Why you need a LSD if you like cornering!

[wparsons]

Originally Posted by Manji

What difference does the amount of power make, as to whether someone needs a diff that locks under decel or not?

It doesn't, it just further shows he has no clue what he's talking about.

[wparsons]

Originally Posted by Ubersuber

For good examples of what is important consider the 70's rwd stock Euro racers like the BMW 2002 and Alfa GTV which uniformly show the inside FRONT wheel lifting completely off the ground and the 80's VW GTI showing the inside REAR wheel lifting completely off the ground.

What relevance do cars from the 70's or 80's, including a FWD one, have to this discussion?

Originally Posted by Ubersuber

If you need to seriously increase roll stiffness at the traction end of the car then the Torsen isn't going to cut it. Trouble is for fwd (or awd for that matter) the Torsen front diff is your only real choice unless you can fit one of the new clever electronic diffs.

Until that drive wheel is lifting the Torsen should be good enough and actually works better than a clutch type lsd. Once you get that rear roll stiffness up to critical level only a clutch type lsd will work properly for you. If you FI this engine then I recommend you do not try to also increase rear roll stiffness but if you do then a mechanical clutch type lsd will be the only solution to your traction issues out of corners.

That's only an issue if you only substantially increase roll stiffness at one end of the car, and is worse if you're using sway bars for the increase in stiffness. If you ran springs to give a 200% increase in rear roll stiffness and ran no sway bar you would be less likely to lift the rear tire than running 100% increase from a bigger sway bar alone.

If you increase roll stiffness by roughly the same amount front and rear you also don't have this issue. To lift the inside rear tire you have to be transferring more weight to the outside front (and to lift inside front, it goes to the outside rear).

You'll see a TON of fwd cars lifting the inside rear because they're running stiff rear bars to help them rotate, but anyone doing that on a RWD car has no clue what they're doing.

Look at the CSG shop BRZ (before the clutch LSD). They run 10k/12k springs (which we can all agree is a substantial difference in roll stiffness), and I'm pretty sure you won't hear @CSG Mike say he was fully lifting the inside rear tire.

[Ubersuber]

You should try to avoid contradicting yourself.

As far as I know physics hasn't changed since the 70's. Anyone watching the current Australian V8 racing will see that suspension design hasn't changed that much and the extent to which it has can be "blamed" on the fantastic improvements in tires since the 70's, exploiting physics not changing it.

The only reason you could need a clutch type lsd on these cars is if you are transferring weight in roll.

Stiffer springs lift weight more quickly as the car rolls (that would be the "transfer" effect, a bit of a misnomer but common usage).

All that can extend the spring is the unsprung weight whereas the sprung weight compresses the spring, there's quite a difference.

Whether you effect weight transfer by increasing roll stiffness with roll bars or springs doesn't affect the fact of weight transfer. Stiffer spring rates however achieved will transfer weight in roll more quickly than softer spring rates.

It is this lifting force that unloads the inside drive wheel causing it to lose traction and for which any form of slip limiting device should result in improved acceleration out of a corner. The discussion here is the extent to which these cars benefit more from a clutch type than a Torsen type.

I am keen to know how a 2 way (or quaintly named "1.5 way") diff can possibly help these cars but I concede it might be possible.

So far, the actual numbers linked to in this thread suggest only very marginal improvements can be effected by fitting a 1 way clutch type lsd. More objective numbers would be a lot more useful than continuing this apparent disagreement about physics as applied to rolling cars.

[Ubersuber]

Originally Posted by Manji

What difference does the amount of power make, as to whether someone needs a diff that locks under decel or not?

I just assume a slow car like these cannot be made to go faster by locking the diff on overrun. It seems like a reasonable assumption. The real question is can a two way locker make any car quicker around a circuit. If yes then how since it can only slow the car down more rapidly, if it works at all.

[wparsons]

Originally Posted by Ubersuber

As far as I know physics hasn't changed since the 70's. Anyone watching the current Australian V8 racing will see that suspension design hasn't changed that much and the extent to which it has can be "blamed" on the fantastic improvements in tires since the 70's, exploiting physics not changing it.

Physics may not have changed, but suspension technology sure has. Better tires, better dampers, more consistent springs, lighter wheels/tires, etc. The overall idea might be the same, but if you look at the details they're MUCH different than 40 years ago.

Originally Posted by Ubersuber

The only reason you could need a clutch type lsd on these cars is if you are transferring weight in roll.

