LSD ramblings

[nomis3613]

Hi,
Interested to discuss the effects of the LSD? Read on!

For the LSD, I couldn't think of many tests which could be measured in numbers. But there's a few things, see the attached spreadsheet. I realise it's highly confusing, bear with me, I'll explain it as I go. For all tests, a strength of zero means the standard differential installed (the lowest setting for an LSD is 5).

-----------------------------
RWD CARS- ACCEL SETTING
The test car was a Mercedes 300SL (gullwing). This was chosen because out of the factory it wheelspins horribly, so it is in desperate need of an LSD.

This is the simplest form of LSD tuning. You just want to stop the inside rear spinning up in tight corners. The first thing I tested was straight line acceleration, to see if tightening the diff would improve straight line traction (see test #1). It didn't.

Now, onto the important test: wheelspin out of tight corners. Test #3 (at Apricot Hill) is the maximum speed the car will wheelspin to. For example, with the standard diff the Merc would wheelspin all the way to 205 km/h if the throttle was held open. Increasing the accel setting meant it regained traction at lower speeds (eg earlier).

So increasing the accel setting enabled the inside wheel to stop spinning earlier, as expected. A little bit of accel greatly reduced the wheelspin, then the results plateaued at 110 km/h. This is because below 110 km/h, the engine has enough power to spin both wheels (which the LSD can't do anything to stop).

What the numbers don't show, though, is how the car feels at higher accel settings. I was expecting more understeer, since the diff is opposing the forces needed to rotate the car. However, what I felt was actually better turn-in as the strength is increased with no side-effects whatsoever! Strange... (but useful...)


-----------------------------
RWD CARS- INITIAL SETTING
Same tests and car as the accel setting and the 0-1000m times (test #2) were also were not affect by the initial setting.

For the "tight corners wheelspin" test (test #4), the results were similar to the accel setting. Slightly more initial strength was needed (compared with accel) to get the same result. But the feel was totally different. It was horrible! It gave more understeer in general (consistent with the scrolling text talking about "the car being difficult to turn"), but also entry oversteer when turning-in at full throttle. Also, when the rear did finally let go, it was very sudden and uncontrollable. Yuk.

-----------------------------
RWD CARS- DECEL SETTING
Just like the accel setting, I was expecting the Merc to understeer more with increasing LSD decel strength. Again I was surprised! To me, increasing the LSD decel strength felt like adding negative front toe. Power understeer (aka exit understeer) was reduced, but at the expense of turn-in. I can understand why turn-in would reduce (as the car turns-in, the inside rear needs to decelerate slightly, and LSD decel would oppose this), however if anyone has any theories about why it is reducing power understeer, I'm all ears!

-----------------------------
TBC: FF and 4WD
Hmmm... this post is already getting a bit TLDR, so I'll leave the FF and 4WD cars for another day. Feel free to agree/disagree/ask questions, I'm really interested to hear your thoughts.

Simon

 

[Parnelli Bone]

That Merc is a good car for this test...you're right, it does "lose it" very easily out of corners without a limited-slip unit. Also, it's notable that the real-life Gullwing had a swing-axle type rear suspension

http://en.wikipedia.org/wiki/Swing_axle

which had some wicked problems with high-speed stability. I believe PD took this into account because the Gullwing feels horribly nervous at speed with a stock suspension. Putting even a default sports suspension will cure most of this. This goes for the Triumph Spitfire and the Ginetta too, I think...

I agree about the better turn-in you noted with stronger accel settings on this car, too. The 300SL is definately in need of help from the time we win it as a prize.

 

[F68U]

increasing the Decel strength will generally enduce an amount of oversteer into the corner. Essentially it is applying a compression brake to the rear wheels (the same as a heavy rear brake bias) hence the car wont feel like it is understeering out of the corner as the approach angle into the corner will already have an amount of oversteer.

Try increasing the rear decel on a high performance car running R compound tyres, you will find yourself entering a corner and the rear end wanting to swap around with the front.

 

[Greycap]

I can understand why turn-in would reduce (as the car turns-in, the inside rear needs to decelerate slightly, and LSD decel would oppose this), however if anyone has any theories about why it is reducing power understeer, I'm all ears!

