this likely to be an issue?
I've struggled to find much information but it seems that SFA ballnuts are a simpler, less precise and less load bearing? Would it be better to cancel the order in favour of identical SFU Ballscrews?

This is precisely the reason I went with linear actuators over rotary ones. I originally built a 2 DOF with wiper motors and it was a bit underpowered, but it also had issues if you weren't careful with how much you let the system rotate. With more powerful motors that's less of an issue though. Linear stuff gives a more straightforward tuning setup and consistent loading to the motors, though as Aerosmith said, software can potentially address this. However, if you have limited PID tuning controls, it could mean that your rig is more optimized at one point in the travel over another. My Kangaroo controllers actually auto tune by specifically moving the rig to a couple different positions along the full range of travel and doing some oscillation tests at each stop. I think it then averages the results at each location to determine the final PID values.

C5 SFA are higher precision ball screws than the normal SFU c7, so should be better in every way, except maybe more expensive (What Grok told me)

I am planning for SFA2010 C5 on my 6dof build. Working on the first actuator design now.


Edit: Confirmed.


The differences between SFA C5 ball screws and SFU C7 ball screws stem from their design, precision grade, manufacturing process, and intended applications. Here's a breakdown:
1. Precision Grade (C5 vs. C7)

• C5 (SFA C5): This refers to a higher precision grade. According to industry standards (e.g., JIS B 1192), a C5 ball screw has a maximum lead deviation of 0.018 mm over 300 mm. This means it offers greater accuracy and lower positional error, making it suitable for applications requiring tighter tolerances and repeatability.
• C7 (SFU C7): This is a lower precision grade with a maximum lead deviation of 0.050 mm over 300 mm. While still precise, it is less accurate than C5, but it is often sufficient for many general-purpose applications and comes at a lower cost.

2. Manufacturing Process

• SFA C5: Typically, C5-grade ball screws are ground, a process that involves precision machining to achieve tighter tolerances. Grinding results in smoother surfaces and higher accuracy, which is why C5 screws are more expensive and used in high-precision environments.
• SFU C7: These are usually rolled ball screws, made by rolling the threads rather than grinding them. Rolling is a more cost-effective process, leading to slightly rougher surfaces and less precision compared to grinding, but it’s still reliable for many applications.

3. Nut Design and Circulation

• SFA: The SFA series often features an end-cap circulation design, where the balls are recirculated through end caps at both ends of the nut. This design can provide smoother operation and lower noise levels, making it suitable for smaller-diameter screws or high-speed applications. It’s often paired with quieter or more specialized setups.
• SFU: The SFU series uses an internal circulation design, where the balls recirculate through an internal deflector within the nut. This is a common, robust design that’s space-efficient and widely used in industrial applications. It’s generally simpler and more cost-effective than some other circulation types.

4. Applications

• SFA C5: Due to its higher precision (C5) and potentially quieter operation (end-cap design), it’s better suited for applications like precision CNC machines, semiconductor equipment, or other systems where accuracy and low noise are critical.
• SFU C7: With its C7 precision and internal circulation, it’s more commonly used in general automation, industrial machinery, or CNC setups where high precision isn’t as critical, and cost-effectiveness is a priority.

5. Cost

• SFA C5: More expensive due to the grinding process and higher precision grade.
• SFU C7: More affordable because of the rolling process and lower precision requirements.

6. Performance Characteristics

• SFA C5: Offers lower backlash (or even zero backlash with preloading), higher rigidity, and better repeatability due to its precision manufacturing.
• SFU C7: May have slightly more backlash and less rigidity, though it’s still highly efficient and reliable for less demanding tasks.

Summary

• SFA C5: Higher precision (C5), ground threads, end-cap circulation, suited for high-accuracy and quieter applications, more expensive.
• SFU C7: Lower precision (C7), rolled threads, internal circulation, suited for general-purpose use, more cost-effective.

If you’re choosing between them, it depends on your application: go for SFA C5 if you need superior accuracy and smoothness (e.g., precision machining), and SFU C7 if you’re looking for a reliable, budget-friendly option for less critical tasks (e.g., basic automation).

@runar totland has already explained everything. But a shorter answer would be:

Precision doesn't matter at all for simulator applications. We are talking about 1/100 of a mm here. But the higher quality manufacturing process (grinding vs. rolling) means a smother surface and less noise. Also, end-cap ball deflection is superior to internal deflection because the balls are in contact with the screws continously over multiple turns. This means higher load capacity and better robustness for excentrical loads (tilt forces).

Thank you all for the feedback/reassurance - my main concern was the load bearing capability of the ballnut however if I have not misunderstood, it seems to not be a practical concern.

I'll be pairing these with 80ST-M02430 750w 2.39Nm AC Servo motors and have decided to go for the motion4sim motion controller. The motors will have plenty of power to drive the setup and the motion4sim controller seems to cater well to this combination of motor, actuator type (linear) and screw length/pitch (600mm, 16mm diameter, 10mm pitch - (1610's)).

