B7 Mura 990 Small lean angles create large lateral oscillations

[HornetMaX]

Anyway, I also think that the virtual rider is the source for most physics problems we have currently (apart from the missing swing arm pivot adjustment on the stock bikes).

But keep in mind that on a real bike, the pivot adjustement is only a few mm (+/- 2-4mm), I'm not sure that would make an huge difference.
Easy to test tough: stock 990, add/subtract 4mm to the pivot min/max. Not sure if it's worth though, as we'll have beta8 someday.

[h106frp]

Thanks for the info - I'm not trying to make a physics mod with this stuff, just playing around out of curiosity to see how it works.

The huge default value for spg1 is not good though, I am quite convinced that it causes the rider to overreact at speed, and the low value for kdamping1 causes the 'jagged' low speed steering - spg1 seems to be trying to compensate for this.  With a modest increase in kdamping1 (0.5) and reducing the magnitude of spg1  (-150) you can get really nice smooth low speed handling without the erratic (jagged) steer inputs.

The steering damper (damper, damper power) is bugged and as far as i can tell does not work at all in any conventional sense (for a steering damper) that i can fathom, trying to use it too calm high speed handling is a waste of time. 

It does encourage me that the virtual rider model is capable of doing what is required though, it just seems to need a bit of tuning to finish it. I think i have also come to the conclusion that the virtual rider issues make any attempts at tuning the bike physics very difficult as its difficult to separate bike physics and the virtual rider control loop.

@vin97 - totally agree about the rider - i altered the bike so that it was super stable under power and braking with really nice high speed handling, low speed handling then gets a lots of high frequency vibration, but this get magically fixed by raising the control loop gain but then the high speed handling gets a bit 'flappy' - but our bike model has not changed only the rider inputs and the bike suddenly appears to have poor high speed stability 

I know this is all very 'trial and error' and probably a waste of time but it has kept me amused and given me a reason to do lots of practice laps if nothing else. 

@MaX - starting to think that unless the 'dev diary' shows some work on virtual rider tuning B8 will not be much fun 

[Vini]

I think PiBoSo is doing that at the moment (uphill/downhill fix and general lean angle change).

Combined with the improved rear suspension simulation, we might get surprised by beta8.

[Napalm Nick]

I know this is all very 'trial and error' and probably a waste of time but it has kept me amused and given me a reason to do lots of practice laps if nothing else. 

Totally agree that its fun to learn about this stuff. My only concern is that if figures and changes are made to make the bikes work, but which don't reflect real life measurements and physics, then all we are doing is creating arcade bikes. Good for interim fun of course while the man is fixing the simulation but I think it is important not to get frustrated thinking he can fix it by just doing this or that (an unrealistic change).

Having said that you may know my stance on any simulation is that it will be totally impossible to reflect real life physics, so some compromise can (and must) be made.

Don't get me wrong, fiddling with this stuff is good fun and interesting to learn about I just worry that people start to think "well we can fix it so why can't PiB" and get frustrated   

[HornetMaX]

The problem is that all this stuff works together: the stabilizing controller, the yaw controller, the steering damper ... What one typically does when dealing with multiple control loops, is to try to separate them, i.e. having one controller acting over a frequency range that is different from the one of the other controllers. This greatly simplifies the tuning.

If you want to dig into all this stuff, I'd suggest to start removing the steering damper (Damper = 0) and the yaw controller (KYaw = KDamping0 = KDamping1 = 0) and try to play with the stab controller (spg0/1, spi0/1, spd0/1) first.

The steering damper is only supposed to help in dealing with wobble (wobble, not weave). Wobble is essentially an oscillation of the front, weave is more an oscillation of the rear (you can have a look here). Also, the stab controller should not interfere with the steering damper: I'm not sure that's the case today, as the stab controller seems to have no explicit rolloff (it doesn't cut its action at high frequencies).

The yaw controller is a bit of a mystery to me. I'd need to play with it more, but it seems to me it mainly help exiting a turn on the throttle (it avoids lowsiding when accelerating and/or picking the bike up).

[BOBR6 84]

Anyway, I also think that the virtual rider is the source for most physics problems we have currently (apart from the missing swing arm pivot adjustment on the stock bikes).

But keep in mind that on a real bike, the pivot adjustement is only a few mm (+/- 2-4mm), I'm not sure that would make an huge difference.
Easy to test tough: stock 990, add/subtract 4mm to the pivot min/max. Not sure if it's worth though, as we'll have beta8 someday.

lol yeah seen a pivot kit from yoshimura $266 for 2mm adjustment lol!!

