Proton Persil: v1.2 available 2008-12-05

Tomte wrote:
1. I will quickly move the model in MI
2. the driver shifts the balance a bit to the front, so I now assumed 38/62 (Z=-0.52), I could probably argue a tiny bit more.

1. I wouldn't bother with that, we have cars made both ways that work, and you'd have to redo the whole .car. The wheels are likely placed to combat the load shift...I've been sitting on the ktm results from the wiki formula for a about a year, so I may have described my method incorrectly.
2. On such a light car, it might do more...40/60?

Coming from the standpoint of where RL applies its forces...to the CoG point relative to the 0 point of the model, and hence the car. The axles are simply placed to deal with it, and don't need to be part of the equation, from my perspective. Through lots of r/r messing, i've found these types of values better simulate the experience than by pre-assigning wheel load.
4. I also have:
- Z inertia @ 400 based on your suggestion to raise it for stiffer springs
strength@ 15000, hooks the road well.
diff@ 125. R/R vehicles seem to need more to get the same effect in RL, probably due to the weight being on the wheels.
Front:
wheels.pos {-0.80, 0.34195, 0.93257}wheels.
maxSuspension 0.075
Rear:
wheels.pos {-0.8, 0.34445, -1.41244}
wheels.maxSuspension 0.08

load sensitivity 0

DANDY wrote:3. Based on LOA, I have -0.21142 at 38/62

what is LOA? Length overall?
The weight distribution is usually measured at the wheels. I actually found photos from race corner scales showing the load of each wheel of a 245 hp American Atom made by Brammo, which is about 60 kg heavier than the British original.

personality to spare
Running 1:18-1:20 on Mtn, 57 @LRP, 2:13.7@Rocky Pass

tested 081127b as much as I could. Good news is I can drive this build... as opposed to previous.
You guys know my level of driving so take these for what they are. I ran a few laps and whichever was my best I wrote down.
about three to 5 laps per car.

Black HSL Lime
Atom 2.36.86 58.86 57.38
X-Bow 2.28.80 58.96 55.46
Radical 2.44.01 55.05 55.54

DANDY wrote:
your values

load sensitivity 0

My main reason for raising the Z Inertia is a slower, better controllable load transfer. I guess this is what I meant with stiffer suspension. Until now I only raised it to about 20% above the calculated value, but I will give 400 a try.
15000 sounds good, dunno about the supsension friction. I found that lowering that to near 0 gives me the best chances to survive the hill and Lime Rock, but it makes the car quite wobbly. I'd like to raise it, 600 sounds where I would like to have it.
The loadsensitivity in your .car is set to -0.1 and -0.14.

One question though: any specific reason why you raised the front axle? Did you notice some improvements in drivability?

As you, I also thought that the model placement does not matter as everything is relative to it. Obviously, it changes the inertia calculations, but I thought that Redline would deal with it, given the own coordinate system of the car. So why do I feel a difference between the original undrivable version and the replaced one? Mystery.

Anyway, thanks for your help: I think I'm on the right path now.

SF, thanks for the testing, I hope I didn't make you staying up late. I still need to do some work to get it up to speed, now that I feel that I'm getting closer in the 'drivable' departement.

I've tried to catch up to the posts in this thread, but I'm still a bit mind-boggled by the speed at which changes are made. I think with build 081127, the car is much more difficult to read. With version 1.0, I always knew why the car went into understeer instead of oversteer in a corner, but now the handling seems much more clouded.

I think you should start with establishing a Z axis weight distribution, and then take the rest from there. I've seen sources that claim anything between 40/60 and 35/65, so I think your guess would be just as good as any data source. May I recommend driving around Buckmore Park for some low-speed handling tests, where sway bars and aerodynamics won't confuse you? I think that track may the best way to figure out what Z weight distribution you should have (and if you should modify the Y inertia). With those values established, move on to the values for the other axes (try watching the car being nudged in all sorts of directions around Lyngaas). Then set up the suspension, and set up the sway bars on medium-speed tracks like Lime Rock / HSL that don't have too much aerodynamic influence. Then, set the front/rear lift balance when driving on high-speed tracks.

The hard and important part of car testing is learning the sweet spots of all the different settings, then starting over with one value changed and re-learning the sweet spots and compare it to an earlier version.

Tomte wrote:load sensitivity 0...The loadsensitivity in your .car is set to -0.1 and -0.14.

a little ; goes a long way

Tomte wrote:One question though: any specific reason why you raised the front axle? Did you notice some improvements in drivability?

only to add suspension travel to near 3"

Tomte wrote:
As you, I also thought that the model placement does not matter as everything is relative to it. Obviously, it changes the inertia calculations, but I thought that Redline would deal with it, given the own coordinate system of the car. So why do I feel a difference between the original undrivable version and the replaced one? Mystery.

Did the driver placement change with it?

