FlyPT 6DOF/Stewart/Hexapod Interface for linear and rotating actuators

Xplane doesn't include an "On Ground" flag in its available output data like FSX does? That's what I used as well as the getting the aircraft's weight to indicate how much to adjust heave when on the ground.

Its even better in xplane than in fsx because instead of using the entire flight model twice you can act on landing gear forces without altering the flight model.

I am substracting the current aircraft weight from the landing gear force.
All you need is a HPF on it and the result is spectacular

Just wanna say :

1-2mm backlash is measured at the end of 100mm crank (servo arm) but yes it could be huge in all 6 legs together.
I made some video of this jerky behavior, but as you said it is the FlyPT interface topic, so I share them on AMC1280USB topic, maybe I get some help to reduse.

Amazing, can't wait to try it

Don’t see any video of slow jerking.
Do the test manually and post the amount of filtering you are using.

The video is in the AMC1280USB topic
https://www.xsimulator.net/communit...imulator-controller.10517/page-25#post-176680

Hi FlyPt ,

First of all thanks for your amazing job !! I build a 6dof and until now I was using an another software to move my sim . Decided with the help of my friend hexpod who encouraged me a lot to test simtools , he convinced me also to test you fantastic job and development .
I've got a 6dof motion rig with rotating actuators (cranck arm ) .
I just tested until now the whole interface , not in motion with a game connected but only all basic movements roll , pitch , sway , surge , yaw , heave .
All movements are ok and top . Surge and Sway are flat ! Top !
Hexpod helped me to setup your interface thanks to him (I use an amc1280usb ).

I noticed 2 problems I would like to report and to let you know !
1- when check the box lock in the lowest position , the sim goes right in a lowest position . OK . After I kill the signal from the amc1280usb (amc1280usb feature , no setpoint send to the rig ) , the sim goes back to the minimum rig position (cranck nearly vertical , the sim goes down on its own weight ) . But after when I uncheck the box and disable the kil switch the rig goes up and doesn't stop to the middle position .I don't know why . I closed the soft and launched again and it's ok to test all movements .

2. When I use the box with the sliders interface to put them again to zero after playing with them to move the sim rig . Everything is ok for all movements except the yaw . When I click on the interface to set the slider to Zero the sim goes up quickly on heave direction . It's like a heave setpoint was given. When I did this the sim is ''online '' ready to move and the sliders are near to zero . Didn't test to click on zero box when the sliders are on their max or min position.

That's all and once again thanks for your amazing job !

Below my setup with your interface

https://drive.google.com/open?id=1PeqNGppol5NrgQNP7mPKQ4XTbiFgZrO0

for information my sim rig :

Thank's for testing!

For #1:
Made some tests, and I don't see anything wrong. You can enable the serial debug and see what is sent to the amc1280usb.
I don't have one, but before turning on the kill switch, did you give some time for the amc to boot?
I have to think better what could be the problem, maybe @Thanos could help here.

#2: Are you sure that was the yaw slider? They are one bellow the other and I made that mistake many times.
I can't reproduce the problem. Could it be the way the actuators are setup... There are so many variables here.

Sorry, I'm not doubting you, just want to test everything so I can identify the problem to solve it.


EDIT:
I used your measures, and I'm seeing a problem in the lowest position...
Give me some time to test it better.

Ahah no problem for doubting !!
For many years I work on my sim and all stuff ( wheel etc ... ).
I know how is complicated to report something and to be understandable and also for the super guys like you who develop fantastic software / hardware etcétéra !!
Will test and report here . Maybe not now because on holidays in the next days !
Thanks
And forgotten to say that also often made mistakes trying to describe the problem and without the help of the diy community and developers unable to solve the problem!

The vid is cool man!

Just Add the some sway on the roll for centrifugal force. You have to roll a bit your rig when you are in the curves

Yes it’s programmed like this but difficult to balance / mixed both effects !
I programmed like this !
From zero to an estimate value ( depending of g-forces and resized from data game ) only sway programmed and up to this estimated value roll and sway mixed to increase the lateral g-forces feeling.
I would like to have a range of g-forces feeling depending the car driven eventhough i know that the telemetry gives g-forces according car power and accelerations / deceleration.
And I noticed that with the VR headset your brain is more sensible when you tilt the sim rig. The feeling that your body is tilted is increased with the headset . So if you use too much roll or pitch for lateral or longitudinal acceleration in order to increase the sensation of pressure on your body using a VR headset , your brain notice this ! It’s what I noticed for me , and I don’t know if it’s the same for orhers.
It’s the reason why one day I will try to build a g-seat in addition to my sim rig !
I think g-seat + 6dof would be a next level of feeling!
Even with a big 6dof with 500mm of travel you are not able to reproduce for exemple a long curve with lateral go-force -You can tilt the sim but you could reach only 1G and with which angle ? And from which tilt angle your brain could not be cheated !

