Well, with Prototype cars and any cars with high-aero, you WILL experience a overcorrection snap because of aero forces. Slides will only be 'holdable' at low speeds in these cars, the faster you go, obviously the more intense the aero effect.

When your front wheels are no longer slipping and pointing the the correct direction, to an extent it's like just turning the wheel sharp and hard when travelling fast as the front wheels start rotating again it will spit you out.

The problem with iRacing is the lowdownforce cars and tin tops still behave this way.

Todd,

Regarding the axis of rotation. It's mostly watching from Far Chase view. What looks particularly wrong to me is where someone gets out of shape at high speed on a straight. They countersteer, get a snapback, and the car seems to rotate on a coin. If the snapback was that violent, I think the opposite rear tyre (outer one after the snapback) would then probably slide too, so you'd get the pendulum/fishtail effect around the front end. The rotation would then be slower. But that rear seems to grip and turn the car as though it was normal cornering. I'd have thought that if the snapback is that violent, then the rear should slide afterwards. If not then it wouldn't be such an issue to bring under control anyway.

BTW, slightly different physics point... Once you've got a difference in lateral force between front and rear, then the angular acceleration produced is a function of I (second moment of mass or 'moment of inertia') about axis of rotation, right? (Torque =I x angular acceleration). How does the programmer calculate I? You need to integrate the equation of the volume of a known mass about an axis, surely? Anyway, if that's wrong, mightn't it explain why a touch from another car, or blade of grass causes a sudden angular acceleration! I've know idea how you guys work out these things.

With the MX-5, you can get a slide going, let go of the steering, and the self aligning torque will generally point the front wheels in the right direction. But good luck catching the snap that way because it's ridiculous IMO. There are only a handful of guys doing faster times than me in the MX-5, yet I've spun out at Okayama at really low speed, and I've seen the situation developing, but you might as well be holding your dick for all the use steering is. Unless you jump on it early and overcorrect, and then you've got the snap to deal with. You can usually muddle through with the brakes as well - which shouldn't really work should it locking the front tyres in car with ABS?



Dave, yeah, when I've seen a really fast RFB lap in the skip it's either been a case of virtually no steering involved in the turn-in, i.e. the driver's made the slightest suggestion of a turn in, and as soon as he's begun trailing the brakes it's rotated in on it's own (usually needing countersteering to control that as well!). Or, the extreme opposite; tons of steering in order to induce understeer so that the front doesn't grip much either to balance it out. Bottom line is that if the front is gripping and the rear comes loose it's bloody hard work.

BTW, the oversteer whilst releasing the brakes I referred to isn't jumping off the brakes, it's just a fraction faster or slower. It may not happen other than when cornering on the limit. So maybe it's to do with this front sliding a bit but then releasing the brakes a *touch* too fast cause them to grip all of a sudden. I notice it doing RFB (where you can't feed in throttle to help); I'm find I'm wanting to smoothly trailing off the brakes but something tells me to pause the process because it'll snap loose if you let it off any more. It all seems to revolve around being too icy generally with the tyres though. They should suck the road up a bit more. If you've seen the RF2 vids, those tyres look good and grippy.



Re grass. ..Yeah, if you get on the grass in 3rd gear stationary, add a bit of steering, touch the throttle to momentarily lose grip at the back, and you can then keep the rear wheels spinning forever (like you were flooring the throttle) just by literally touching the throttle every second or so to keep them going. So again, any rotation driving, and it's either fast and frantic correction or lock the front.

The $64K question is whether the sim physics is actually more difficult than RL, or someone like Todd's mate simply not adapting to the sim environment.

To be honest, I think the grass is iracings biggest flaw.

Pat,

The apparent axis of rotation you seem to be refering to is highly subject to the camera's focal point. I've played with this a little bit and (to my eyes anyway) the car will look like it's rotating around whatever point the camera is focused on. Focusing the camera on the rear bumper makes it look very different than if it's on the front or somewhere else. So what I'd be looking for here are the magnitudes of the yaw velocity and yaw acceleration to spot a problem rather than trying to figure out what point the car is rotating around. I'm not even sure how exactly somebody would define a yaw axis off the top of my head really, with both translation and rotation happening at the same time.