[h106frp]

@Nick, one conclusion i have arrived at is we cannot fix it by tinkering, its down to PB. We can only throw values at the control loops but we cannot control the way they interact and this seems to be the problem area

@Max, sort of where i had got to, steering damper off, yaw off. I will have a go with kdamping off, it will be interesting how this impacts the stab controller as its seems at the moment to be the only way to get the rider to work on low speed turns.

@Bob Maybe it just means the pivot position is way out on the stock bike, revised location in B8 and we may only need a few mm 

[BOBR6 84]

Well whatever the reasons its still nice to have a play around and see what's what.. If anything it shows piboso is not far away really!

Obviously real bikes are ''complete'' lol

So yeah when everything else is done, moving the pivot 2mm should have the right effects.

It really is amazing the level of detail going into every aspect of these virtual bikes!

Puts things into perspective... Lets be honest it would be alot easier to make a bike in a game/sim ''feel'' realistic.. But to simulate everything is like piecing together a never ending jigsaw.

Im looking forward to beta8.. Especially the improved front end on banked turns.

[h106frp]

From my amateur dabblings i have pretty much decided the bike physics are handled quite well if the model is good.

The real failing is the rider, he currently struggles to discern between low speed steering, high speed turning and just steadying and guiding the bike (this is definitely missing - the VR seems to address all bike attitude changes caused by whatever reason with a steer solution however inappropriate i.e. I am making no controller steer input but the VR gives it a huge steer input and i would rather he just steadied the bar torque) when braking or accelerating hard.

You seem to be able to choose one area of riding that you can get really nice and the other areas suffer often in quite erratic ways.

[CapeDoctor]

this lateral 'oscillation' as i like to call it, i found to be very apparent in the (brand?)new MotoGP Yammie M1, for example - most significantly at low lean angles at relatively low speeds.
almost as if something in the physics is constantly causing a tank slapper effect at these points... come to think of it, maybe it's not as bad in b7, but b6 felt as if it was obsessed with doing tank-slappers constantly, as well.... 

[C21]

The yaw controller is a bit of a mystery to me. I'd need to play with it more, but it seems to me it mainly help exiting a turn on the throttle (it avoids lowsiding when accelerating and/or picking the bike up).

Indeed it is.
After testing a bit i agree that it does what you described.
I could not see another effect but changing value from 0-10 makes a stabilizing effect on the turn exit at throttle.
Testing on the Monza 2nd chicane. (yaw conroller=0, steering damper =0)
Set to 0 leads to a strange (better say an expected) behavoir on turn exit. The front shakes on a very high frequency from left to right (no youtube account...otherwise i would upload the video). It was not controllable hitting the rear brake.
Set to 10 decreased the shaking left to right on the turn exit but it´s still visible and leads to crash.
Normal value is 20-25 ?!

I agree on h106frp that the VR is a problem.
Better to say the interaction of the VR and the bike.
You can mess up real good bike physics with the behavior of the VR.

[BOBR6 84]

this lateral 'oscillation' as i like to call it, i found to be very apparent in the (brand?)new MotoGP Yammie M1, for example - most significantly at low lean angles at relatively low speeds.

Yep... the aprilia too! maybe all?

really bad at monza on the fast sections too where you have slight lean angle.. on the straights and at full lean angle its fine! 

[CapeDoctor]

BOBR6, yes, the Aprilia too - seems they all share a common physics atm.... 

and yes, it seems more to happen on straight sections - done a few laps at Mugello, and it doesn't seem to happen (much) on the inclines.

[yoshimura]

[HornetMaX]

this lateral 'oscillation' as i like to call it, i found to be very apparent in the (brand?)new MotoGP Yammie M1, for example - most significantly at low lean angles at relatively low speeds.
almost as if something in the physics is constantly causing a tank slapper effect at these points... come to think of it, maybe it's not as bad in b7, but b6 felt as if it was obsessed with doing tank-slappers constantly, as well.... 

Video or it didn't happen. I need to be 100% sure if it's wobble or weave. The remedies are not the same.

Testing on the Monza 2nd chicane. (yaw conroller=0, steering damper =0)
Set to 0 leads to a strange (better say an expected) behavoir on turn exit. The front shakes on a very high frequency from left to right (no youtube account...otherwise i would upload the video).

Get one, it's free
Or send me the replay file and I'll make and upload the video.