Limited Slip Differential (LSD) theory

[coryclifford]

Instead of understanding I think it would simply behoove a person to simply experiment and find what they like. I have cars with a high number and cars with a low number.

 

[CSLACR]

Some tests indicate initial effects constant locking rate, at least somewhat.
Some tests indicate initial effects how quickly accel or decel activate, at least somewhat.

Probably because it does both?

 

[Motor City Hami]

Take ABS off and several "good" tunes are turned to awful and replica wipes podium.
As long as you're using ABS you can tune really freely and make super fast hotlapper TT cars, when ABS is off order of tunes changes.

I have yet to see someone who makes statements like this actually prove this theory. It should be easy for you to build an ABS zero tune then convert it to an ABS 1 tune and explain the difference in settings and why you made them. People could test the two tunes and really see the difference that you are attempting to explain.

 

[OdeFinn]

I don't get this, if that just triggers how i.e. 35/-/- lock can be triggered, it's empty chamber.
Or what's reason for over/understeer?

@OdeFinn I have tested and tested and tested and tested the LSD for hours. I hate to point to the scoreboard, but if you want to begin to understand how I have earned so many podium finishes in tuning competitions, a look at my LSD settings would be a great place to start.

I have also test driven a number of times in tuning competitions. There are cars that I have tested, adjusted their LSD settings just for fun and saw their cars go faster than mine. This is how I moved on from LSD testing to better understand dampers and springs. There are quite a few of the current FITT tuners who were way ahead of me in spring and damper tuning. But, it's pretty clear that I caught up.

I'm not arguing your reputation on fitt compos, you are just not answering now on question.
If higher number means understeer and lower number understeer, what's reason why it makes such behavior?

Take ABS off and several "good" tunes are turned to awful and replica wipes podium.
As long as you're using ABS you can tune really freely and make super fast hotlapper TT cars, when ABS is off order of tunes changes.

I have yet to see someone who makes statements like this actually prove this theory. It should be easy for you to build an ABS zero tune then convert it to an ABS 1 tune and explain the difference in settings and why you made them. People could test the two tunes and really see the difference that you are attempting to explain.

We're not exploring GT6 ABS. You're just jumping ahead, dodging question what I made earlier.

Some tests indicate initial effects constant locking rate, at least somewhat.
Some tests indicate initial effects how quickly accel or decel activate, at least somewhat.

Probably because it does both?

You're right also here, as discussed on this thread there is assumption of locking speed difference depending ratio of initial and actual lockable value.

 

[Otaliema]

@OdeFinn most of it is how your saying it. It is implying that ABS causes changes in the behavior of the LSD. If it simply for understanding ok I can accept that. But if it's how it reads, ABS 1 messes with the LSD ABS 0 doesn't than its no good.
Yes everyone uses a LSD just like they use tired an engine and suspension and a transmission. It's part of the cat. PD has yet to allow us to remove something other than weight.

 

[OdeFinn]

@OdeFinn most of it is how your saying it. It is implying that ABS causes changes in the behavior of the LSD. If it simply for understanding ok I can accept that. But if it's how it reads, ABS 1 messes with the LSD ABS 0 doesn't than its no good.
Yes everyone uses a LSD just like they use tired an engine and suspension and a transmission. It's part of the cat. PD has yet to allow us to remove something other than weight.

ABS is not messing LSD, but ABS messes suspension movements, LSD and suspension are both used to get proper traction, so it also affects slightly to LSD.
This this cross behaviour is not breaking LSD, but on some cases it "allows" use of i.e.over heavy declaration lock, what will make car braking really pita if noABS used.
In other way to say: "wrong" LSD values what would brake traction more than gain traction are working because ABS won't let wheels to lockup as LSD tries to do.

 

[Lionheart2113]

I'm not arguing your reputation on fitt compos, you are just not answering now on question.
If higher number means understeer and lower number understeer, what's reason why it makes such behavior?
According to tests, performed long time ago by me, recently i.e. @eran0004 there is reason for that, but it's not because of triggering locks accel or decel side.
you think at stock lock is biasing results you can install custom LSD on FWD car and try with settings i.e. 30/5/5 and 50/30/5 (or 30/30/5 and 5/30/5 for direct comparison). Remember to leave ABS off, it's not helping on this test.

