Dirtys 6DOF, AASD Servos

Hey there

finally it all comes together! Totally excited to assemble the rig!

My actuator thread here
My software thread here

A few pics to start with:

This is how it's supposed to look in the end. A few details still missing.


Initially I tested the geometry with a wooden frame, just to check if there are any collisions.


The AASD servo controllers are quite bulky. The ClearPath motors had them integrated nicely, but these AASD are simply unbeatable in price and ease of power supply.


Eventually I tidied it all up and stowed them in a 9H 600x600mm server rack


Actuators have a stroke of 580mm at a speed of 420mm/s. Upper and lower Diameter might still be subject to change, but as it stands right now I get those values:

Heave: 650mm
Sway: >1000mm
Surge: ~900mm
Pitch: ±42°
Roll: ±38°
Yaw: ±30°

...of course not all of them can be at max at the same time, but in gerneral I am quite pleased with the range of motion. I am pretty sure I will eventually have to anchor the feet of the rig down to the ground with bolts, which is going to be a nightmare given that the room has underfloor heating

More updates to come. Let me know if there is anything in particular you're interested in
Find a link to the latest version in CAD here
Cheers,... Dirty

hi Dirty
Your work really knocks my socks off. I think you're a kind of multi-Genius
Everything is so precise and well thought out! Respect!!!!!
I am looking forward to your running sim

Wickie

bond a sheet of mdf to the floor and screw into that

yah,... I might actually do that.
Thing is, that there can be substantial forces pulling up on the feet, even if only momentarily. The glued contact surface would have to be about the size of a human hand per leg.

If I can get my hands on an infrared camera (to find the heating pipes) I might use bolts, but I already know what's gonna come next: I will have to reposition the rig a little which will result in the floor eventually looking like a Swiss cheese

In german I'd say: "It's like having to choose between the Pest, Cholera and Lepracy".

Looking great Dirty! Any chance I could take a look at the fusion project? The links in your actuator thread were no longer working.

Initially I had a different design for the upper frame, but that turned out to have unnecessary many welded joints and was prone to collisions with the fwd actuator:


I switched to a single-beam design that is a lot simpler and cheaper to make:


It is made of 60mm x 60mm aluminum square-pipe with a wall thickness of 3mm. Initially I was anxious if it would carry the loads adequately, but after a finite element analysis that turned out to be unjustified:

I put a 3000N load (150Kg @ 2G) in the center of the frame and the maximum tensile stress was 12 MPa. Garden variety aluminum can tolerate around ~300MPa which results in an overall safety factor for THAT load case of ~25.

It changes drastically when I instead constrain the center section and apply the load to the arms, because the part will now conform differently under the load. Max stress is now ~55Mpa which is only a safety factor of 6, but still good enough!

As with anything computational: Garbage in --> Garbage out. When using FE analysis pay attention which load case you test for. I would put it this way: If FE analysis tells you 'it will break', then it will break! Period. If it tells you 'it won't break', it only means it will not break under this one load case!

The U-joints are 3D printed and all axis run on 26mm or 28mm ball bearings. There is literally no moving part with friction on the entire rig except the IGUS linear bearings:


They were designed to transmit 1000N push- and pull force alike. Which is not much considering an actuator can produce up to 2500 N . We will see how they hold up over time.


Material used is PETG with a max tensile strength of ~50MPa in layer direction and ~25MPa accross layers. Being conservative, I considered 20MPa to be the failure load in all directions, overdimensioned everything by a large margin and came up with this:

A maximum tensile stress of 3.5 MPa at a 1000N load means a safety factor of ~6. Again, we'll see how it holds up over time, but I have already looked into printing PC (Polycarbonate) which could de facto double the tensile strength up to 90MPa and even 50MPa across layers. I hope I won't need it, because printing PC can give a headache to an Aspirin pill!

Here an earlier vesion but same principle:


As with anything that is not accessible for inspection after assembly, I'd use Loctite to prevent vibrating the screwed connections loose:


Cheers, Dirty

I will update those links, thanks for letting me know!

Very impressive.
I have one question. Could you pitch 25 and roll 15 at the same time and post here how much heave is left in both directions ?

thanks

I will as soon as I have power back on the rig. Might be a few days.
In general, I like your Idea, that we talked about: Defining a set of test cases to determine a kind of "motion range score". Not sure, if it will yield meaningful results in the end, but worth a try. It might be interesting to compare different rigs.

Dirty

...says the lifelong engineer who put up a show-ready rig in a few months that others rightfully called the "Mercedes of sim rigs"

to

...the tinkerer who has been wandering around this hobby like a blind chicken in the jungle for two years and thereby explored pretty much every dead-end road that exists in the field of motion sim building.

What a great compliment. Thanks!

Wonderful work @Dirty ,seems like mine in far better

Edit : I have question for actuators :

1-which bearings do you use ?
2-which 1610 ballscrews (chineese ?)
3-do you have noise between Spacer in rod and rod ? and which material .

Thanks a lot

Hey @dureiken

1) I use 26mm bearings for the U-joints. Those are the ones: 10x26x8 6000 2RS. The axial rotation is via 2 12x28x8 6001 2RS in a row. If you mean the bearings in the bearing block of the ballscrew, I don't know exactly. Would have to look it up. I kept the ones that were in there by default. I think those are ordinary skateboard bearings.

