Dirtys 6DOF, AASD Servos

Hey,...

do you have any further information about this? Maybe a link or a tutorial?

I'd be happy to do a test row to compare the dampening properties of different mixtures, but I think the results will be unique depending on rig weight, rigidity, geometry, etc.

Dirty

Some funny thoughts:

TIMBER!

It wasn't me!

Last words: This printed joint can take anything!

But to keep it simple:
It was a hell of a ride and it`s a real beast if unleashed!

Thanks Dirty

McMaster-Carr appears to have nitrile isolation pads ranging from durometer 48A to 80A, with load capacities from 50 to 1500 pounds per square inch. So there's probably something out there solid enough for your use.

Maybe urethane rubber ?

https://www.smooth-on.com/products

You can 3D print the form of your choice and make your experiments.

Tuto :

Finally had the time to make a video of the rig in action



There are still a few more ideas around the motion cueing algorithm that I want to implement, but those are details on the outer edges of the envelope. The lion share of the "core" motion cueing code is finished and I am quite happy with it.
In my day job, I have spent close to 1000 hrs in level D sims, and as far as I can tell, the only difference to what I am experiencing here is the range of motion (quantity) in the translation channels. The motion itself (quality) is identical!

Smoothness has increased A LOT, and that was direly needed I have to admit Now with rubber padding and the right grease it is almost on a level of smoothness that I would call "customer ready". IMO improvements in that regard are only possible with much higher quality ball screws that would end up somewhere in the 400€/pc price range, compared to the 30€/pc that I have been using here.

Happy rig building everyone

Dirty

Hi is www.cncshop.at nothing fore good ballscrews

Like your rig, specially the low noise on the ball screws!
Is that all because of the grease?
I have to make something to mine.

Excellent !

It would be good to compare if the impact on touch down from xplane is similar to the rig output. It seems pretty soft visually.

If you plan to do such vids in the future please think to synchronize a phyphox experience. It’s not so much additional work and it’s so much more informative.

Thanks and congrats. Great job !

I guess they are plenty good for ballscrews !

Question is, are they good for AFFORDABLE ballscrews?

TBH, I have two hearts beating in my chest: On one hand I'd say we don't need those expensive super high quality/ high precision ballscrews. For a motion sim, it really doesn't make a difference if positioning is off by a millimeter or so. That means we can easily get away with cheap Chinese products (SFU 1610 from aliexpress.com)
But on the other hand we DO need smoothness and quietness of operation (at least for flightsims). Which brings us back to those 400 €/pc expensive super high precision stuff. :-/

In the end you get what you pay for,... and it all depends on what you are WILLING to pay for.

I, for my part, can say that I am happy with how it turned out. I did have to jump through a few hoops here and there, but that was all part of the learning experience that I so greatly enjoy .

BUT,... if this were ever to grow into a commercial project, the SFU1610 from AliExpress won't cut it for sure.

Cheers,.. Dirty

Thanks! That's a good idea.
Is it just a screen capture of the phone? Or do I have to do something special?

The touchdown was indeed rather soft on the rig, considering how I was slamming the aircraft into the asphalt . I could make it more sensitive in the heave DOF, but it would mean that I might more likely run into the endstops in-flight.

Possible solutions:

1. Use a signal compression. It would do the job, but if you over-do it, would also make the motion feel "the same" no matter what you do. Also brings up the question: Shall a compression be used on input-level, DOF-level or actuator-level?
2. Use air/ground information to change the filters. It would also be able to bring some interesting features, but it would bring up a similar question: Do I change the input gains, filter settings, output gains,... etc. AND,... switching gains is a big NoNo! You would have to fade them over, and even that would bring some headaches with it.

These considerations are what I meant with "details on the outer edges of the envelope". For now, I will stick with tuning an algorithm that performs reasonably well over the entire operating envelope of the sim before I dive into that rabbit hole of "living filters" or "predictive cueing". I think there is a lot of potential in those ideas,... even if it might just be the potential to spend another year coding stuff that eventually ends in the trash

Cheers,... Dirty

I think most of the improvement came from the rubber pads. I used them on both sides of the aluminum, so there is literally NO metal-to-metal contact anymore. Incredibly cheap but a very effective measure.