This would be easy solution, but it doesn't explain why using real-world setups are working just fine on GT6, I have tested several real-world setups on "hardest" cars as well on "normal" cars and those are settling cars behaviour just like real counterpart car is acting on real-world.


For fitt compo tunes I have to say one thing, several of those aren't working at all if ABS is switched off, ABS on GT6 is just biggest anti-realism thing what confuses a lot of actual game physics, it masks so much of good physics engine. ABS in GT6 is far far far far far away from just being normal brake assist, it kills big part of suspension physics as well other parts of physics.

Take ABS off and several "good" tunes are turned to awful and replica wipes podium.
As long as you're using ABS you can tune really freely and make super fast hotlapper TT cars, when ABS is off order of tunes changes.
I'm looking from this game kinda simulator way of approach to my gaming experience, I'm not crying on online races if not winning, but jumping on joy when all drivers are respecting each other and cars and racing clean tight race.

If this isn't a discussion leading towards the ABS0 vs ABS1 debate, then why so many references to it? I'll be honest and say that it is, and has been, reading like it's going in that direction.

 

[OdeFinn]

If this isn't a discussion leading towards the ABS0 vs ABS1 debate, then why so many references to it? I'll be honest and say that it is, and has been, reading like it's going in that direction.

Because I suggested to do testing without ABS to get clearer results, then people drag to that ABS more than needed.

Btw few examples of locks what are locking over 100% if they were transferred to GT6. Getting idea what @CSLACR said about locking speed. Just draw that linear locking line up til it reaches 100% of throttle and you're way over 100% of locking.. But you can't lock over 100% even on real-life.

Meaning this:

If remaking those on GT6:
Yellow: 0/140/xx
Green: 15/125/xx
Red: 30/105/xx
Blue: 40/80/xx

Now your next thought is clear, no such numbers available, or anything similar. Yes, you're right, that cam angle used on above picture isn't cam angle what's simulated in game. Cam angle what game simulates is smaller, making locking "slower", linear locking line rises slower on similar graph as above, maximum elevation of 60% from throttle off to throttle max. Initial value defines starting lock rate, and i.e.acceleration final lock rate, difference between those are defining angle of locking rate increment.
This is just theory..

And to confuse reader more, above cannot directly use as is on GT6, manufacturer announced locking rate is not same as actual wheel locking rate, manufacturer speaks about percentages of LSD functioning lock amount, how much actual wheel locking rate is depends friction caused by clutch plates. So if above lock is capable to produce only 60% wheel lock rate when it's functioning full 100% capacity then above diagram 100% locking is equal to 60% locking on wheels, and that locking on wheels can be transported 1:1 to GT6.

 

[Motor City Hami]

We're not exploring GT6 ABS. You're just jumping ahead, dodging question what I made earlier.

Not sure what question that I dodged? I was pretty clear and firm on my ideas around LSD tuning. I simply do not see the point in attempting to match real world theory to PD programming because the link has been impossible across most settings. Wait until you try Project Cars. The physics model is so much better. Real world knowledge is an advantage.

I think that you dodged my question about how ABS really effects tuning. You used double speak even in your latest post saying that ABS doesn't effect the LSD but then said that it does have an affect on it, but through the suspension.

 

[saywhat]

I simply do not see the point in attempting to match real world theory to PD programming because the link has been impossible across most settings.

Amen.

I'd love to see someone put a wheelie tuned car into "theory"
Someone please tell me how PD defies gravity.

 

[OdeFinn]

ABS is not messing LSD, but ABS messes suspension movements, LSD and suspension are both used to get proper traction, so it also affects slightly to LSD.
This this cross behaviour is not breaking LSD, but on some cases it "allows" use of i.e.over heavy declaration lock, what will make car braking really pita if noABS used.
In other way to say: "wrong" LSD values what would brake traction more than gain traction are working because ABS won't let wheels to lockup as LSD tries to do.