When I suspect something is up I crack open the iRacing telemetry and usually find that what I'm looking for is really happening. I wrote a lot of posts in the iRacing forums showing things like yaw moment reversal at high slip angles that I'd been complaining about for a long time, but couldn't clearly illustrate to anybody until the telemetry stuff came out, regardless of how blatantly obvious it was to me after one minute on the skidpad in this car. This is what was illustrated in my video earlier, along with the other video of the car doing a quick 180 that doesn't look unusual to most people. The same problem can be seen there too, but is tougher to spot. Most people miss it and say there's nothing wrong, but they aren't looking closely at the yaw acceleration and velocity which can both can be seen to speed up considerably after the throttle is released and the slip angle grows a fair bit. The yaw velocity roughly doubled in situations like this which was highly questionable and is what people felt was wrong with this car. All most people see is the car pointing this way, then that way, over so much time and that's about it. Sure enough, the very first test with telemetry confirmed exactly what I felt was happening all along. The next patch it was fixed on the GT and HPD.

Unfortunately they won't show slip angles in the telemetry. I don't blame them one bit really. I probably wouldn't show them either.

Moment of inertia: In VRC Pro we use data straight from the manufacturer's Pro-Engineer design files for inertia moments, so in our case it's integrated from the mass properties density, volume, and location of each part, right down to every little screw in the whole car (these get pretty small in RC cars). I don't calculate this myself. It comes out of the engineer's software. They send me giant readouts with inertia properties and I plug the relevant values into my model. The car itself is only part of the picture though. I also get all the drivetrain parts including all the rotating shafts and gears, tires/wheels, and so on. I've done it this way for ten or eleven years now.

I doubt most sims get that good of data, but the idea is to try and get it from the manufacturer if you can. The engineers are fairly likely to have good computed data on it. If that's not possible to get, there's a database online somewhere that has something like 100 or 200 cars and trucks in it that were measured and/or compiled by the NHTSA. If I'm not mistaken, this is done using a "swing test" or something similar, where the principle moments of inertia around the yaw/pitch/roll axes are measured directly from the real car, like weight or anything else would be. When I play around with regular sized cars in my hobby sim, I refer to that database a lot and just pick a car that's close to it, then make a semi-educated guess at massaging the numbers a bit. I think that paper has been posted here at the LFS forums many times over the years. I've got it here on my drive so will look up the title if you're curious. It's a free paper you'll find quickly.

Another option is to try and estimate it from the wheelbase, mass, and other properties of the vehicle. One approach is to estimate the inertia moments with boxes of different sizes and masses positioned in various locations. Maybe one for the engine, another for the chassis, etc.. There's a gentlemen by the name of Brian Wiegand, a retired Northrup Grumman "Senior Weights Engineer & Mass Properties Handling Specialist" (that's the title in one of his papers, "The Mystery of Automotive POI Values"), that has done a lot of work in this area. Funnily enough, he contacted me recently about an article I wrote online asking for attribution information for a research paper he's working on, and we got to discussing this very topic. If you're curious you might see if you can find the paper online somewhere. It's something the SAWE charges for though so I'm not sure if you'll find a free copy around. I wasn't able to so had to ask him directly.

Another paper of his that he pointed me to is SAWE #3490 Automotive Mass Properties Estimation which deals with this very thing. This is what you might try to do if you don't have measured data from a swing test of the entire vehicle or really good CAD data like I use in VRC Pro.

I'll try to address the rest of your post in another reply later. Just wanted to touch on the rotation axis and moment of inertia stuff while it was on my mind.

OK, that sounds quite reassuring then, about moments of inertia. Thanks for that info Todd.