Did you even read the data and articles? There are reasons beyond the inside tire lifting to want a clutch LSD over a Torsen.

Originally Posted by Ubersuber

Stiffer springs lift weight more quickly as the car rolls (that would be the "transfer" effect, a bit of a misnomer but common usage).

If you increase roll stiffness you get quicker weight transfer, but you also get less body roll. Less body roll means the unloaded side isn't lifting as much.

Originally Posted by Ubersuber

All that can extend the spring is the unsprung weight whereas the sprung weight compresses the spring, there's quite a difference.

Not true. On the stock springs/shocks, the spring is partially compressed even with the shock/strut out of the car, so it's not just unsprung weight extending the shock, it's the spring rate pushing the shock into the droop range. Have you jacked up your car, did you notice there is a good 3-4" of droop travel from stock ride height? Not all coilovers work the same, but anything with a good setup (helper spring, etc) will be similar.

Originally Posted by Ubersuber

Whether you effect weight transfer by increasing roll stiffness with roll bars or springs doesn't affect the fact of weight transfer. Stiffer spring rates however achieved will transfer weight in roll more quickly than softer spring rates.

You're right on the weight transfer, but wrong on the overall effect. With a stiffer sway bar the inside tire won't droop as much, so it'll lift much sooner. If you compare two cars with identical roll stiffness, one from springs alone (no sway bar), one with stock springs and a sway bar to make up the difference the car with the bigger bar will lift the inside tire first. Every single time. It's not just weight transfer that affects if/when the tire lifts.

Originally Posted by Ubersuber

I am keen to know how a 2 way (or quaintly named "1.5 way") diff can possibly help these cars but I concede it might be possible.

There is a difference between a 2 way and a 1.5 way LSD, they aren't just different names for the same thing. A 2 way has the same amount of locking under both acceleration and deceleration, a 1.5 way is less aggressive under deceleration.

Originally Posted by Ubersuber

I just assume a slow car like these cannot be made to go faster by locking the diff on overrun. It seems like a reasonable assumption. The real question is can a two way locker make any car quicker around a circuit. If yes then how since it can only slow the car down more rapidly, if it works at all.

You really don't get this at all, do you?

1) It's not a locker, clutch LSD's don't go full lock under the slightest application of torque, it's progressive and only as much as needed.

2) How would a 2 way LSD (or even a full locker) slow the car down faster when coasting in a straight line?

3) Are these cars slow enough that they can run every track at full throttle the whole way without ever lifting or needing brakes? If not, your point about the locking on deceleration is totally moot. Having a 1.5 way over a 1 way helps with rotation when you're off the gas, and is more predictable to drive than a 1 way. Every single track I've driven requires the use of brakes, and not entering every corner at WOT... so yes, it would be an improvement, even with these "slow" cars.

You've been proven wrong by tons of people on everything you try to sound smart on, when will you just give up and stop spreading bad information?

[Ubersuber]

Some of us are just slow to cotton on I guess. Sorry about that.

I was prompted to go back to the initial post though, which is very interesting and the only reason I thought to post to this thread in the first place.

Mike is to be congratulated for going to the work of getting and graphing this data.

Re-reading all the useful posts I got to thinking what the wheel speed traces should look like with a clutch type lsd on the same car. Some threads suggest that the Torsen should work just fine and others are sceptical.

So, do the wheel speed traces actually tell us what is going on at the contact patch, i.e. effective driving torque or torque robbing wheelspin?

It is axiomatic that the inside drive wheel will travel more slowly than the outside drivewheel until it spins up after breaking traction. Actual wheelspeed will be higher than geometric wheelspeed just before traction is broken however, as a tire will develop maximum grip at around 10% or so overspeed. For maximum driving force from both drive wheels you want both drive wheels to enter this state. I'm leaving aside the concurrent need for lateral grip for the moment although both need to be vector resolved for a complete analysis of what is going on. When Mike says he is modulating the throttle this is exactly what his brain is doing: vector resolving the desired combination of forward drive and lateral grip, all without a calculator....