Pretty much answers your question, was posted one message later.

Not sure what question that I dodged? I was pretty clear and firm on my ideas around LSD tuning. I simply do not see the point in attempting to match real world theory to PD programming because the link has been impossible across most settings. Wait until you try Project Cars. The physics model is so much better. Real world knowledge is an advantage.

I think that you dodged my question about how ABS really effects tuning. You used double speak even in your latest post saying that ABS doesn't effect the LSD but then said that it does have an affect on it, but through the suspension.

Whole point of this thread is to get understanding of why PD modeled LSD work how it works.
Project cars face same problem as GT6 what comes to modeling realistic LSD, only open, full locked (solid/welded) and Torsen style differentials can be simulated "100% realistic". Other differentials are facing problem where you have to decide some values to be static/linear, i.e. Slip factor of clutches, ramp angles, pin shapes, temperature changes to friction etc.. this leaves programmers only choice to go with static locking curve instead of real-life active locking curve, programmers can tweak manually differences on locking curve behavior, but it's always "static" to compared how it works on real-life.

Even I'm "debating" this LSD theory as is, I'm simultaneously checking possibilities what could be on programmers way to simulate realistic way and figuring out is this done by some other way.
Currently investigating alternate theory of numbers to represent kgf-m or lb x10 values of lock resistance, trying to get time to tests if breakaway point could be found on those resistance values matching engine+gearing kgf-m torque.

Programmers can go few way, torque sensing geared (Torsen, Quaife, etc. not used as we know), speed sensing (viscous, speed difference between wheels, can be counted, not used as well, or weird simulation of it) or torque sensing with clutch pack ramps preload etc, practically most user friendly to setup is clutch pack LSD and therefore mostly used on simulation coz variety of values what to alter by user.
Assuming at GT6 simulates clutch pack LSD makes values "understandable", but behind values PD have to lock few things to static, effective ramp angle, determines speed of locking as well amount of locking on cases were there is insufficient amount of torque for pin to climb end of ramp for spreading pressure ring enough to lock 100% on lock mechanism. Normal cases pin travels to "end" of ramp, or it travels at much what will allow lock to be fully engaged. When lock is fully engaged are wheels locked only by amount of friction caused by clutches, than can vary.

If clutches under maximum pressure caused by pressure rings and preload spring are making 40% effective locking rate to wheels you have plenty of things to do for simulating that.
You had to make static locking curve from initial to full lock, pretty easy, but what's breakaway?
That 40% can be achieved and keep inside some particular torque range in real life, defined by clutch capability to handle heat, pressure amount what lock is giving to them etc. Same lock with different clutch amount/type can provide i.e. 80% effective locking on wheels..
Are we adjusting effective locking rate with static breakaway, or is breakaway changing along effective locking rate? I'm guessing at we're adjusting effective locking rate, but there is still few questions in the air, clutch of car on PD model slips constantly, is reason to slipping to give feel of clutch pack of LSD slip? Better clutch on car activates LSD better in PD model, partly true on real-life but not as PD uses it, but it gives higher breakaway feel on LSD in PD model, well is that for us to give slight smoothness on LSD? LSD reacts quicker on carbon propeller shaft, is that for us to get bigger cam angle(or U.S. way smaller ramp angle)?

As you see there is lot of things "hidden" on PD's programming, and similar things are hidden also in project cars, or if not they aren't even trying to simulate it "right".
Apologies for message overall look/order, writing in several parts as having time..

For programmer easiest lock to do is speed sensing lock, simulating viscous lock with unrealistic locking curve.

 

[Motor City Hami]

 

[OdeFinn]

Hoping at this isn't flying above..

After few tests here is more for LSD breakdown on GT6.
First there is "hidden" static value for fully engaged lock, it's always on 75% of throttle, that's pretty much hardcore fact.
Then based on feeling, not yet tested on "picture it/video of it" way is that lock engaging starts on 50% of throttle, it's not linear, more like Bezier curve rising really progressively. Until that 50% there is only initial locking amount active.
I.e. 10/40/- lock looks like this:

Then this 75% of throttle has other hidden static feature, i.e. 35/-/- (prelude) is giving up initial on that point, breakaway. It keeps lock fine till you hit throttle over 75% and it starts gradually slide towards open differential.

