Universal 6DoF Recessed Simulator

news ? videos ?

Only once completed I will post again.
I think no more posting of development stuff.

Unfortunately the process of developing a product can seem very schizophrenic, to people on the 'outside'. Agree that often it can be better to keep things under wraps while you make design decisions and iron out the wrinkles . Look forward to seeing the finished product.

of course I understand, it's just that as a simu enthusiast, this new simulator makes you want to see it moving in beautiful video

Well the problem is if I upload half finished things, people keep on thinking it's final... And I prefer to work on specific problems, one at a time.

Anyway, I've finally finished a new motion controller now that I have some industrial servo drives that are pretty bullet proof, unfortunately also expensive. So we connect to simtools via STM32F4, then do some formatting of the data with a bit of filtering, then push out via CAN bus to the drivers. But we have to work within the limitations of the software and I'm trying a few things to get around that.

I think the problem is that with rotation and acceleration forces there is a different speed control required. The only solution with Simtools currently is that you have to bias it either to rotation or to force, so if we bias it to force for racing, rotation is a little slow and it can lag a little bit(keep in mind the video lags the capture in any case), see in the video when the guy hits me and the car lifts up, it doesn't move correctly, of course there is still some filtering happening to avoid huge crash movements so it's also interfering but even with the crash filter it should be a faster, but if we change the scale lower then it shifts up the small movements again. It has to be this way because if you look at cornering, acceleration etc. those are all what I feel mostly in the car. For flight we can bias it differently. I will have some encoders with a small OLED feedback display that will allow changing of these settings on the fly. Also the scale, because for F1 we have to scale the forces a lot more than for other cars, if using the F1 settings on the slow cars, they barely move. Doing it via simtools on a per car basis is not as easy as doing it with an encoder mounted in easy reach on the sim (or even in the steering wheel).

Excuse the quality of the video, I re-purposed a high speed security camera with a fisheye lens so that I can do the dual recording as suggested by noorbeast, which works a treat, thanks again!


If we push a little bit more on rotation speed then we get a lot more aggression, but this is something you can choose in Simtools, the firmware is the same. I prefer the movement above. It's not easy to see the small fast movements on camera.

it seems good !
for a formula 1, I prefer the fastest
how to push more rotation speed? the speed of the motors?
How many% axis for heave? heave seems rather discreet no?

I am of your opinion, a speed of engine or a speed by the software, different for the accelerations would be interesting.
we can also increase the values of accelerations in Simtools to reduce the violence.
if the resolution of the simulator is important, that the reactivity is good it can do this without too much loss on the quality of the effects.
I compensate for the poor frequency response of my Asynchronous motors by the Buttkicker

The video doesn't quite showing how rough it can get, but the data that comes out of the games is actually very poor, when I look at the data logs (see the plot on the previous page) the amount of movement sent over 1ms from the game isn't realistic, the swings are physically big enough to be larger G forces than possible, sometimes from 20% to -20% in 1ms. For me so far F1 2017 is completely unusable, I don't know if it's the game or the plugin. Having a slower acting motor has a "natural" smoothing, but I am experimenting now with with a much higher roll center for sway and surge, so with pitch and roll the roll center is underneath, with sway and surge it is above, it's difficult to do without IK math and it won't be perfect, maybe 80% as good as proper IK. And also I will experiment with a kalman filter to help with the abrupt changes in direction.

Yes increase the speed of the motors, right now I think we are limiting at less than 1000rpm at max speeds... at higher speeds it becomes too aggressive, at full speed, forget about it. You can't rely on the motor controller "soft start" because then you lose all small movements at the bottom. With the higher roll center I'm able to increase the speed and it will feel more acceptable, but the question is if we want it much faster, do we want to travel to max sustained G fast and stay there so meaning we start out at 1G or 1.5G then end up at 0.5 for the remainder of the turn, or do we want to build to peak at 0.5G then hold and back off as it is reduced, I prefer the scaled approach and not the peak and hold approach.

All forces are at 100% on at least one of the driving axis. Heave is all at 100% but reduced in tuning center. I prefer @ 100% peak on all forces because I get around 92% axis utilization with it. I don't use sway because it requires washout to be turned on to work correctly, and it only works on single axis output in this way for full 6-dof I can't get it to work satisfactory it screws with the rest of the motion, in the second video I think you can see some traction loss movement, so instead of sway I use extra1.

Speaking of F1, the seat came out great, sits great too!

