Showroom My take on the typical linear actuator 6DOF

I'll try to get some pictures on here soon as I'm hoping to assemble the mechanicals of my rig over the weekend, but figured I'd get a thread started up detailing my plans and designs for anyone interested. The long version of the story is that I got the idea to build a rig many years ago and ended up with a box of wiper motors from my old job. I put together a DOF Reality style 2 DOF platform mover using them with some metal I took from a scrapped treadmill to see if it was something I wanted to expand on and to see how it worked in conjunction with my Oculus. It was okay, but had some issues with a motor overheating and generally just didn't get a lot of use due to the hassle of my PC being in my office while my rig was in the basement.

Fast forward a couple of years and my dad started getting into flight simming, which got me thinking about airplanes again. I had always toyed with the idea of updating to a 6DOF, but really didn't use my current 2DOF enough to warrant it. I figured building a simulator was a cheaper way to practice flying than renting a plane though, so I gradually starting collecting parts over a good year or two while I modeled everything up in CAD to see what I would need. The following is the general basis for my design:

"Typical" style linear actuator design seen on the forums

• 1605 ballscrew, 500 mm long
• MY1024 scooter motors (24V DC, ~250-350W depending on who you ask, cheap on Amazon)
• Openbuilds 20x60 aluminum extrusions, 500mm long
• 100 ppr optical encoders (Amazon)
• Limit switches on either end (lever style microswtich)
• Igus linear slide bearings
• Some 3D printed parts as needed (bearing housing, gears, slides, couplers, etc.)

The main differences I have to a lot of the builds I've seen here are that I'm using 3/4" actuator tubes and gears for my encoders instead of belts. The idea with the smaller diameter tubes was that I wanted the tubes to be a close fit to the ballscrews to help control any whip since they are only supported well on one end. I was thinking this could also be used to help retain some grease for the screws, but we'll see how that goes. I originally was looking at a slightly larger diameter tube with a separate nylon tube I was going to slide inside so I'd have a softer wear surface, but we'll just see how it goes with the 3/4" (plus it was harder to find linear bearings in weird sizes). I have a feeling they are going to be a bit loud, but I'll have my Oculus headset on anyway, so hoping it's manageable. I also flipped my linear bearing housing around to what I see most people do so that the bearing is inside the actuator instead of on the outside. As it turns out, this puts the bearing almost exactly where the ballscrew ends, so I think it will help support it better. For the encoders I printed one gear in PETG and the other in a TPU style softer plastic. The goal there was it would give me a little bit of a pressure fit to remove backlash concerns and also hopefully keep them a bit quieter.

My control system consists of Sabertooth 2x32s with Kangaroo x2s and 3x 24V 1500W power supplies (~63A). The idea being that I'll have one "control box" (a power supply, Sabertooth, and Kangaroo) for each pair of actuators since the Sabertooths will run two motors at once. By the numbers my power supplies should be able to run a pair of motors each at rated capacity, but I'm guessing it might be a little short on surge current. Time will tell, but I can always see about adding some batteries or something or just turning down the motion.

I'm reusing the original "spine" I built for my 2DOF, but adding arms to it to mount the actuators. My current design plan is to have short tube "sockets" on the base where I can slide in different controls as needed based on the type of game I want to play. So if I want to race I can put a steering wheel in the center post and a shifter on the right, or if I want to fly a jet I can swap that to a joystick in the middle and a throttle quadrant on the left, or even mech combat with a joystick on either side. Or, I can take them all off to play something like VTOL VR that uses all motion controls. It will be a little weird having the steering wheel tube between my legs, but I don't think it will be a deal breaker given the flexibility it should offer and I'm hoping it will make getting in and out easier (my old hoop design on the 2DOF was a major pain).

The base is simple 3x1x.120 wall tubing. Depending on how much the thing wants to move around I may weld some sleeves into it so that I can bolt it to my floor in the basement, but until things actually move I'm holding off on that.

At this point I have all my actuators built and the base fully welded together. The upper spine is mostly welded as well, but needs my control socket mounts attached and some tabs for seatbelts that I hope to get welded on today. With any luck I'll have the mechanicals all assembled tonight. I have the beginning on one control system mocked up on a piece of wood, but it needs some work as I plan to mount some fans and stuff to keep everything happy and I'm guessing I'd like to have some kind of cover in the long run. I'll see if I can get some CAD model screenshots and build pictures uploaded later today or this weekend to better explain the design as I'm sure it's not that easy to follow my ramblings and descriptions in just text.

Here are some screenshots of the CAD showing my initial design ideas. For the most part I carried this all through without any major changes. I ended up adding some shaft collars on the rods that are on the base that the actuators mount to. I figured it was some cheap insurance to provide some more support in case the set screws got loose.

Here's the main base. It's pretty simple 3x1x0.120 rectangular tubing. The rods are drilled in at a 60 degree angle and rotated roughly 22.5 degrees off center. The goal there was to have them as straight as possible when the rig is in its neutral position.



The "spine" as I like to call it was loosely based on a DOF Reality setup originally, but the dimensions were tailored to try to match my own car more closely and built to use my wheel and pedal set (Thrustmaster TX/T300 with the T3PA Pro pedals and H8A shifter). The center tube the wheel mounts on is removeable to allow for swapping to different controls like a joystick or flight yoke and would let you customize the height to your taste.


The actuators are the usual design seen around here, just built with DC scooter motors. Nothing too fancy or ground breaking, but I did swap the bearing on the end to point inboard as the ball screw ends right there, so it seemed like a good idea to have the extra support there. For the time being the bearing housings are 3D printed out of PETG as they were going to be one of the more expensive parts if I bought them in metal, and I wasn't sure if my 3/4" tube idea was going to work out or not, so I didn't want to have to buy them twice. We'll see how they hold up.