Yeah, I'm very aware of the illusion created by chase views focussed on the centre of the car, and in all sims the movement looks odd from that point of view. I *think* I'm taking this into account. I'm seeing this movement based on simply staring at where the wheels are on the road before during, and after the process, and that sometimes they seem to follow pretty much the path I'd expect if the car were cornering normally at 10mph with grip instead of violently snapping back at 100mph. I'd expect another bout of sliding at the rear after that. I'm open to persuasion on this apart from anything because it's horrific if they've got this wrong, but it still looks to me that it appears on rails all of a sudden. It could, I suppose, be a consequence of the sim producing stupid/unlikely inputs from us drivers i.e. in RL you'd almost always get the front wheels straighter quicker, after the snap. To my mind it comes back to inertia again though; *visually* the rotation seems to accelerate too fast after the snap and I can't see the rear tyres holding on perfectly afterwards with that much mass moving that fast sideways.

It is worrying that you immediately spotted the yaw velocity issue with 2 cars but DK didn't. It does rather blow to pieces the notion that these cars undergo precise tested for realism, and by pro drivers too. It seems more a case of stick it out there, and if enough people complain about something, then change it next update.

Trouble is, I think generally the pros probably drive within realistic limits but sim drivers take the physics to extreme places - where there be "singularities" and the like.

BlueFlame: I've an image in my mind of someone driving on the grass in a smooth circle, and DK confirming the lateral g is what it says on a table on his 'puter, but no one actually noticing that if you lose a bit of grip for a moment at the rear then you spin around in circles for ever!

On the Skippy, yes, the real one does the "trailing brake oversteer". The instructors just call it TBO. It is incredibly noticeable at Laguna Seca in turn 2. If you carry tons of brake as you go charging into the corner, you'll notice it as you get into the middle part after the first apex with the brake still applied. Snap off the brake at this point and you'll have a big moment.

I don't know exactly what causes it but it must be linked to the rear weight bias, because Porsche 911s do the same thing, just much less extreme.

And yeah, the Skippy was better during the first NTM release. Now it's really unpredictable. Almost random. Chronic understeer one moment, chronic, repeated snap oversteer the next - both with seemingly on or off characteristics.

iRacing's latest work is really hit-and-miss with me...

Pat,

You probably are seeing something a little funny then. I tend to notice these oddities in a little different way because I'm looking at different things, probably. The point of rotation of the vehicle doesn't enter my mind. That doesn't mean it isn't wrong, I just might be spotting it as an odd change in yaw velocity or something like that. A "rotation axis goof" might be a simultaneous symptom of the underlying cause of that, but it's not a programming thing where they got the basic physics calculations wrong. It's all in the tires.

One thing that might be at play here, and this is something that bothers me about a lot of the iRacing cars with an exception or two, is that they seem to have largely gone the route of rFactor and friends with regards to over the limit tire force behavior. I.e., how forces change beyond the peak lateral force slip angle. Most of the cars drive like there's a great deal of drop off. Not nearly as much as rFactor and it's brothers had when they came out, with the exception of the street tire cars in GTL which to me were just marvelous, but still too much. Thanks to the telemetry we can now see pretty accurately how much there really is. In the cars I skidpad tested with telemetry, it's somewhere between LFS and rFactor in this regard.

This drop off in force after the peak can add significantly to what you're describing, especially if the balance in drop off between the front and rear tires isn't matched up well. You end up with behaviors that feel strange, like a car that understeers strongly up to the limit, then a few degrees beyond that becomes neutral steer ("drift"). At that point countersteering doesn't really do anything so it suddenly feels like a different car. It doesn't blend into the drift and feel natural, or respond to small countersteer and throttle inputs. You have to countersteer further and further while nothing happens, then steer a bit more and wham, it starts to react. The higher the stiffnesses the quicker it reacts. At 5 degrees slip angle you're fine, but at 5.2 you're just going along for the ride. Sound like the MX5 or any other cars to you?

