Showroom Sven's rig, 6DOF servo, car + heli

Sven is not a member of this forum but he's my friend and as we build the rig together I'd like to share some photos.

He uses the same linear actuators as I do for my bigger two-seater. The actuators look a lot shorter than those we can usually see on rigs built like this. But the reason is their better stroke to total length ratio. If we would add, say 200mm of empty tube where the motor is located normally it would look the same.

Interesting base idea. I had thought about doing that "3 spoke" approach myself, but not sure why I talked myself out of it. I guess I was thinking it would come across as less stable, but after thinking about it some more I'm not sure it would. You still have the same hinge line between any two points, regardless of whether you have material there or not.

Worst case I just planned to bolt my rig to the floor anyway, so wouldn't even matter a lot in the long run.

The spokes are super heavy with 160x80mm extrusions and 20mm solid aluminium plates. If had to buy the aluminium I would design it differently but Sven had lots of spare material in stock so he took what he got.
We probably have to add more weight or bolt the whole thing to the ground if it jumps around. But I'm not too worried about stiffness.

It's alive!

The actuators run super smooth. When they run fast the timing belts make some noise. We hope we can shield that by adding covers to the belt gears. We also found out that the pedal base can collide with the front actuators and the base plate at extreme yaw+roll or yaw+pitch down positions.

What are you using for controls? Aka drivers, controller, feedback?

For the first test we used a Beamicon2 CNC-controller. This allows only manual and pre-programmed movement, no simulation. But it's easier to check the limits and tune the servos. Later we will use my motion controller.
The servo drives are described here. The motors have internal magnetic encoders, so no need for external feedback. We use a hard stop at the bottom for the homing. The drives set an output if the torque exceeds 40% of the rated torque.

Slowly we are making some progress. Yesterday, we ran the rig with the new 6 axis controller for the first time. We tested some surge, sway and yaw movements which we haven't done before on the 3DOF rig, of course. We were very carefully because we haven't implemented the anti-collision feature. I have to make myself familiar with the new FlyPT Mover version which should support it.

Unfortunatelly, there is still a software bug that causes the actuators to move at different speeds. That leads to uncomfortable yaw components when doing heave movements for example. I have to fix that. And XPlane took very long to download on the new PC so we couldn't test with the real sim software.

Do you have any physical limit switches (aka for each actuator)?

I have thought about adding limit switches (rubber bumpers like used for garage doors) at the sides of the pedal area that triggers in the case of a collision. But if it bumps at high speed there's not much braking distance left.

Yeah! Today we did the first "hot" test: XPlane with FlyPT Mover, several flights with a Cessna and a R22 helicopter. We limited yaw to +/-10°, roll and pitch to 20°. With theese limitations we came very close to the actuators with the pedal area but haven't got any collision.

We tested with a LCD screen, first, and then with an oculus VR headset. With the LCD screen the motion is very smooth. However, we had big problems with the VR headset. First, the headphones stopped working after the first flight for no reason. Second, the VR software somehow seems to "sandbag" the PC. The motion controller frequently reported communication timeouts and repeatedly moved to the parking position. I increased the timeout from 20ms to 50ms which solved the problem but due to the unreliable timing we had much more vibration and jerking. But at least the differing speed problem I reported in post #7 is solved.

How did you determine if you were within the Cartesian limits?

I don't really know. I'm not even sure what you mean with "cartesian limits". We determined all parametrs by trial and error and with a nervous finger on the E-Stop button.

Allthough this can be called a step in the right direction and shows that all systems are working I'm still not really happy with it. The filters in FlyPT Mover are a mystery to me. I understand the basic concept but until now I only took one of the sample Mover files ("xplane_final_3-5-3_heli.Mover" in this case) and just played a bit with the gain values. The rig moves in the right directions but it doesn't feel as realistic as it should.

For example, in a longer coordinated turn where you just apply enough rudder so the ball in the spirit level is centered the rig should be exactly horizontal and only the horizon on the screen should roll. But there is a pose in the xplane_final_3-5-3_heli setup that includes the absolute roll position which causes a roll and therefore a false cue of being pushed towards the center of the turn.

I have to dig deeper and find out how the filters and poses work.