Love the seat and your profile seems to be heading in the right direction. Have you actually ridden in the rig yet? It always surprises me how violent it often looks from the outside when I throw someone else into my rig - feels right though when I'm sitting in it with the VR headset on. As you may know, I got around all of Simtools and 'small' simulators in general shortcomings by fitting a Gravity system.

""All forces are at 100% on at least one of the drive axes. Heave is 100% but reduced in the center of adjustment. I prefer @ 100% peak on all the strengths because I use about 92% of use of the axis. I do not use the swing because the washout must be turned on to work properly, and it only works on a single axis output this way for 6 full dof I can not make it work satisfactorily it screws with the rest of the movement, in the second video, I think you can see a loss of traction movement, so instead of swinging I'm using extra1.""


can you put a screen of your Simtools settings?
It's not very clear to me.

I would test your approach

It's Quite simple - when you apply multiple DOFs to a single axis in Simtools, then any filters applied to even just one of those DOFs will effect all DOFs applied to that axis.

I rode a more aggressive profile with a low pass filter(youtube isn't taking the video of that profile), the driver was a little worse than me so you can imagine I spend most of my ride on the grass and it is a bit rough (probably the same as when I once went off on the grass in my kart a few years ago) but it's tricky striking the balance between getting high speed on the full length along with removing the noise at the bottom while still retaining enough definition. I'll try a few adjustments tomorrow and see, but right now we have about 30% more speed with about 5-10% more bump than the videos above.

I'll do so tomorrow...

I did not understand that in his message ...
he says that the forces are 100%.
I understand that the forces in Game Engine are 100% ... that's why I asked for an explanation, a screen of its configuration Simtools is more meaningful for me.
it's not always easy to understand, my language is French ...

The moves of platform are strange, they do not follow the video. There are no Surge movements and the movements are quite slow. I think you are setting movements while you are outside platform. When I sat down in mine, what seemed to be a good setting from outside had nothing to do with the correct settings. Nothing will give you more than a feeling on your own body.

Yes you are correct it will probably change a bit when testing from inside, I do not want surge movements or sway linear movements, they do not work well, it works better when we convert them to rotation since you can keep the force illusion longer, but the rotation is a bit different, I tried to make it some linear moving into rotation but it's only possible with IK, we get it more or less to do it, but not smoothly. Today a few people tested it at much more speed than those videos and said we can push faster still, so I will do that, but first we need to tame some overshoot on the servos.

I have in the past requested two sways and surges to allow some sway/roll and Surge/pitch mixing, but there are not too many members with 6DOFs, so obviously it was never a priority. If your lucky there are sometimes another sway/surge in the Extra1-3 (I built my own NL2 plug-in so I made sure it had this feature) which you can use. You can also get close by playing with Simtools settings. Watch the surge in my rig in the below video as it pulls back for braking and noses down as well.

Hope my explanation helps . FI I also have used BFF software which uses 6DOF math and DOF mixing/washouts - don't use it now, as I cannot also run all my G systems at the same time.

It looks amazing!

I always find it interesting how 2DOF is pushing you forward when braking, while 6DOF is pulling you back. And in the end, it has pretty much the same effect in your mind.

maybe consider adding three vibration motors.... just a suggustion. see the video at 0:37




Edit: Some earlier studies suggested that with certain configuration it may produce stable vibration on such 6dof platform:
https://s3.amazonaws.com/academia.edu.documents/33109909/Newdesign_for_a6dofvibrationsimulator_with_improvedreliability_and_performance.pdf?AWSAccessKeyId=AKIAIWOWYYGZ2Y53UL3A&Expires=1527714012&Signature=NwNTNbjnjBoI5ZNQy5e4ff8mrtk=&response-content-disposition=inline; filename=Mechanical_Systems_and_Signal_Processing.pdf

I’ve been eating this thread up. Thank you for your inspiring documentation and ingenuity. This is really awesome!

From what I’ve learned on your website, you use different actuator setups in your motion simulators:
Gear/crank, ballscrew (why not rollerscrew?), and inverted belt drives.

I understand you are designing your own actuators due to performance/value shortcomings of available products. Why not use those on all four feaatured models?

I would like to know how your inverted timing belt drives compare to ballscrew or gear/crank setups in terms of noise emission. Would be nice to see a video without the overlaid game sounds.

I personally would be very intersted in the upgradeable 3-Axis version.

One more thing. The discussion here about profiles and filters etc. is very much centered around driving sims. How about flight? Do you, in general, consider flightsim motion ‚easier‘ to set up?

Thanks again & greetings from Hamburg.

https://s3.amazonaws.com/academia.e...=Mechanical_Systems_and_Signal_Processing.pdf