I did a lot more messing with it yesterday, not sure it got better from what I posted, but here's the path I ended up on....even worked in some loadsensitivity...seriously though, the car should handle well without, it should only be used as a seasoning.

1127 D2

DANDY wrote:

Tomte wrote:load sensitivity 0...The loadsensitivity in your .car is set to -0.1 and -0.14.

a little ; goes a long way

Hahaha, you did it again and I missed it again. I think I need to make all mistakes at least twice before I learn...

On repositioned model I also changed the position of the driver in the .car

C14ru5 wrote:

• sway bars
  
• lift
  

We have no swaybars apparently on the Atom.
Do you think this car produces any kind of lift to speak about? The air should go straight through it Then again, even the Caterham is producing lift, and that thing hasn't much body to speak about either.

DANDY wrote:
I did a lot more messing with it yesterday, not sure it got better from what I posted, but here's the path I ended up on....even worked in some loadsensitivity...seriously though, the car should handle well without, it should only be used as a seasoning.

1127 D2

Dandy, you're quite a purist, no swaybars, no loadsens, no downforce

I did some laps on Rocky Pass with D2. Unfortuantely, I cannot get a single clear lap at halfway decent speeds. The X-Bow does that track without braking sweat, I would at least expect the Atom to be within 5s of it's lap time.
True, Rocky Pass is a relatively fast track, but we designed it especially to stress suspension and handling.

Tomte wrote:no loadsens,

it's back in D2...which was called D1 in carname, just changed that... just did 2:20 at Rocky Pass, not sure how that compares.
I may need to re-examine the xbow now, I'm suspicious of myself.

I dont want to debate reality with the much brighter people than I working on this but.....

081127 D2 rides like its on a sheet of ice for me. It skids in curves at at 60 mph.... certainly that cant be right?

DANDY wrote:

Tomte wrote:no loadsens,

it's back in D2...which was called D1 in carname, just changed that... just did 2:20 at Rocky Pass, not sure how that compares.
I may need to re-examine the xbow now, I'm suspicious of myself.

I just had some sweet rides with what I tought was a rear loadsens on -0.07 on one of my versions... it turned out to be -0.7.... I guess I ned to back off that a bit
I think a RP time of 2:10 or there abouts is ok for the KTM, given how quick other cars are ont this track. 2:20 is definitely slow.
Come to think of it, 2:15 on Mountainside is a good time as well, faster than say Don's early saloons, 911 (I think), M Coupe, but not as fast as the really quick supercars.

Tomte wrote:Do you think this car produces any kind of lift to speak about? The air should go straight through it

Where there's turbulence, there's suction. The direction of the vector is yet to be determined, though. If anyone has a toy model of this car: Please put the car on your floor, position a fan in front of it, light a match and record a video of how the smoke moves around the car's body. We need that data!

Last one....2:12.8 fear-restricted@Rocky Pass
Changed suspension, heaped on the loadsensitivity, weighted the wheels, minor adjustments here and there...probably as far as I can push it. Throttle control is key.

I did not follow all development steps, to be honest...

For me the car is much more difficult than the one Tomte originally released. I understand that this was intended and we might have a comparabla class with the xbow now.
The 20081128D1 version needs much attention but is controllable. I like it - but its not easy anymore.

would need to do some timed tests, but won't do that right now...
for me a release i can find fun with...

alphonse

Dandy wrote:3. Based on LOA, I have -0.21142 at 38/62

I honestly think that the CoG position in your .car files is off. I'm certain that the CoG is dependant on where the model is placed in the 3D environment. As the wheel position is taken from the model in it's specific cordinate system, the CoG needs to adhere to that system.
That means in this case: in the old model, the one you are using, the front axle is at 0.93 and the rear at -1.41 (or there abouts). Therefore, the 50% mark is at -0.24. You placed the CoG to -0.21, which would mean a forward weight distribution of 51.5/48.9.
According to this topic, the weight distribution will be based on the axle positions in the .car.
The tread also sais that as long as CoG, wheel position and collision boxes are set accordingly, the model could also be places some 1000m out in Z direction without changing the handling.
This of course should also include the intertia calculations.

I will continue to use the replaced model

Code:


model "ArielAtom_Frame_v1.2.mdl"
interiorModel "ArielAtom_Interior_v1.2.mdl"
shadowModel "ArielAtom_Shadow_v1.2.mdl"
driverPos {-0.287, 0.34, -0.20}


with all the Z coordinated adjusted accordingly (even the lights). This is already done since 081127b. I wouldn't know why except for the changed inertias, but the Atom does feel different and a tad easier to handle.

There was the undrivable 081127 D1 I think
and now the modified 081127 D

This is almost as hard as the 911 ! With this build I cant do anything under 1.03 on HSL. The car feels ok but it seems as if it's always hitting a patch of oil on the road. Real slippery, I even managed to lose the back end simply by full throttle in third gear?

Dont know what to say but ... it's really hard to drive now. Same comment as before, at speeds of 60 MPH, does not seem it should be so slippery unless it's rainning.