And for @DolHaus or @Otaliema mentioned switching between initial and acceleration is having one bug, visible on low power engines, it's not affecting actual locking, but grip can probably suffer.
It's tire "cooling" bug, when throttle goes slowly from over 75% position lock engages fully and that makes really weird sudden cooling from red to blue and then rising back to red on better gripping tire.
There is no good logic behind, it will happen i.e. 5/40/5 and 40/5/5 locks, so no related to bigger initial etc.
Video of mentioned behaviour:

Small light on end of tunnel, 75% throttle is static, but you can alter your RPM's on 75% mark, changing clutch to better your lock activation place will change, you will have different RPM running on engine when lock engages. Clutch also makes initial and lock grip stiffer.. So no free dinner..

Edit/addon:

Carbon propeller shaft changes lock activation really linear, and makes it start to engage locking earlier, testing more but seems like carbon makes locking start linear from beginning of throttle till 75% reached where it's engaged fully. It also keep linearity during deceleration. Linear or smooth curve.
Locking curve with carbon propeller looks like much of those Cusco diagrams posted earlier.
I'll draw a picture for easier understanding..

Locking with carbon propeller shaft, blue is my guess, based on PD's love of Bezier curves, but difference is not huge is it that linear or curve. Point of starting to rise locking percentage from beginning is clear.

Breakaway point is affecting all sizes of initial preload, same 75% of throttle is there and LSD friction breakaway and transition towards open lock starts, don't stick this too strictly, there is things what's needed to understand along that, will draw few pictures how it probably goes based on feel/seen (indicators etc) behaviour on game.
But basically i.e. 60/5/5 lock breakaway on 75% of throttle, then how acceleration side changes that, I have few pretty accurate assumptions, tests are standing behind them, will update more later.

 

[OdeFinn]

Uff, creating a test pattern for this is kinda multiplexed thing to do.
My feet 'o meter report transition on throttle response between clutches, tested and yes there it was..
Approximations presented from actual physical G27 movements vs. GT6 throttle range 50-100%, to simplify tests there is carbon propeller installed, propeller might have response effect too, probably 0-50 range changes most because of it.(below numbers 100% represent total range from 50% to 100%)
Triple: 50-75=50% 75-100=50%
Twin: 50-75=45% 75-100=55%
Std: 50-75=33% 75-100=67%
Clarify, i.e. Std clutch you have longer range on actual movement to get game throttle to rise from 75%-100% than 50%-75%, approx twice longer move have to perform for that.

Considering this physical pedal reacting curve change along with game effect change makes this even more fun.. Fixing up testing procedure.. Thanks in advance for i.e. @Lionheart2113 to provide MoTeC data, from his older tests and ability to help on forthcoming tests.

 

[Ridox2JZGTE]

@OdeFinn : Saw this on one of my old docs, from Xtrac fwd plate LSD manual which shows how preload/initial torque works, the Xtrac offers different preload spring up to over 200Nm.

P45 power ramp angle at 45 deg

C60 is coast ramp angle at 60 deg

Imagine if the LSD is set with 125Nm preload, but with 4 weak friction plate and 85 ramp angle on both power and coast ( lowest lock ), it would be similar to say 30/5/5 LSD in GT6 ( just a rough example )

 

[demonchilde]

MCH, that image was worth 1 million words, lol.
I have been following this 'friendly discussion' for a bit now, and due to my personal rule of "10 foot pole distance on anything that smells like gas when theres a lit match present", I have kept quiet.
But that image was priceless and perfectly timed. I spit out my coffee when i saw it. There has to be a warning before such things, lol.

////darts back to the shadows/////

 

[SiTHNiSMO]

Very interesting Mr Ode, I think what your saying makes sense. Going in DEEP. I do my LSD tuning looking at how my drive wheels are reacting to everything in my set up and tune it based on that. I'm not drifting and not after that full time locked diff set up, I think the Drifters and NASCAR guys both run full time locked set ups.