Compared to a looser deceleration value the car won't turn in like it used to and you'll have to enter the corner slower in order to make the turn. As a result you have lower inertia forces trying to wash the nose out during the exit - simply said, you're going slower and thus having more front grip available.

 

[DK]

Interested to discuss the effects of the LSD?

Yeah, I heard they planned to give that to alcoholics to wean them off alcohol. Which means hobos are going to be very different people; "Excuse me sir, can you spare some change, I got a unicorn to feed."

 

[nomis3613]

That Merc is a good car for this test...you're right, it does "lose it" very easily out of corners without a limited-slip unit.

Thanks. Also, I can't deny that it's in my garage cos it's such a pretty looking car!

I agree about the better turn-in you noted with stronger accel settings on this car, too. The 300SL is definately in need of help from the time we win it as a prize.

Cool, glad to hear I'm not imagining things! Before I get too excited, I spose I should remember that this effect could depend on the car...

increasing the Decel strength will generally enduce an amount of oversteer into the corner...

Hi, are you talking about GT4 or real-life? Could be down to driving styles, but what I felt was less exit understeer, but slightly less turn-in.

Compared to a looser deceleration value the car won't turn in like it used to and you'll have to enter the corner slower in order to make the turn. As a result you have lower inertia forces trying to wash the nose out during the exit - simply said, you're going slower and thus having more front grip available.

Aha. Just wondering, is this also how negative front toe works to reduce exit understeer?

Yeah, I heard they planned to give that to alcoholics to wean them off alcohol. Which means hobos are going to be very different people; "Excuse me sir, can you spare some change, I got a unicorn to feed."

Haha! I was sorely tempted to make a very bad pun for the title of this thread. Clearly my subconscious got the better of me, though! Hope you unicorn had enough fairy floss for dinner...

 

[Parnelli Bone]

Cool, glad to hear I'm not imagining things! Before I get too excited, I spose I should remember that this effect could depend on the car...

It does.

Haha! I was sorely tempted to make a very bad pun for the title of this thread. Clearly my subconscious got the better of me, though! Hope you unicorn had enough fairy floss for dinner...

Hey I never saw a unicorn! I tried to fly out a window, though, while a green-headed alien chased me quoting the Bible's Armageddon verses!

 

[nomis3613]

Now for some thoughts on LSD settings for FWD cars....

-----------------------------
FWD CARS- ACCEL SETTING
The test car was a Honda Civic with Stage 2 turbo and SM tyres. With approx 270kW through the front wheels, this is a good example of an overpowered FF car struggling to turn horsepower into forward movement!

For straight line acceleration (test #5), there was no improvement due to LSD settings (like the RWD car). So the important factor must be accelerating WHILE TURNING...

At low LSD Accel, the car would rampantly spin the inside wheel up to 160 km/h (test #6, track is Autumn Ring for its many low-speed corners). Higher LSD Accel reduced this speed until wheelspin was only present when the engine had enough wheel torque to spin BOTH wheels (110 km/h), which occurred at a strength of 25.

Above this strength, I was expecting exit understeer to increase (since my theory was that a tight LSD would not let the wheels travel at the slightly different speeds necessary to corner, therefore reducing front end traction). But noooooo, exit understeer decreased with higher and higher LSD Accel settings. It felt like the diff was sending more torque to the outside wheel, which was rotating the car to help cornering. At the highest values, I was actually getting power oversteer... in a FWD car!!!

-----------------------------
FWD CARS- INITIAL SETTING
(Since the LSD Accel test gave the same result as LSD Initial for 0-1000m times for RWD cars, I couldn't be bothered repeating this test for FWD)

Like LSD Accel, LSD Initial reduced wheelspin out of tight corners (test #7) but the "outside tyre rotating the car out of corners" effect did not seem as strong. Also I felt that (even though the inside wheel wheelspin was being controlled) it seemed slightly slightly easier to spin BOTH WHEELS. Hmmmm...strange.

Just like the RWD car, I did not like the handling felt with higher LSD Initial. It's hard to describe, other than to say it felt stodgy and unpredictable.

-----------------------------
FWD CARS- DECEL SETTING
Well, around Autumn Ring in the Civic, this setting sucked! Higher values made the car feel less responsive. Overall grip didn't seem affected but the car wasn't as keen to respond to inputs. This was a totally different effect to how RWD reacts to LSD Decel, anyone got any ideas?!