Without any doubt a proper g-system is a game changer. The problem at the moment though is there aren't any available that are "proper" without having a military like budget. However that's about to change soon and will be announced here first so keep an eye out .

ok will keep my eyes open

@pmvcda ,
One ignorant question:

What’s difference between a DEMA and a „second order“ filtering ? Is it the same, like a filter over the same filter?

I'm afraid of answering ...
So, let me read something first to be sure I'm not saying something stupid
Where are my old books and notes...

Hell!, I'm going to take the risk:

DEMA is two 1st order filters with the objective of getting a faster response.
It's a multi pass filter (fast, so good for real time).
You keep the "bad" attenuation from the pass to no pass frequency.

Second order filtering has more "definition" between pass/no pass.
Bigger the order, better the definition.

What I mean is that if we want a cutoff at 100 hz, an ideal filter would let pass everything to 100 and nothing beyond that (infinite order ?).
In first order, we might start loosing at 50 hz, up to 150 hz from pass to no pass.
Bigger the order less this range (simplistic/ultra basic way of saying it).

But I might be wrong...

It’s absolutely great we finally have someone who’s understanding the capital use of proper filtering.
I am sure you agree, that’s the key to get into a convincing result.

Although the result is already spectacular with ema dema I still have a small issue.

It’s mostly about the longitudinal acceleration in xplane while the plane is on ground and deals with the friction while braking.

Having a chance to chat about it with Austin, he did his best to tweak his ground model and statuated that the physics there correspond to what he is expecting.

As I don’t have any knowledge about this topic I will try to present my issue in a very simplistic way:

precisely on the moment when the plane stops, we have a opposite longitudinal force going up and down with a pretty big spike.

Although your washout and smoothing do the job in a very coherent way I still have a pretty big „kick“ especially when I put the surge on the pitch axis.

Do you see any other filtering method which could “round” or reduce this huge spike and keep the other smaller details present?

Here is the curve:


Trying to achieve what I want with your band pass but somehow not getting it.

That’s why I was thinking about “second order” filters.

But in thinking I might let you do because obviously, you are much better in it. ;-)

Thanks

EDIT:

I don’t know how deep you want to go into those details. Maybe, beside you ema dema tema, you could consider, reconsider, other various approaches like butterworth or chebyshev if you see a benefit to do this programming effort for a better final result.

Thanks a lot for your fantastic job.

It’s mind blowing what exists once you open a book.
I am pasting it here, maybe at some point you will find relevant to dig into it.

https://books.google.de/books?id=rmO5BQAAQBAJ&pg=PA39&lpg=PA39&dq=butterworth+vs+dema&source=bl&ots=GQaQMUKDOl&sig=ACfU3U3LRFPAiZC29d82iVWSxHgdUuGFeA&hl=fr&sa=X&ved=2ahUKEwjTmqHvgsThAhWOZlAKHST7CQIQ6AEwDHoECAMQAQ#v=onepage&q=butterworth vs dema&f=false

Ok I am stopping spamming.

Best

You are not spamming!

I can take a look at other filters, but they need to be fast.
I have the code for non exponential laying somewhere, but they are not going to add anything to the result.
I might add them just for testing, but it might generate more confusion than help.

I will look for some efficient 2nd order filters. But I don't want to guaranty anything .

Right now I have a big mess with inputs on the new interface version.
I'm making it modular and trying to make it work for remote and with games (like XPlane I think) that need to receive info to send back what we want.
Also, I want to have multiple inputs/sources if we want to (some games use it) and allow the input of more parameters.
It might not be visible in the new version, but I want that change made internally.
Should be in small steps, but this one is large, and I'm also working on a kind of auto calibration for the next version.

Might not look interesting, but what I'm making right now, might solve the problem of XPlane, because in the future I will be able to treat inputs in a separated way, and mix them in the end.
When I say treat, is apply filters. When I say mix, is mix the effects to obtain a result.
So mix landing gear info with aerodynamics info generates a result. If XPlane allows (still have to take a better look at the telemetry).

For xplane, I strongly recommend you take the main telemetry from “motion platform stats” which’s adding the longitudinal approximation of “center of rotation” unless you want to apply same formula internally to make it more crossover for other games.

What would be good it is the similar approximation for the side axis as well (which xplane doesn’t have for the moment) which would shift the center of rotation from “between the seats” to the capitain seat.

The landing gear force you can find under the checkbox 66

As I said, Austin bug correction will go public in the near future, if you want to test it right now I can provide an unofficial build.

Let me know if you need it