My differential tuning looks at Grip, weight, downforce, and set up style all as contributing factors. I consider the overall grip to power ratio when tuning the LSD. For example Ill use a FR Sports Car. A light car with lots of power and low grip Street Tires & little to no downforce takes a particular style of set up from me. More springs than bar to keep my inside tires planted in light / low grip situation. More springs & less bar call's for a little more Camber & a light LSD to use the high power effectively. The lower the power is OR more grip from stickier tires will increase the grip to power ratio & this will let me tighten up the diff to put more power down.

A good example car would be my Bowtie Sports Car the M4, Performance in a Suit. My 550pp SS set up is relatively light at 1250kg, doesn't have significant downforce, and running 505hp on Sport Soft tires ABS-1 its power to grip ratio is relatively high on power.

The M4 is easy to light up the rear SS tires and drift out of corners, she's got more power than tires. My set up for the M4 runs stiffer springs than bars, a bit of Camber & a light LSD set up. The rear bar softer to have that bit of extra exit grip on the back end for throttling out of corners the springs stiff to keep it going down the straight, the LSD is light to keep from blowing off the tires, only as light as she needs to be. Too tight lights up the outside, too soft lights up the inside. . What's too tight a diff on Sports tires may be Just Right on Racing tires. So much depends on pretty much everything else.

The Power to Grip Ratio is very significant to LSD set up. One car can have anywhere from a light to tight LSD depending on the power to Grip Ratio of the set up. How any one setting is tuned is afected by the entire set up of the car.

There is no one answer or rule of thumb to generally apply to all cars, even if broken into platforms. A car doesn't react the same if weight, balance, power, grip level, downforce, set up, or track conditions change. What the tuner needs to adjust to achieve a desired result depends on all settings already applied to the car, its specs, tires and so on. Adjustments to settings just don't do the same thing to handling no matter what, so much depends on so many other factors. Often times a combinations of settings can render another adjustment more Bad than good when other combinations that adjustment will be more good than bad, all things need to be tuned to work together.

My M4 has what I consider balanced handling. As she sits adjustments are very predictable for me and the settings are working well together. I like to think of sub 7min at the Nurburgring as a Good benchmark for 550pp on SS. Im not crazy fast as some Hot Shot Hot Lappers are, but its a general Good in my opinion cracking under 7min. I figure if somebody can Best my Seasonal Time driving a Gift Car by a few seconds they should be able to cut my time with my own M4 by similar times. I don't have a tuner garage here but I'll donate my M4 Tune to the Debate if anybody is interested in seeing clear cut examples of theory applied in game practically.

 

[MrGrado]

Welcome to the forums. Please share your tune. Great first few posts also.

 

[SiTHNiSMO]

Welcome to the forums. Please share your tune. Great first few posts also.

Thanks, & will do, I'll be home in a couple hours.

 

[SiTHNiSMO]

BMW M4 Coupe 505hp 1250kg SS ABS-1

Power Modifications
Intake Tuning, Racing Exhaust, Power Limiter @ 97.7%

Body Modifications
Stage 3 Weight Reduction, Carbon Hood, Window Weight Reduction, 23kg Ballast @ 46%

Suspension Tuning
Ride Height 91/114, Spring Rate 7.40/12.46, Damp Comp 4/5, Damp Ext 6/9, Roll Bars 3/2, Camber 2.2/2.9, Toe -0.19/0.36

Racing Brakes
4/6

Drivetrain
Tripple Clutch, Carbon Shaft, Adjustable LSD
6/18/9

Transmission is always a Pita with the "Max Speed" adjuster manipulating gears adjustment range, a few steps are needed to get desired results, please bear with me.

Using the Final gear to get the ratios I wanted (stock gear ratios or as close as possible) I first adjusted the FD to 3.350 then adjusted the Max Speed to 273mph this gives me the ability to use stock gearing for the most part but effectively lets me adjust the final drive gear with stock gearing. I don't like how PD does the Max Speed on cars with multiple overdrive gears like the M4.