SF

First things first: I have to thank Dandy, who spend quite some time trying to find a setup for the Atom. I really appreciate that (even if we differ on some points).

After a week of fiddling and trying, I could not get the 'no swaybar' option to work without some serious amount of loadsensitivity (>0.3). Even with that, the car just didn't feel right or interesting or fun to drive.
So, with a heavy heart I went back to the 'with swaybar' option. For me, it appears to be the lesser cheat with improved feel and understanding of the car. Call me weak...

Anyway, here's my last attempt. If I don't hear anything negative (which I can't discuss away ), I'd like to release that as v1.2. In any case, if one of you finds a setting that works on most tracks, I'm happy to incorporate it.

Dandy, I think the KTM is fine as it is, I don't see a need to change it . The way it handles is believable, the brakes are astonishing, but there is nothing wrong with that.

Atom 081204

Tomte wrote:I'm certain that the CoG is dependant on where the model is placed in the 3D environment. As the wheel position is taken from the model in it's specific cordinate system, the CoG needs to adhere to that system.

Isn't everything relative to the origin of the 3D model?

NoNameBrand wrote:Isn't everything relative to the origin of the 3D model?

Yes it is. I was just discussing the position of the CoG in one of Dandy's .car files.

081204;
This version is fun to drive but is pretty tough. Feels right on the edge... pretty safe.. and bam... spin out of control
It's not slippery like one of D's .car mods. Feels more like suspension making it tough. There is on twist on black forest, in the cliffs, that I spin out 7 out of ten times if I try to get top speed. I really have to slow down and the only thing that seems to happen different then otherc curves is there is a drop in the road.

Going uphill on limerock is also quite tricky... harder than downhill

I think it's a good match, will do better than Xbow on certain tracks, flatter ones. Just as Radical will do best on real flat tracks but will be damn hard on tracks like black.

Best time I got on Black was 2.34.94

I cant tell anymore when you guys are happy with your work or not.

RP in 2:08'58 in 3 laps. Well, surely is the better then old version. This new build approaching the hardest parts of the track with a sensible high speed. She has behavior and is not flat. She need to be drove carefully. Yes, I like it.

The latest version is much more like version 1.0 in its handling, which means that I like it.

The oversteer/understeer behavior seems to solely depend upon if the weight transfer is to the front (braking into a corner) or to the rear (accellerating out of a corner). I don't know if that's the definition of a perfectly balanced car, or if it's an indication that one variable may be out of proportion in the .car file. You seem to have tried very many approaches, so I think it's best to leave things as they are now that they work OK.

In both my incarnations of the Top Gear Test Track, I get times between 1:17 and 1:18.5 - Not the 1:19.5 of the Ariel Atom 2 300 driven by The Stig. First of all, you shouldn't trust that I've been completely accurate in my measurements and track layouts. Second, most Redline plug-in cars are much faster than their real life comparisons, sometimes as much as 10% faster. I take your 3% faster as a sign that you're within the right range.

Since I think this car handles well, if you are to do any performance adjustments compared to other similar Redline cars, then only do so by changing the mass and changing the body's rotational inertia. That's probably the most consistent way to do it, just like how FIA uses their weight penalty system to balance the field.

Thanks a lot for all your replies.

@SeeFlat: your assertion sounds reasonable. It's a mid-rear engined car, which will be tricky and bite your head off when you try too hard. Once the rear end loses grip, you will be hard pressed to control it with opposite lock.
There are certainly tracks where it has a hard time to match the speed of the X-Bow, especially everything with sudden elevation changes.

@DonaemouS: 2:08 is astonishing, I never manage below 2:13 on Rocky Pass. There is only a very narrow path at WALT and the following turn, despite that the road is so wide there. You really need to know what you're doing (which you do).

@C14ru5. The Atom is far from perfectly balanced. I mean, with the engine that far in the back and not much weight, I wonder how they get the car to work in real life in first place.
I read somewhere that the Atom 2 with the singular springs at each end was sometimes tricky to figure out. One did not know if the Atom would understeer or oversteer in the next moment. In this aspect, my car is easier to read and understand.
When the rear wheels get light under braking, the back wants to come around. When you accelerate and take load off the front tires, it understeers. Powerslides are hard to do or impossible, only as a reaction of sudden load transfer.
All that reminds me of the M Coupe... I agree that a way to adjust the speed/competitiveness would be to adjust the weight.

Anyway, I'm happy with what I have now. A bit more throttle response in corners (off-throttle/trailing) would be nice, but that might get in the way of keyboard players.

Expect the release soon.

Oh, one more thing: I hope that the KTM won't change! From what I read it is easy, somehow unspectacular to drive compared to the Atom or Caterham et al. I think this is what we have with the current car. The tests I read compared the KTM with the 300 hp version of the Atom, which is faster than what we have. That's why I made the 245 in the first place.