G-SEAT

I've been looking at the position of the back myself lately and i've decided to go with a top mounted hinge so you get the increased pressure in your back but not so you get pushed forward. Sway will need to be on sliders though so it will be more complicated but i wont be adding that just yet. Since your bottom seat more or less does this already it will be about the same effect.

I'd be temped to do a hinge a bit lower like Tino has on his seat. It looks interesting.

Do your thigh paddles use the same cues as shoulder paddles? If so I might link the shoulder and thigh paddles and drive each side with only one servo.

No, the thigh paddles use sway only. The shoulder paddles use sway and surge. You could maybe try to blend them together somehow. I might say that if I didn't have the budget for both I would focus on making the shoulder paddles as good as possible and not even have the thigh paddles.

How is homing handled?

The motors bump into hard stops with current detection on startup. I would like to get rid of homing in a future iteration with an absolute encoder on the final output.
If you're sitting in the sim and have a power cycle it will home on you. I've got all the homing directions away from the user but its still not cool, even if its rare.

Been doing some Googling cannot find info on Thanos AASD board encoder input. Only see limit switch option. Absolute encoders would definitely be the way to go.

No thanos' board would not take care of this. You would have to add it as a custom solution. Probably with an arduino and as5048a or similar

On vacation right now but should have a pile of parts waiting at home to begin building my copy of your g-seat. I'll only use the peg leg portion for roll pitch and heave and will only use the seat itself to provide the lat and long acceleration cues. That being the case and once I have it properly tuned:

I should not need to use motion compensation, correct?

Sick Erik! I want to see your progress.

But you will still need motion compensation if you are in vr. Reason being the view in vr is fixed to your car. Your vr system will reference your room, and your sim will pitch roll and heave in your room. So the sim to the room will still need to be subtracted from your view.

If the motion is small enough then you wont notice the difference enough to matter (<±5°)

I would however recommend having lots of pitch and roll (±30° if your crazy like me) so you can feel driving up and down hills and banks well as the motion of the car through jumps in rally or curbs on the Nürburgring etc. For heave you dont need much because you can only really simulate high frequency which doesn't need much travel (10-20mm).

I could go on and on but will stop here.

I'll start a build thread once I get going. Plan to use FlyPT so hope that I can use his output for compensation via OpenXR motion comp.

Above you said you could 'go on and on' with details and ideals about your g-seat. PLEASE DO! I'm sure all of us following this thread would love it!

I guess I felt I started rambling about the way I thing we should all build simulators rather than the gseat specifically.

I could give an overview; this is based off 8 years of thinking about this and trying stuff.

Break it down like this:

Chassis
G-belt
G-seat
G-helmet
Tactile
Wind

For chassis I think I have the most controversial view.

First where our “industry” is today:

Most people, it seems, ignore rotation data and try to deliver surge, sway, and heave through whatever means they have available. Usually, they have chassis pitch and roll and maybe yaw as traction loss. So, they simulate surge with a pitch motion and sway with a roll motion, and maybe they try to do a cueing effect of tipping a user to get gravity to make you feel like you are accelerating in a direction. Unfortunately, this gives really strong false cues and that is why motion gets such a bad wrap and why so many people say “less is more”.

Motion cueing with gravity is cool for planes but crap for cars because things happen too fast. So if we forget about motion cueing because of the false cues then we are left with pitch and roll motions delivering surge and sway data with small angles. This works well and also plays into the “less is more” mentality because at small angles the motion is nearly linear. Especially when most sims center of rotation for these degrees of freedom located at or near the floor. That leaves the user far away from the center of rotation to pendulum around. When you use small angles you get small pretty much linear motion. In mathematics this is called small angle approximation. But the motion is super limited and can only deliver high frequency information to the driver.

I advocate for using large range pitch and roll and yaw motion with a center of rotation at the center of gravity of the user. Send the actual 1 to 1 pitch roll and yaw data into those motions. Now you can use your rotation motion to give the actual vehicle rotation information as well as accurate gravity vector information. This will allow you to feel the suspension when launching off the line, the jostle in roll when your vehicle jolts up a big curb on the Nurb, or climb or drop down steep hills. I have built a sim with the center of rotation at the center of gravity and it’s a big ass deal!

Of course this means you must get surge and sway through other means. I currently have a pitch, roll, heave 3 dof sim(with the center of rotation at the center of gravity of the driver) plus a continuous yaw dof and I literally have no surge and sway information coming through my chassis. Zero. You will also notice the gseat isn’t on there right now due to other factors.





My next sim I will build a stewart platform style sim with more range of motion for pitch and roll as well as a full 360 degrees.

So with the pitch roll and yaw information covered, plus the stewart platform giving us linear surge sway and heave motion (to the tune of 25 to 50mm of travel I suspect) we still have a gaping information gap for the driver. Where is the low frequency (or long term) acceleration data???

Well that must come from the g belt, g seat, and g helmet. Those tools will give us the correct information and a very natural way – especially the g helmet because your head is low fatty body (low squish) and is the only thing cantilevered outside the car (seat, floor, pedals, wheel).

Personally I am most excited by the g helmet. I think it offers the most potential for excellent data transfer to the driver. But for now I will leave it for a rebuttal because I have been typing for long enough and a lot of this information is best shared through discussion.

Check out this video. The strange looking seat is an actual 3d scan of my rig and I use it to get an understanding of size and form.

The scan is animated with telemetry data and shows 1 to 1 rotation of the vehicle. There is no translation motion right now even though there would be in the next version of the sim.

This stage in richard burns is interesting because it has something like 28 degrees up and down for pitch. Imagine what that would feel like during a drive!

Based off 11 years of thinking about this and trying stuff as well, I do agree with your analysis!

I love your infinite Yaw and your center of rotation at the pilot's center of gravity, it's bulky but so better.

Other subject but related, I like to sort a progressive trajectory for a simbeginner in respect of cheaper, easier & more efficient immersion ratio in the following order: FFB wheel, bucket seat in a rigid static cockpit, tactile, G-belt, actuators (Gseat or 3 DOF)...
I wanted to try Wind simulation but I really dislike draughts, so...

I'm eager to see your Gseat on your rig! That will be awesome

Hi, I was looking at the lichuan kits in the past because they seems interesting and affordable but they are not using an AASD-15A. How do you interface this motors with you pc?

By the way I found this when searching for g-seat, not sure if how it's made works for your requirements but seems a really nice and compact solution.

Oh I use AASD15 drives with the thanos setup. I only ever bought gearboxes from Lichuan. I'm migrating things more towards odrive though as I get more into direct drive custom stuff.

Yeah that is one of two manufacturers supplying g seats to the sim racing world. They use rc servos so pretty low power compared to this thread. Also the relatively flat panels are pretty limited in what they can deliver. Much better than nothing and lots of people love them. Barry has a good video on the GS5 from back in the day.

Glad to finally see the rig and looking forward to the next iteration!

Im a little curious on the limits for this one though as i plan on going this route before i also go for a steward platform when money becomes available. When you have full pitch or roll wont the distance between the points shrink and create some tension on the actuators? im not 100% sure on this but i imagine it. Have you done any testing to see how far you can push it?

The bottom frame that holds the actuators can flex a bit to account for the angle change.
This was an experiment that worked but but has a drawback. When the gseat is installed and you have strong sway motion there is some wobble. Any wobble is a false cue so im keen to change it all for more stiffness.
The range of motion is no problem and I can get about +/- 15 degrees in pitch and about +/-20 degrees in roll.

I will likely move to a rotary direct drive stewart platform for better stiffness and larger range of motion.