On top of this, too much drop off makes the situation much worse because when you first begin countersteering in a slide, the car not only doesn't start to slow down its rotation as you'd expect it to, it speeds up and you just spin out even faster than if you had done nothing at all. So you wind up with this tiny little window of steering that you have to snap the wheel to as fast as you can in order to catch a slide. Too little and you accelerate the spin. Too much and you spin the opposite direction.

Want a quick way to check if there's a lot of drop off or not? Next time you slide a car, steer into the spin instead of out of it. If the car straightens up right away it's all because you reduced the front lateral force a bunch by increasing slip angle. The more you feel that, the more difficult the car is to catch probably. There's a lot that I like about iRacing so I don't want to sound too negative about it, but one of the things I don't like on some of the iRacing cars (the high end slick cars primarily) is that it's much easier to save a spin by steering into the turn instead of out of it. Countersteering just makes things worse on many of the cars. In the 842,305,201 in car videos I've watched, I have never seen anyone do this in a real car. Why not? You'd think somebody would have figured this out by now. It's a much safer way to catch a slide, apparently.

Live For Speed got it very close (if not perfect) and does this very well. When you overstep the bounds and countersteer a little bit the way you would in a real car, the yaw velocity slows down as you'd expect rather than the exact opposite. There isn't a tiny little window of acceptable steering angles that work to catch a slide. As a result, Live For Speed is much easier to drift than other sims are, and it feels (to me at least) much more natural to dance around the limit with. There isn't some switch that's on at 6 degrees and off at 6.001 degrees. Maybe some people say it's too easy, but I think they're just comparing it to other sims where it's much harder to do. I see this easier sliding and ability to drift the car in a natural feeling way as a plus, and a sign that LFS handles this area better than even iRacing does with one of my heroes, Dave Kaemmer, at the helm.

When I ask real drivers about it the majority seem to say that drifting is too hard in their favorite sim and should be much easier. If LFS is "too easy" to drift, I'd say "good, it should be a lot easier than it is in <insert sim name here>." Rock on, Live For Speed.

People need to pay attention to what exactly the comparison is being made to on that. Another sim or experience from real driving?

Anyway, point being that this excessive drop off is probably a large part of the oddness you're seeing with the Nascar vehicles (among others). I haven't spent much time in those cars since the NTM came out so can't fairly comment on how it is now. Next time I do I'll remember to check and post something here again.

To be a little fair and balanced, iRacing's SK Modified is remarkably better over the limit than many of the other cars.

I almost never drove a rear wheel drive car, but testdrove my friend's stock RX-7 and was drifting it inside of 20 minutes. Since then I've owned an MX-5, and if my MX-5 was as lethal as the iRacing one I'd be dead several times over.

My MX-5 gets snappier with the R888 tires on, but still easily driftable and 'incidents' are fun, not lethal.

Quote from BlueFlame :

To be honest, I think the grass is iracings biggest flaw.

I agree the grass feels funny. It's similar in a way to the behavior I showed in my video where the rear might have more grip than the front until you get in the grass where the situation reverses. If a car is dramatically pushing on asphalt, why would it not keep doing that in the grass? I don't really know how that should work, but agree with you that something about that feels funny. I care a lot more about onroad behavior than what happens in the grass though so don't make much fuss over it.

Quote from Hyperactive :

To me it seems iracing have the knowledge about tires and would know how to make them better but they also still have this ideology that driving race cars fast or getting over the limit is really really difficult. A bit like the gtr game developers. They had the necessary data but they did not want to believe it and adjusted the numbers to get the difficulty up where they thought it should be. I think the ideology that driving race cars is difficult sits very firmly still in the minds of the sim developers and while there are people who are not afraid to put the real data into the sim and believe it there are still many who adjust even very good data to make it harder just because "that's the way it should be".

I agree and have frequently thought the same thing. Before I made my video I played around with tires in my hobby sim to see what it would take to make a car do what the FGT and HPD were doing at the time. It turned out to be quite a bit more difficult than I thought. I was surprised, and yes, "difficult to make it do that" would be the right phrase to use here.