So, put in a clutch type lsd, and for the moment just a 1 way for simplification. As the inside drive wheel unloads and is unable to deliver all of the available torque to the road then what happens? That inside wheel needs to spin up as the lsd locks up (doesn't matter if it's progressive or not, the fact is the lsd limits "slip" between the axles and if enough input torque is delivered they lock up 100%). The clutch type lsd tries to bind the two axles together and with sufficient input torque and assuming the car does not slow down or change radius of turn the inside drive wheel is driven up in rpm to match the rpm of the outer drive wheel. These lsd do not transfer torque, they permit more torque to be utilized by the outer drive wheel and necessarily spin up the inside wheel, equal torque is delivered to both drive wheels once the lsd locks up as it does if wheelspin commences. Tire smoke and wasted engine torque are the necessary result.

Keep on the throttle and eventually the outer drive wheel also reaches its grip limits and both rear tires will smoke. Mike's onboard calculator then applies countersteer. (Evo TV does a nice comparison video between the new Cayman GTS which finally has that clutch type lsd everyone was crowing for and a 996 911 GT3, on his "warm up lap, but significantly not on his fast lap, Bovingdon demonstrates this action of a clutch type lsd rather neatly). Of note are the lack of pre-load and the very little ramp angles in this Cayman lsd:

http://www.planet-9.com/987-cayman-b...dissected.html


With the Torsen, once the inside wheel begins to spin then that inside wheel cannot "drive" the outside wheel beyond its grip limits. The Torsen biases torque but can only do so if one wheel is applying some torque and the other wheel can use more than that. Only if the outside wheel cannot use the amount of torque being applied to the inside wheel times the bias ratio will both rear wheels spin up with a Torsen. In the dry this would be unusual though quite common in the wet or on snow. I mean 151 lb ft divided by two or by 5 is still a small number even if in the latter case it is multiplied by up to 4 for the stickier tire.

However, in both cases the wheel speed sensors will indicate that the inside wheel is picking up speed.

Unless I am misreading the graphs, and before anybody jumps on me for that concession I freely admit that is entirely possible, I see the rear wheel speeds as close to the same regardless of left or right turn is involved. Even the dips and spikes in the traces seem to follow each other, except for box 3 which is interesting in itself. This seems to indicate that the Torsen is doing a pretty good job of maximizing the available traction of both drive wheels. The inside wheel trace is spikier than the outside wheel trace but that is what the Torsen demands for maximum efficiency in torque biasing. In a sense the total usable torque depends on the inside wheel reaching maximum efficiency and remaining there which then maximizes the torque biasing.

There must be something else going on if the lsd is indeed quicker than the Torsen.

And, just btw, the "binding" you get with a Torsen at low torque levels and high wheel speed differences is in the nature of the beast. The wheels are still widely differentiating but the Torsen "sees" the outside wheel as delivering much less torque than the slow moving inside wheel and tries to bias torque to the inside wheel. That scrabbling is the opposite of failure to differentiate and is indicative of the "one way" nature of the Torsen. Apply more torque in that tight corner and the scrabbling will stop. In winter or other low friction surfaces this characteristic of the Torsen can be quite disconcerting.

One official Toyota video I watched, linked on here somewhere to an australian weblink, suggests this Torsen is a 4:1 bias ratio (the commentary says "up to 4 times" the torque can be transferred) which is a lot. Most fwd Torsens are in the 2 - 2.5 bias ratio range.

[wparsons]

Originally Posted by Ubersuber

And, just btw, the "binding" you get with a Torsen at low torque levels and high wheel speed differences is in the nature of the beast. The wheels are still widely differentiating but the Torsen "sees" the outside wheel as delivering much less torque than the slow moving inside wheel and tries to bias torque to the inside wheel.

I'm WELL aware of why and how it is happening, I was simply pointing out to you that the stock Torsen does it when you claimed only a clutch LSD would do it.

Originally Posted by Ubersuber

That scrabbling is the opposite of failure to differentiate and is indicative of the "one way" nature of the Torsen. Apply more torque in that tight corner and the scrabbling will stop. In winter or other low friction surfaces this characteristic of the Torsen can be quite disconcerting.

This has NOTHING to do with it being a 1 way. A 1 way, 1.5 way and 2 way all react the exact same under power. The only difference is when decelerating with the clutch out and in gear.

I can assure you that if I give it more gas, it doesn't stop the inside tire from scratching for grip, it gets worse until both tires break loose and the rear is sliding.

[CSG Mike]

Suffice it to say that we've seen a substantial improvement in the car's ability to power out of a corner. With zero changes to the car's engine output, even the vmax in straights is higher, as well as exit speeds.