Here's a close up with transparency to show what I mean about the ball screw ending inside the linear bearing with the housing turned this direction:


And here's a closeup of the encoder and gear setup. The gear on the coupler is just "pressed" on. It's made of TPU though, so it's pretty soft and not so much a "press" as a "stretch". It's moderately grippy stuff though, and so far seems to hold on okay. If I see issues with it twisting on the coupler I'll probably print another that has some tabs that lock onto the notches of the coupler where the rubber cross block goes to more positively hold it on.


I got all the physical parts bolted together last night and am pretty happy with how it went together so far. I ended up adding some tabs for seat belt mounts to the spine as well as that seemed like a good idea. Hoping to get all the electronic assemblies together today and will see about getting pictures of the actual build while I'm at it. So far things seem pretty solid, so I'm cautiously optimistic.

Got my control "panels" built and everything wired up to do some testing and tuning. Looks like I might have an issue with the limit switches on one actual and another is giving me a funny limit error, but I'll dig into that tomorrow. The original CH340 serial to USB converters I bought don't want to talk to the Kangaroos anymore, so I've got some FTDI based ones on the way that will hopefully be here tomorrow. I have one spare FTDI based dongle I've been using in the meantime to program each unit one at a time. Each control panel has a 1500W 24V power supply, Sabertooth, Kangaroo, and a clamp resistor to protect the power supplies (a Sabertooth feature). The resistors are technically undersized at only 100W, but I'm going to run them and see how they do first as I don't think there should be a lot of regen with this design as the motors don't want to freewheel very fast if at all. I plan to add a standard PC style fan to the end (bottom corner of the board in the picture) to blow over the resistors and the Sabertooth. Main issue I have with that at the moment is that I only have 12V fans and the power supplies are 24V, so just kind of a pain to have another entire supply just for fans. Might see about getting some 24V fans instead, but haven't dug into it yet.


Here is a quick picture showing the collars I added on the lower actuator mounts at the base. They have set screws of their own, but I just welded them on the bottom side to hold them in place. Some have more gap to the u-joints than others, but these are more just a safety catch than anything anyway. Astute observers will notice that the u-joints are very much not in line with the actuators. As it turns out, when I made my 3D model I had one set of them rotated the wrong way, and by Murphy's Law that was the one I measured to get the numbers to build with. So the 60 degree tilt is still there and correct, but the 22.5 degree twist is going to the outside of the base instead of the inside. It doesn't seem to hurt anything at the moment though, so I'm just going to roll with it. Drilling those holes was one of the most tiring experiences I've had in a while and not something I want to do again. I could cut the welds on the base and flip the tube around there to fix it, but until I think it's a problem I'm going to leave it be and see how it does.


This is the 3D printed linear bearing housings I made. I didn't go crazy with them, just my fairly standard 20% infill. They seem stout enough and they really shouldn't be seeing much force anyway if everything is doing its job. I had the tubes hard chromed to make them more slippery and to help resist corrosion. Otherwise they are just typical DOM tubing. I originally test fit them raw and thought about trying to run them that way, but they seemed a little sticky in the linear bearings and all of the literature on the bearings talked about either hardened shafts or stainless or other corrosion resistant material, so figured this was cheap insurance. Fun fact, the clevis brackets you see at the top where the spherical joints bolt to the spine are actually U brackets intended for bathroom stalls. The models I got from McMaster for the spherical joints measured out at 3/4" wide, so I had bought some tubing I was going to cut down to make them, but the actual joints were 7/8" wide (which matched the datasheet if I would have been paying more attention). I looked at a couple different ways of making these until I finally found the right search term and found these. They worked out perfectly!


And here's a closer picture of the encoder setup as built. The red material is NinjaTek Cheetah, which is just a TPU blend that's easier to print. It's pretty interesting stuff because it's actually quite stiff if you print it a couple of walls thick, but reasonably spongy if you keep it thin. If you squeeze the gear in your hand it's suprisingly stiff, but if you squish it between your fingers through the thickness of the gear it's still pretty soft. All of the other blue parts are PETG. The tube is "attached" to the ball nut block by just friction. I printed a reducer bushing that presses into the block and the tube then presses into it. I added a slit with a hole for a screw to potentially add clamping force to the tube if needed, but so far I haven't used it. You can pull the assembly out of the nut block fairly easily, but gravity is also helping hold it together, so I think it should be okay for all but the fastest of moves.



Here's a quick shot showing the tabs I added for seat belt mounting. They were scrapping a bunch of parts at my work the other day and I snagged these. They are a little big for the job, but worked perfectly. I've got one at the very back that both shoulder belts attach to and then one of each side on the rear tube just behind the seat for the lap belts.


And here's a general overall picture of the assembly. It's a bit messy and I don't have peripherals attached yet, but I wanted to make sure everything worked before getting too far. I haven't made plates to mount the wheel or shifter yet either, so I can't even mount them if I wanted to. I've been debating whether to make one plate with multiple bolt patterns on it for a wheel or flight yoke or whether I want to make one stand for each peripheral. I don't actually have a flight yoke yet though, so that's still up in the air. I do have a joystick that I might make a base for, but it would mount a lot lower than a steering wheel and it doesn't really have a good way to bolt it down as it's an ancient MS Force Feedback Pro that's just designed to sit on a desk. Might have to get clever with the 3D printer to make some mounts for it.

Nice work. The only think you'll probably not be happy with are the 1605 screws with only 5mm pitch. 10mm would be better. Depending on the torque of your motors this would probably require a 1:2 gear between motor and screw but as you already have a gear for the encoder it wouldn't take any more part or space. Half the RPM at the screws means less than half of the noise and only 1/4 of the inertia.