1st 4.678 (closest to stock first gear) 2nd 2.593, 3rd 1.701, 4th 1.277, 5th 1.000, 6th 0.853, 7th 0.671

After all gears are set I re-adjust the FD to 2.999

I have plans to work over the gearing a little bit but it works well for now as she is.


I'm not a Hot Shot driver but I can get her around Nurburgring in under 7min consistently, that's a Good Set up IMHO, its not a lap on the verge of losing it at any second but a high paced run around the Track that's a fun drive, I leave thinking about how I can push it harder through sections, I don't feel like I'm at the limits at all.. I'm sure a TT top leaderboard driver could do much better with a 0/0 set up but I also believe that same driver could do just as well with my set up if they put in the time they do with 0/0 set ups. I'm guilty of the same thing, I believe we all are somewhat going to push set ups we think are faster harder than ones we think are slower. A set up one decides in advance is faster (even subconsciously) will be pushed harder by them to realize it while we shrub off set ups we don't think are good, accepting slower times as if that's all they can do, but is it really?

I'm a replay whore too, I enjoy watching replays when I carve the track well and I like my cars to look as REAL as possible going around the track, I like a good stance that drives well and tune my cars to look as good as they feel. I work to make the car drive well and react predictably to inputs while looking right.

 

[Ridox2JZGTE]

BMW M4 Coupe 505hp 1250kg SS ABS-1

Power Modifications
Intake Tuning, Racing Exhaust, Power Limiter @ 97.7%

Body Modifications
Stage 3 Weight Reduction, Carbon Hood, Window Weight Reduction, 23kg Ballast @ 46%

Suspension Tuning
Ride Height 91/114, Spring Rate 7.40/12.46, Damp Comp 4/5, Damp Ext 6/9, Roll Bars 3/2, Camber 2.2/2.9, Toe -0.19/0.36

Racing Brakes
4/6

Drivetrain
Tripple Clutch, Carbon Shaft, Adjustable LSD
6/18/9

Transmission is always a Pita with the "Max Speed" adjuster manipulating gears adjustment range, a few steps are needed to get desired results, please bear with me.

Using the Final gear to get the ratios I wanted (stock gear ratios or as close as possible) I first adjusted the FD to 3.350 then adjusted the Max Speed to 273mph this gives me the ability to use stock gearing for the most part but effectively lets me adjust the final drive gear with stock gearing. I don't like how PD does the Max Speed on cars with multiple overdrive gears like the M4.

1st 4.678 (closest to stock first gear) 2nd 2.593, 3rd 1.701, 4th 1.277, 5th 1.000, 6th 0.853, 7th 0.671

After all gears are set I re-adjust the FD to 2.999

I have plans to work over the gearing a little bit but it works well for now as she is.


I'm not a Hot Shot driver but I can get her around Nurburgring in under 7min consistently, that's a Good Set up IMHO, its not a lap on the verge of losing it at any second but a high paced run around the Track that's a fun drive, I leave thinking about how I can push it harder through sections, I don't feel like I'm at the limits at all.. I'm sure a TT top leaderboard driver could do much better with a 0/0 set up but I also believe that same driver could do just as well with my set up if they put in the time they do with 0/0 set ups. I'm guilty of the same thing, I believe we all are somewhat going to push set ups we think are faster harder than ones we think are slower. A set up one decides in advance is faster (even subconsciously) will be pushed harder by them to realize it while we shrub off set ups we don't think are good, accepting slower times as if that's all they can do, but is it really?

I'm a replay whore too, I enjoy watching replays when I carve the track well and I like my cars to look as REAL as possible going around the track, I like a good stance that drives well and tune my cars to look as good as they feel. I work to make the car drive well and react predictably to inputs while looking right.

Since you've been banned, just try this little changes that may improve your M4 to handle better :

damper comp 4 6
damper ext 5 8
ARB 3 3


Just damper and ARB tweaks that should do well even on CS tire. The ride height glitch setting dulls the car front end response and rotation, so lessening the height difference a bit may also improve it, but not much.