The thicker grease then added some dampening on top of that. I would love to try out an even higher viscosity, but I am just not willing to spend another 300+ € on grease *facepalm*

I mean, those two measures complemented each other nicely, but looking at them through an economic lense, they couldn't be further apart on the cost effectiveness spectrum! Definitely try the rubber first!

Btw, my voice was recorded through the headset mic (Valve Index). My mouth was about 10cm, from the mic while the actuators were around 90cm away. That might distort it quite a bit. Ya know,... remember the whole sound-dissipates-with-the-square-of-the-distance thing?

They are still very well audible. I can make a recording that captures the room audio better.

Dirty

Hi
Youre ballscrew is c7 precicion class yhis means 0.05mm ballscrew backlash, and tou stay with 0.1mm precice over 30cm.
For more precicion you need c5 screws, further its steel quality and seals from nuth thats the differince between
expencife and chinees ones

Probably the best flight tuned 6 DOF that I have seen to date . Yes, I had this heave issue in my old 6DOF, got around it now in my current sim with two independent heaves that work on different Tuning Center and Simtool GameEngine settings. The main heave moves the whole seat base (in my new sim it will be the whole seat/rig), while the more sensitive heave drives the seat base paddles. There are also the constant heave cues that are being conveyed by my harness systems - Ie when you are in a turn the thigh harness pulls you into the seat base to simulate the increased gravity that you would feel through the turn, with the base seat paddles also gripping you to help. FI I also have multiple surge and sway actuators that can be individually tuned. Similar story.. When I tell friends to lift the VR headset during a turn it surprises them that they are actually level .

Oh, FI my cheap ball screws that I originally purchased for my 6DOF are very quiet now after a few years of use.

@Dirty

The noise reduction doesn't so much as to come from the density or material but from the waffle pattern contact surface.

I don't think the quality of the ballscrew will make much difference as you are practically using it vertically, almost all the load will be on the lower contact point of the screw groove.
All ballscrews will wear, even if it is more hardened or more precise . You can't have 2 hard surfaces (balls and screw) one of them has to wear, usually the screw looses out.

I am sorry to say OZHEAT you need hardend steel upon hardend steel, its the only one that last with grease like a normal bearing.
Hardend steel upon softer steel wont last a day

Here's a thing about those parameters that I think a lot of people are not aware of. Some parameters (e.g. Pn115) are not effective unless you set Pn258 to 0 (default on my drivers was 1). From the driver manual:
"For automatic gain adjustment, the mechanical rigidity setting consists of the following 21 types. In setting the gain adjustment mode (Pn258) is 1, the mechanical rigidity level (Pn259), will be based on the servo gain parameter setting table to automatically select the servo gain (position loop gain, speed loop gain, speed loop integral time constant, torque command filter time). At this point, gain parameters such as Pn115, Pn116, Pn153~P156, Pn196, and Pn197 are not valid in automatic gain adjustment mode."

Hi @Dirty !

Now in this thread, as I am doing my research on the full parts I need while I wait from vendors to deliver.

Do you think having the servos UNDER the rig would have problems? I think having shorter "noisy" cables would help... right? I would replace the power cables to the motor for shielded ones, but I think shorter would also help...

Maybe an enclosure grounded to the motor ground could work/help?

And another "casing" as a guide for cables with aluminum metal duct (like @Dschadu mentioned)

I did this quick model to explain...

Let me know. Also maybe @Thanos could give me some feedback if time permits

I wonder if Aluminum is a good shield. I google it but could not find that much straight information about it...

Thanks all!

Oh, sorry! Forgot to reply...

You're right. Some parameters are not effective in automatic gain. And also 110, 111, and 112 (smoothing parameters) depend on 109 (smoothing mode). There are plenty of parameters that depent on other parameters to be set to activate a certain mode.
Because of this I always made a test to see wether the parameter is even effectice. If small changes didn't yield any effect I tried bigger changes until I noticed a substantial change in behavior.

None the less, in the end I was unable to get rid of the slow speed stuttering completely with servo controller settings alone, but the rubber pads did make a significant difference.

Cheers,...