It's entirely possible to get inside wheel spin without lifting the wheel; you just need to overcome the tire's grip. When you unload the inside tire under cornering, that's what you're doing.

[mike156]

"1.5 way" vs. "1-way"
From my experience, if you get some degree of lock up on decel, it gives you more opportunity to control the car with the throttle. It's similar to using trail braking to induce oversteer. With under powered RWD cars, there are corners where you might go through it at heavy throttle and the weight transfer you get keeps the rear planted well enough. With the 1.5way, you can drop throttle to transfer weight forward and the reduced tire load allows it to induce oversteer. You then get back on the throttle to tighten up the line.

How useful that is on a high speed track, I couldn't say. It's very useful in wet conditions and is fun on the street though. Would you be faster with a 1-way? Couldn't tell you. It's definitely more fun on the street though with an underpowered RWD.


Torsen vs. Clutch really seems to come down to two different thoughts. Some will argue that with proper suspension setup, the car won't lift the inside drive wheel. While this is likely true for most situations, when the road gets bumpy or off camber, or you hop curbing all that goes out the window. At that point, driver confidence is all that matters and with grip and drive torque going all over the place like you get with torsens in that situation, it's bound to slow you down.

FWIW, BMW on the E36 uses a pretty light preload with fairly mild ramps. The OEM upgrades done on them tend to reduce preload even further but then make the ramps more aggressive. Just one more datapoint.

[CSG Mike]

Originally Posted by mike156

Torsen vs. Clutch really seems to come down to two different thoughts. Some will argue that with proper suspension setup, the car won't lift the inside drive wheel. While this is likely true for most situations, when the road gets bumpy or off camber, or you hop curbing all that goes out the window. At that point, driver confidence is all that matters and with grip and drive torque going all over the place like you get with torsens in that situation, it's bound to slow you down.

Lets say that the suspension is set up to maximize the rear grip, with a balanced car. Now add a clutch type LSD. Now the rear grip is improved, and the car is able to power out better, but when that extra power is applied, now the front pushes. You compensate by softening the front rates slightly, restoring the balance. Now you can fully utilize the increased rear grip, and increase net corner exit speed.

Absolutely correct, the suspension's job is to keep the tires in contact with the ground, but under cornering, if the inside load is reduced enough, you can still spin the inside tire, temporarily reducing the stock toresen to an open diff.

[ZDan]

Originally Posted by CSG Mike

Lets say that the suspension is set up to maximize the rear grip, with a balanced car. Now add a clutch type LSD. Now the rear grip is improved, and the car is able to power out better, but when that extra power is applied, now the front pushes. You compensate by softening the front rates slightly, restoring the balance. Now you can fully utilize the increased rear grip, and increase net corner exit speed.

This can be a path to ever-increasing diff lockup coupled to ever-increasing rear roll stiffness. Add stiffness in the diff, get push, add rear roll stiffness, increases unloading of inside rear, requires more diff lockup, which requires more rear roll stiffness, ad infinitum.

Can lead to wildly different setups on the same or very similar cars, handling totally differently, yet yielding similar lap times. A lot of diff lockup with higher rear roll stiffness vs. minimal diff lockup with less rear roll stiffness. Either approach can give you the drive needed out of corners.

IMO, the best diff is the one that interferes the LEAST with the outputs while ensuring that power gets to both rear wheels and never wastes power on inside wheelspin.

Have to keep the higher-level picture in mind. More diff lockup isn't always "better". In the end, tires used, driver preference, etc. will influence the overall solution.

There's no one answer! Driving styles and preferences are different, and this is PRECISELY one of those areas where the setup that works best for one driver might not work best for another.

[Captain Snooze]

Originally Posted by ZDan

There's no one answer! Driving styles and preferences are different, and this is PRECISELY one of those areas where the setup that works best for one driver might not work best for another.

I'm thinking that an aftermarket diff is a plug and play item. I don't have the money or experience to be experimenting with diff settings. (I'm guessing a few people would fall into this group.) So long as there is some benefit straight out of the box I would be happy.

[s2d4]

Originally Posted by Ubersuber

Some of us are just slow to cotton on I guess. Sorry about that.

You must be suberman.

[CSG Mike]

Originally Posted by Captain Snooze

I'm thinking that an aftermarket diff is a plug and play item. I don't have the money or experience to be experimenting with diff settings. (I'm guessing a few people would fall into this group.) So long as there is some benefit straight out of the box I would be happy.

That's why we do the testing