Coming soon...LSD for 4WD...in the meantime, all comments welcomed.

 

[Parnelli Bone]

Very good. I have used mid to highish Accel ratings for FWD with results like you note...at a low-speed track like Autumn Ring. But at a track like Route 246 with its high-speed curves, understeer still shows up under throttle out of those faster areas. I remember using high Accel LSD anyways when I drove a Civic in the Turbo races, in an effort to kill as much wheelspin as I could when the car was in 2nd gear.

I rarely use Initial or Decel ratings higher than 5 or 10 in a FWD, too, as you noted, because a front-drive car's advantage is usually best while it's braking & turning-in. Why kill this with entry-understeer a highish INitial and/or Decel can cause?

 

[sukerkin]

The 'both wheels spinning' effect you noted is a very real problem with LDS's - 'real' as in it happens in the real world as well as the game world.

In a high power car with a stiff diff it is possible to destroy your acceleration as both tyres light up. Depending on how the diff operates (locks and unlocks the power transmission} and the events leading up to the 'light up' you can get some vicious snap-spins too.

 

[nomis3613]

Very good. I have used mid to highish Accel ratings for FWD with results like you note...at a low-speed track like Autumn Ring. But at a track like Route 246 with its high-speed curves, understeer still shows up under throttle out of those faster areas. I remember using high Accel LSD anyways when I drove a Civic in the Turbo races, in an effort to kill as much wheelspin as I could when the car was in 2nd gear.

Thanks, Parnelli. I took your advice and tested the Civic at Route 246 and, like you say, stronger LSD Initial gives more high-speed exit understeer. Personally, I'd be happy to just compensate for this using downforce, but thanks for pointing out the side effect.

The 'both wheels spinning' effect you noted is a very real problem with LDS's - 'real' as in it happens in the real world as well as the game world.

In a high power car with a stiff diff it is possible to destroy your acceleration as both tyres light up. Depending on how the diff operates (locks and unlocks the power transmission} and the events leading up to the 'light up' you can get some vicious snap-spins too.

Cool, thanks for confirming this.

 

[Parnelli Bone]

Thanks, Parnelli. I took your advice and tested the Civic at Route 246 and, like you say, stronger LSD Initial gives more high-speed exit understeer. Personally, I'd be happy to just compensate for this using downforce, but thanks for pointing out the side effect.

I did the Turbo races in the Civic without downforce! Which is pretty sick against the crowd of rear and all-wheel drives. Had to rely on LSD, suspension, and tranny settings to win here. I remember it not being very fun to be honest.

 

[nomis3613]

Now for the long-awaited section on AWD cars. So I tested with 3 different cars and got 3 different results...here goes...

Audio Quattro
To keep overall grip levels low, so that LSD effects would be more noticeable, SH tyres were used. For downforce, I used as little as possible (though some was needed to balance the car). Gears were set as short as possible, so that difference in power-down could be noticed. Testing was done around Autumn Ring. Each setting was initially set to minimum, comments below are the effects as the setting is increased. VCD was initially set to 10 (more on that later...)
REAR LSD ACCEL: improved turn-in (eg same effects as for rwd cars)
REAR LSD DECEL: slightly less turn-in, less exit understeer (eg same effect as for rwd cars)
REAR LSD INITIAL: Not tested. Since the previous tests gave the same result as an rwd car, I figured initial would be the same- and I hated the initial setting during rwd testing.
FRONT LSD ACCEL: brilliant! Even at VCD = 10, I could feel the outside front wheel pulling the car out of the corner, eliminating exit understeer. (same effect as for fwd cars)
FRONT LSD DECEL: car less twitchy, or more stodgy (same effect as for fwd cars)
FRONT INITIAL: Not tested.
VCD: Surprisingly, increasing the power to the front wheels did not increase the effects of the front LSD. Exit understeer was slightly increased. I'm not saying the overall feel was better or worse, just different. But it definitely felt like the front LSD ACCEL effect of eliminating exit understeer is not "diluted" when the VCD is sending all of the power to the rear.