As for the actuator design, I'm currently leaning towards the PeaceMakers design (since I don't have or currently use aluminum profiles and believe the weight of the PVC pipe will be preferable). However, I’ve been considering making an alteration by replacing the 3-metal aluminum rod ballnut design with a single central carbon fiber tube. I still need to do more research, as I don’t believe carbon fiber will withstand the wear from the linear bearings over time, since it’s not particularly durable in terms of friction.

Keep in mind that actuator weight may not really make too much of a difference in the long run. It's more the platform weight that matters as that is what is getting thrown around. Yes, the actuators are attached to it and some of their mass would be considered "sprung" to use a vehicle suspension term, but since they are fixed to the base as well you are only moving a portion of their weight.

You might see if you can calculate the estimated weight of the different designs just to see how much they might vary. I was surprised my actuators were as heavy as they were, but the bulk of their mass is the motor on the end of them, which is on the end that's fixed to the base, so that weight isn't really moving much.

I totally agree. Also keep in mind that carbon fiber has a very good tensile strength to weight ratio but can't take much compression load or bending. And if it fails it fails catastrophically (see Oceangate desaster). I don't want to be sitting in a rig with a broken carbon fiber tube when it collapses. A metal tube bends before it breaks and most likely doesn't have razor sharp edges.

I'm planning on doing a build guide for the linear actuators once I've finalised the design - this is in part to consolidate a lot of information I've had to piece together across many sources and posts, aswell as to make it easier for anyone in the future wanting to follow along and understand what/why/how it all works.

This community has been invaluable and I really hope to contribute back in the form of easy to find/follow/understand resources. It's amazing having all the information here but my head is spinning after reading so many posts/threads to find stuff out (some of which is quite simple/basic but not easy to find/obvious from a newbie perspective). If possible, I'd like to make it easier for future newcomers like me to find and wrap their head around the fundamental information

With that said, I do have a few more questions...

I'm struggling to find the most basic information on how to plan/design the platform. I see L1, L2, L3 measurements thrown around, Leg angles, etc and someone mentioned to use FlyPT mover to visualise all of this, but I feel like I'm missing the very first step and understanding of the platform dimensions, references and angles etc.

I've read https://www.xsimulator.net/community/threads/basics-6-dof-stewart-platform.9396/#post-119948, but it doesn't reference L1 - L4 dimensions, platform height or any other language that is commonly used today when discussing 6DOF Platforms. It seems to focus more on how the dimensions can influence the movement.

Sorry if I'm missing something so obvious or maybe my brain is just a bit fried, but:
What are the main differences between a circular base and hexapod base?
How do the dimensions affect the motion platform?
What angles/locations am I supposed to mount the actuators to the base and the platform/rig?
What are the L1 - L4 dimensions?

Finally, the couplers for the ballscrews.
For the application of motion simulation and directly coupled actuators/motors, is it better to go for a flexible shaft coupler, rigid or double diaphragm?
How do you know which coupler is required in terms of dimensions (D#L#)?

As always, thanks for your patience and help! I'm preparing to have all of these kinds of basic info compiled to make it all beginner friendly to those of us without engineering degrees

I'll offer my opinion on a couple of those things at least.

Regarding the base, the shape of the base really doesn't matter that much. What really matters is where on the base the mounts are. You can have a triangle, a circle, or three spars sticking out from a center point that all have the same joint points if you wanted to design it that way. So you might want to be a little more specific and potentially even draw some stuff up to better explain in case I'm interpreting that incorrectly.

As for the dimensions, generally speaking the larger the radius of the platform, the more stroke you are going to need to get the same amount of tilt. Picture a seesaw and imagine pushing down on one end. If you are really far from the pivot, you have to push a lot further to get large angles. Closer to the pivot and it takes less movement to get the same angle, but more force. So a larger platform needs more stroke, but less powerful motors. Linear motion is less affected by platform size and more just directly related to stroke.

Angles and locations are what define the capabilities of the rig and are harder to describe as they are very 3 dimensional. If you keep your upper and lower platforms the same size it gets a little easier as it keeps things more in-plane and leads to less complex angles and also gives you a pretty even trade-off for rotation vs linear movement. The link you posted covers this a lot better than I am likely going to be able to, but as a general reference, if the upper or lower platform is smaller than the other, it trades angular motion for linear motion. Think of how the actuators are pushing on the rig and it might make a little more sense. If they are leaning inward or outward they are going to tend the push the platform away more than up, so you get a higher ratio of linear motion to rotational.

The L1-L4 dimensions to the best of my knowledge are all straight off of the FlyPT Mover software. That is just what they use to define the different points in the system and the dimensions between them. It's easier to open up the software and take a look (or maybe just check the website, I think they have it spelled out there as well). For the most part they just define the distance between pivot points and the stroke of the actuators.

Regarding couplers, I've only ever seen people use flexible couplers of some variety. I'm not as familiar with the different names for them, but the general recommendation I've seen is the Lovejoy/spider style ones. The "spring" style that has a spiral cut piece of metal doesn't like constantly reversing motion and tends to come apart. The spider style is designed with both directions of rotation in mind.