2) Yes, those are cheap... uhh.. sorry, "cost effective" Chinese SFU1610-700mm. From a precision point-of-view they are way good enough. They were made to move a CNC milling tool with a precision of 0.01mm, they will be good enough to move my cute little butt around. They can generate a quite noticeable CHRRRRR sound when moving, but I doubt that the expensive European or US manufacturers would make a difference that is worth paying 3 times as much for them. @PeterW bought some very expensive brand product. Maybe he can describe the difference. I plan on using plenty of grease on them to muffle the noise of the balls and am also already considering using 100x100mm square profiles to have the noisy part of the actuators well encapsulated. On the dimensions I would have to honestly admit: I have read the manufacturers specs and the way we (all of us!) use them is simply out of limits! By a large margin! If you look in the table for "bearing on one side only", you will find a max RPM of ~700, not 2500 or 4000 like some here do. Also the max axial load is given somewhere in the order of 400N, not 2500N as my motors are capable of. From a rational standpoint SFU2510's would probably be appropriate, but that being said, the SFU1610 work just fine

3) I printed a wide variety of spacers. Here are the last five in chronological order from left to right:

I tried to get the outer diameter of the spacer as close as possible to the inner diameter of the rod (24mm) taking into account the inaccuracies of the printer. Took a little trial and error, but eventually it worked satisfactorily with a design that was slightly oversized, but also flexible enough to give way a little. I printed them with PETG for now. I will inspect them after a few hours of operation and might switch to polycarbonate if I find excessive wear. In general I would say, give them a tight fit. They will grind themselves down to the correct clearance. And use high quality lube. I use PTFE lubricant.

Dirty

Thanks for all your answers !

For 1- I talked about the linear bearing, the one for the rod. I used IGUS 40

And for noise, did you try closing the sides ? with polycarbonate plates ?

Thanks

Ah,...

The igus bearings are JFM-283239-20. They were about 4€ a piece. A total of 12 were needed, because I use double upper bearings to increase buckle resistance at large extensions.

I did close the side of one of the actuators with 2mm aluminum for comparison, but that alone didn't do much. The vibration is still being transmitted to the outside through the metal. Aluminum a pretty good sound conducter. It changed as soon as you touch the actuator with your hands, wich attenuates the oscilations.

I think for a second version of the actuators, I would use those 100x100 extrusions with a few strips of Dynamat or alike on the inside for sound deadening.



Dirty

Quick little update

I got the main frame welded for 340€. This thing is rock solid and can easily carry a single person with some control hardware.


These double nut blocks are on the inside of the aluminum square profiles to hold the aluminum plates.


Before I drill the holes I mark the drill point with a punching tool.


To be precise, I normally print myself a jig first.




Then the laser cut aluminum parts are screwed into the double nut block.


I plan on putting 5mm rubber pads between the aluminum plates and the square tube to decouple the main frame from vibrations from the actuators. It's unbelievable how much smoother and quieter they make the entire motion experience.

Dirty

Hey ,

Rig works fine!

I've had it running for a few days and did a few hours on it until I got frisky and ran the actuators into the end stops. One of the parts that connect the U-joint to the rod broke and severed the connection. With one actuator detached, the rig collapsed and a few more PETG parts failed.

The "fragility" of the printed parts turned out to be a blessing in disguise . If they hadn't snapped, the actuators would've bent. To be clear: the INITIAL failure was not supposed to be, and I have since reprinted the part significantly stronger. But all the other parts that failed as a consequence of that failed in the best possible way. They failed just before the metal bent.

After about two weeks the sim was back up and I was finally able to do more testing.


A bunch of things I learned about motion simulators in the process:

1. Using rubber pads (or something similar) to decouple the actuators from the platform reduces vibrations significantly, but is not as straight forward as it sounds. Too hard a material and it doesn't absorb enough energy. Too soft a material and the upper platform is no longer rigidly connected and generates movement where none is intended :-/ Sorbothane works OK, but is very soft and therefore requires a lot of surface area to carry the load. I had ~50cm^2 or ~8 square inches and it wasn't enough to hold the platform tightly. I will try some different rubber to see if that works better.
2. Filter tuning is everything! - OK, it's not EVERY-thing, but without proper tuning everything is nothing when it comes to generating motion for a sim. I thought tuning the filters would "improve" the motion somewhat, but it's more like: Tuning the filters "creates" the motion in the first place. I can imagine that many don't want to wrap their heads around filters and their interaction in the time and frequency domain, but it's definitely worth it! A few days ago I have spent literally half a day staring at curves, imagining how they change with changing parameters. Still a lot to tweak and tune, but the motion improves steadily.
3. Cars and planes are not the same thing when it comes to writing a motion cueing algorithm! It's not like they just need different filter settings, they need a different algorithm altogether. I don't want to bore you to death, but I have implemented a few compensations for false cues that result from flying at high G-load and high angle of attack. Made a huge difference in the end.
4. EMI from these AASD servos is a bitch! I somehow managed to mitigate its impact to an extent that would allow me to get a barely-OK experience. I put ferrite cores around the motor power cables and the cable to the headset. Without them, it would not even be possible to get a joystick recognized by the USB controller, let alone seeing anything other than a blank gray screen in the HMD.
   
   ...and lots of aluminum foil with an underlying copper braid for shielding.

Dirty

...oh,... and I just noticed that I haven't posted a picture of the actual thing here yet .

...rudderpedals still to come.

Cheers

I remember @Thanos made a vid where he replaced the original cable to avoid the noise. Did you saw this ?

Good to see you are making a great progress.

can you please post a video of that thing in action?