By the time I'd done it and made the car worse than the FGT was by intentionally making a lot more drop off at the rear tires than the fronts with the sole purpose in mind of getting the car to accelerate the spin after a certain slip angle, I realized that I'd almost have to do it on purpose and spend a fair bit of time tuning it to get the "right amount" of acceleration going into the spin. I came away thinking there was at least some chance this was done intentionally because so many people expect cars to do that. That could be wrong, but it was really harder to do than I thought it would be. I can't see myself accidentally making a car that pushes as much as the FGT did exhibit that behavior. I'd probably have to spend hours to make it work that way.

I also agree with you about the Corvette: I played with it for a little while (on the OTM) and dropped it pretty quickly. That one was a disappointment. I tried it very briefly (maybe a lap or two) with the NTM after a very long period of not touching the car and to me it seemed to be better. I'm not too sure about that car though. Need to play with it more.

Brake release oversteer: Haven't forgotten about this one yet. Still need to give it some thought and maybe play with it in my hobby sim to comment.

Quote from atledreier :

I almost never drove a rear wheel drive car, but testdrove my friend's stock RX-7 and was drifting it inside of 20 minutes. Since then I've owned an MX-5, and if my MX-5 was as lethal as the iRacing one I'd be dead several times over.

My MX-5 gets snappier with the R888 tires on, but still easily driftable and 'incidents' are fun, not lethal.

Great to hear. Thanks for sharing.

This is the same sentiment I see echoed over and over from real drivers (including the friend in the Donkervoort) and not just in iRacing. Harder = more realistic is nonsense a lot of the time. That was true when we went from Pole Position type games to things like Grand Prix Legends, but not so much these days. A great deal of the engineering involved in tire and vehicle design is on aspects of control-ability. You don't want vehicle responses to be outside of certain ranges in certain situations. They need to be engineered to handle in a way that drivers feel comfortable with. If you have to baby the thing around the track just to keep it pointing forwards, it's time to take it back to the garage and call in the tire engineers to figure out how to improve it.

Quote from jtw62074 :

Next time you slide a car, steer into the spin instead of out of it. If the car straightens up right away it's all because you reduced the front lateral force a bunch by increasing slip angle. The more you feel that, the more difficult the car is to catch probably.

I've not heard much about RL drivers using understeer-type behaviour in the way it can be used generally in sims. Could be it destroys the tyres. and also that outside of the US it's mostly RFB. But it doesn't seem to have the same advantage. In RL they just correct with countersteering, as well as adding throttle and coming off the brakes. It does seem there's a common factor in a lot of weird behaviour to do with the over the limit grip falling off, and once tyres start sliding their continuing even though the cause has been removed.

But there's the other thing - how difficult would it be in RL to control a car on the limit without your inner ear and other physical sensations telling you when it's starting to yaw etc? That does make it difficult to compare 'difficulty'.

MadCat360: interesting about the Skippy. Do you think that's with or without the front sliding a bit due to excessive braking. Because that would cause sudden oversteer if you come quickly off the brakes in any car when the front is sliding and the rear is already loose.

Quote from Postman Pat :

But there's the other thing - how difficult would it be in RL to control a car on the limit without your inner ear and other physical sensations telling you when it's starting to yaw etc? That does make it difficult to compare 'difficulty'.

I think the "missing g-force" feel factor is not nearly as important as people generally seem to suggest. In my work on VRC Pro with top level racers (even many who are not so top level) in the world, the amount of feel they describe in the car is surprising. When analyzing some of their setups in telemetry during testing recently with the new 1:12 scale electric car, they were adjusting things by one degree slip angle or so. I saw three out of four testers set up their cars to turn at one slip angle within +/-0.5 degrees of each other, saying that 1 degree less than that was "too pushy," while another guy was a little bit outside this range. In our simulation it makes an especially interesting case because the inputs they have really are identical to the real thing. No FFB, they use the same radio as they do in real racing, and there's no g-forces to be felt in either case.