Mitsubishi Lancer Evo III
Same setup as Audi (minimal aero, SH tyres, Autumn Ring)
REAR LSD ACCEL: improved turn-in, maybe slightly increased exist understeer
REAR LSD DECEL: no noticeable effect, except for slightly more tyre chirping under braking
FRONT LSD ACCEL: maybe a tiny bit less exit understeer, hard to tell (much less effect than Audi)
FRONT LSD DECEL: better braking, less twitchy
VCD: Again, did not seem to "dilute" the front LSD settings. But the Evo III had better handling at ~35%. At higher VCD, the Evo III had less exit oversteer and seemed more "planted" in general. Whereas the Audi seemed best at the minimum (10%), so that it could use the power at the rear to compensate for understeer.

SKYLINE GT-R NUR SPEC
SH tyres, minimal aero, this time the track was George Paris (the Skyline seems to big and heavy for the tight and bumpy Autumn ring), VCD initially set to minimum.
REAR ACCEL: same as Evo (improved turn-in, maybe slightly increased exist understeer)
REAR DECEL: a lot more tyre chirping under brakes
FRONT ACCEL: no effect
FRONT DECEL: no effect
VCD: Like the Audi, increasing the LSD only increased power understeer. But then I removed the custom VCD to see what the stock Nissan 4wd system would do. Wow!!! Amazing, the engine power was being used perfectly to help the car through corners. The effect is incredible, so using a custom VCD in this car is a huuuuuge mistake. Without the VCD, the front diff settings suddenly became effective (like the Audi). The only downside of the stock 4wd is that it can be a bit slow to react, so last-minute adjustments can't be made during a corner. Also through a switchback it can get the tail penduluming a bit. Still, overall the benefits of NOT using a VCD are incredible.

What I find strangest about these results is that the Audi was the only car (with the VCD fitted) where the front LSD greatly improved handling. Any thoughts (on this question, or anything about GT4 LSDs in general)??

 

[DK]

Still, overall the benefits of NOT using a VCD are incredible.

Well, if it ain't broke, don't fix it.

 

[nomis3613]

Well, if it ain't broke, don't fix it.

Haha! What about "better to try something and fail than to try nothing and succeed"?

Unlike the Skyline, I've found most 4wd cars don't do Torque Vectoring. So usually the VCD helps tune exit understeer/oversteer/wheelspin.

 

[nomis3613]

Almost 6 months later, and my quest for LSD Understanding has taken me back to the theory behind how the FC LSD settings work. I've tried to get help from these pages but I'm still not sure if I'm understanding it right. Also, I could be on a wild goose chase if GT4 behaves like a Torsen instead of a clutch pack LSD??

Can anyone help me with some questions about how LSDs work in real life please:
1) is this how a 75% locking factor LSD works for acceleration: inside wheel spins > speed difference is detected > clutches engage to send 75% of torque to outside wheel and 25% to inside wheel? So a LSD can actually send more than 50% of torque to the wheel with grip (which is better than a locked axle can do)?

2) if so, then I assume that high speed exit understeer for an FWD car is caused by: outside wheel rotating faster causes diff to react as if the outside wheel is spinning > clutches engage to send 75% of torque to the inside wheel > extra torque on the inside wheel rotates the car in the opposite direction to the corner?

3) I am told pre-load applies to braking force too (ie not just acceleration). Is this correct?

4) Here is my understanding of how LSDs act in braking:
- diff open: each wheel is independently slowed by its own disc brake, engine braking torque takes the path of least resistance (eg if a wheel is partially locked, the engine braking will further slow it), wheels can rotate at different speeds to each other.
- diff locked: total force from both discs brakes and the engine braking is split 50/50 between the wheels, wheels must rotate at same speed
Can someone please check if I am understanding this right?

7) Preload: When coasting, is a diff with 0 preload acting as an open diff? And a diff with preload will lock the axle until the torque difference across the axle (eg one wheel with more grip than the other) exceeds the preload value? Is the reason that a car with high preload is difficult to turn because the axle is often locked?

6) just out of curiosity, do any diffs in real life have the option of changing the deceleration locking % independent of anything else?

And some questions about the GT4 settings:
7) Preload is usually in Nm (or ft/lb), so what unit is the "Initial Torque" strength in? A percentage of max engine torque perhaps? Or just an arbitary number?

8) Do you think the accel and decel strength values are actually locking %, or are they just an arbitrary numbers in GT4?