That's just one quick example, but this is all purely visual feedback they're using just as we use in big car sims. So personally I think this missing feel stuff is generally much overrated, especially with experienced sim racers. When slip angles in a big car sim are tightened up only a degree or so, many people can tell right away something was changed.

Quote from jtw62074 :

I think the "missing g-force" feel factor is not nearly as important as people generally seem to suggest. In my work on VRC Pro with top level racers (even many who are not so top level) in the world, the amount of feel they describe in the car is surprising. When analyzing some of their setups in telemetry during testing recently with the new 1:12 scale electric car, they were adjusting things by one degree slip angle or so. I saw three out of four testers set up their cars to turn at one slip angle within +/-0.5 degrees of each other, saying that 1 degree less than that was "too pushy," while another guy was a little bit outside this range. In our simulation it makes an especially interesting case because the inputs they have really are identical to the real thing. No FFB, they use the same radio as they do in real racing, and there's no g-forces to be felt in either case.

That's just one quick example, but this is all purely visual feedback they're using just as we use in big car sims. So personally I think this missing feel stuff is generally much overrated, especially with experienced sim racers. When slip angles in a big car sim are tightened up only a degree or so, many people can tell right away something was changed.

One F1 team tested the yaw sensitivity of two sets of F1 drivers. Those that had karted at a young age, and those that hadn't. The experiment used a rotational chair used by the airforce if I recall correctly. Anyway the test concluded that drivers who had karted since a young age were more sensitive and even more so than fighter pilots who use the chair for training.

So there is some physiological evidence of sensitivity to yaw being a real thing. However, how important that is I don't know.

Very interesting. Thanks for sharing.

I didn't mean to suggest people can't sense movement and that it's not a factor at all, just that there have been lots of discussions over the years I could point to where it's rather overblown imo. For instance, seeing yaw moment reversal in a sim does not require seat of the pants feel. It can be spotted visually, yet in the discussions on this the argument that 'you can't feel g-forces so you don't really know' was used. There are plenty of examples of this type of thing.

Not trying to argue, just clarifying what I was getting at a bit better.

Probably works both ways. E.g. Todd recognised something visually wrong with the sim, that I expect an F1 driver looking at it wouldn't register. But put an F1 driver in a real car and they'll unconsciously feel subtle changes in balance and yaw very early before anyone's eyes could pick it up and that's how they can drive a car on the limit.

Quote from jtw62074 :

I got involved in a really long thread on this issue at the iRacing forums. The thing that I was pointing out was that the yaw moment in steady state cornering was pretty strong understeer. As the rear slip angles increased, the understeer moment decreased slightly, then decreased more rapidly at higher slip angles, and eventually reversed directions. I.e, understeer moment turned into oversteer moment which should not generally happen.

I'm not quite understanding the issue here, so I'm going to prod at your post a bit of you don't mind:

steady state cornering was pretty strong understeer.

- Okay...

As the rear slip angles increased, the understeer moment decreased slightly, then decreased more rapidly at higher slip angles

- Isn't this expected? As the rear slips more than the front you get less understeer?

and eventually reversed directions. I.e, understeer moment turned into oversteer moment which should not generally happen.

- It seems to make sense though? From steady state you increase speed leading to higher slip in the rears (definition of oversteer?) then eventually gets to a point where the car is yawing into the direction of the turn.

Can you explain what situations and how to replicate what you're talking about in game?

I'll try to explain it in more detail later. Here's the video illustrating it:

http://www.youtube.com/watch?v ... DskLoU9NMCUmLAQOK5xmabvji

I don't have a problem with this happening in the rain on a near neutral car, but in the dry this is a no-no I'm afraid.

todd have you tried this with the clutch to make sure this isnt caused by something wrong with either the diff or the engine braking?

No, I didn't. It shouldn't really matter.

iRacing pretty much fixed it on this car by adjusting the tires after this in the next patch.

Quote from jtw62074 :

I'll try to explain it in more detail later. Here's the video illustrating it:

http://www.youtube.com/watch?v ... DskLoU9NMCUmLAQOK5xmabvji

I don't have a problem with this happening in the rain on a near neutral car, but in the dry this is a no-no I'm afraid.

That clears things up. Wasn't what I had in mind and definitely looks awkward.

Quote from jtw62074 :

No, I didn't. It shouldn't really matter.

well it does look a bit like what the car would do if youd pull the handbrake so the effect could theoretically come from way too strong engine braking

Quote from Postman Pat :

MadCat360: interesting about the Skippy. Do you think that's with or without the front sliding a bit due to excessive braking. Because that would cause sudden oversteer if you come quickly off the brakes in any car when the front is sliding and the rear is already loose.

All of the rear-weighted cars I've driven (basically two different generations of 911 and the Skippy) exhibited a sort of lock-up like effect when hard trail braking was used. It was worst in the Skippy. When I was doing braking drills in turn 11 at Laguna Seca, it was entirely possible to get a lock-up like turn-in effect without actually having locked tires! I could see the fronts were still rolling, yet even with hard braking (which would pivot my MX-5 like a top) the thing just wouldn't turn.

The brake gives some kind of stability to the Skippy. The more you use, the more pushy it gets. The car turns best with no throttle and no brake.

I feel this front-locked-but-not-locked effect is a by-product of rear weight bias somehow. And it must be what is producing TBO.

In this video, you can see TBO the first time I go through the Andretti hairpin at about 40 seconds. It is not a big moment but it happens just as I fully release the brake in the middle of the corner. Incidentally, you will also see power-on oversteer on the exit which I have always had trouble getting in iRacing. The real car drifts easily in 1st and 2nd gear.

http://www.youtube.com/watch?v=Sk-RECJHkxQ

That was also the first school I ever took in the car, so I was making those mistakes a lot and going slow.

Very, very much looking forward to hearing Todd's impressions on TBO.

Quote from Shotglass :

well it does look a bit like what the car would do if youd pull the handbrake so the effect could theoretically come from way too strong engine braking

If it was engine braking, the behavior would manifest itself right away at small slip angles rather than just large ones. In the first part of the video where I'm blipping and releasing the throttle, the car straightens up when the throttle is released (understeer yaw moment). The trouble is that this fundamental behavior reversed course once the slip angle grew enough. The understeer moment decreased, passed through 0, and then changed sign.

If the car would accelerate into a spin like that at all slip angles rather than just large ones, engine braking indeed would be one thing to look at. It seemed there was a cut off point or critical angle where the behavior changed. I plotted out telemetry showing this happening too where the angular velocity actually doubled at some point from the beginning of hitting this "critical angle" until it started slowing down enough for the car to nearly stop. The trouble is the tires would twist the car to the left to straighten it, but then actually reverse and twist it to the right.

The first video of the car doing the quick 180 shows this too, but it's not obvious unless you're watching the yaw acceleration closely. There's a period there where the throttle is released and the car not only keeps on spinning (that's not a problem), but the spin briefly gets faster. The vid is so short though and it looks so much like a regular half donut it's easy to miss, so I made the other one to show it more clearly.

Ok, back to comment a little on releasing the brake and getting oversteer finally. I didn't do any testing in my hobby sim to look more closely at this, so am somewhat speculating.

What comes to mind primarily is that this is something that ought to be improved with brake bias. Madcat, you talked about how using heavy braking was like locking the tires without actually locking them. In other words, the lateral force decreases on the front tires as you increase braking. This happens with almost any decent amount of braking, except very light braking where the reverse is usually true. So for the sake of discussion, let's call the area where the lateral force drops with increased braking "moderate to heavy braking." Sound good? Ok..

You don't need to lock a tire for this to happen, or even be operating at the limit of the traction circle/ellipse to see this reduction in lateral force come in to play as braking is increased. The same goes for the rear tires of course. One thing that happens under moderate to heavy braking and cornering is you're losing lateral force at both the front and the rear tires. I.e., the yaw moment contribution (the torque that twists the car) decreases on both. If you get them to drop the same amount so the yaw moment doesn't change as braking is increased, the car should drive very easily into the turn with no surprises regardless of how much braking you're using. That's a very well built car though. Most aren't that good. This is where stability control systems shine.

On top of this the forward weight transfer is influencing the forces too which complicates the picture quite a bit. With one wheelbase to center of gravity height ratio you might get understeer, with another you might get oversteer. So it's not easy to point to a single cause that applies to the case of every vehicle (real or simulated) that displays the seriously annoying behavior the iRacing Skippy does in this situation.

Anyway, if the front lateral force drops more than the rear (more precisely, if the yaw moment contribution of this lateral force goes more toward 0 than it does at the rear), the car will understeer under braking, which matches Madcat's description of the real Skip Barber car's behavior.

The opposite is then true of course if the braking is moved more toward the rear. The rear could lose lateral force more quickly than the front. In that case as you increase braking you get less understeer, and if you go far enough you're effectively using an e-brake to spin the car like Shotglass described. So at first thought it sounds to me like with the Skip Barber car you could improve this with a brake bias adjustment. The trouble then though is that you might be making a big compromise in straight line braking where ideally you might want all four tires to peak simultaneously. So there's a trade off to be made in this if ABS or stability control isn't allowed which theoretically can overcome this very problem.

What about releasing braking? If we start with the first example where the front lateral force was lowered more than the rears under moderate to heavy braking by the forward brake bias, then it would follow that when you release the brake you'd get all of that lateral force back and the front end would tuck in more again.

However, this explanation doesn't adequately address the big spins you tend to get in the iRacing Skippy. When you release the brake, the car doesn't just return to the normal amount of "no throttle and no brake" steering. Instead it goes dramatically toward oversteer. I tend to think the difference there might be in the damper rates, something along the lines of what Shotglass was suggesting. This area of vehicle control needs some fairly heavy analysis because so many things are at play, so it's tough to make a generalized "here's why it happens" statement that points to a single cause. Some things have easier answers than others...

Maybe I can give the Skippy another drive and pay more attention to this, maybe look at telemetry a bit, despite how much I dislike this car since the last time I drove it.

As to whether or not it's "because of the rear weight bias," the best answer is probably a not very helpful "yes and no." The rear weight bias largely determines what brake bias you need to use to begin with. When you then match a set of tires to that car you'll get either understeer or oversteer under braking while cornering. It's really a matter of design choice (and success in making it do what you meant it do when you designed it!) A real car can go either way. So it's a combination of all of these things at play here.

If I had to pick one thing though, my gut says it's more about the brake bias than anything else. People that are fast in the car are using brake and gas at the same time. The only reason to do this that I can think of is so you have a dynamic brake bias. Under braking by itself you have more rearward bias that, in the iRacing Skip Barber car, overcomes the push you get in the real car under moderate to heavy braking while cornering, but this is too much rear bias for straight line braking, so in that case you use a healthy does of throttle to shift it forward. Some people are good enough at this to adjust the brake bias with the throttle all the way from the straight right up to the apex.

A way to really find out what's happening is to plot out the yaw moment contributions from not just all four tire forces, but also their lateral and longitudinal components. So you'd be looking at 8 different plots really to try and analyze what it's doing (easier said than done). If I worked for iRacing and my job was to engineer these cars, I'd be begging Dave K. or somebody to plot this data out live so I can see what's happening while driving. I do this in VRC Pro. Then for example you can point at exactly how much the front right tire's braking force (longitudinal force) is contributing to understeer, and how much the lateral component is doing just the opposite. Do that for all four tires and see how they change during that little transient where you release the brakes and it should give a better picture of what's happening. I wish it were simpler than that, but in this case it's probably not.

I've often wondered what the real Skippy does, so I'm really happy that you got into this discussion and explained what it feels like in the real car, Madcat. Thank you.