I'm not assuming anything. My 240Z with ~200rwtq and ~170rwhp (then) is a very different story from a similar-weight Miata with what, 110rwhp?

I don't have excess torque available from the engine when my foot's on the floor. Which it is as soon as possible after apex.

That's a totally craptastic plan if you only HAVE 150lb-ft! (I'm of course assuming you're talking about engine torque and not the *actual* torque at the diff input and output).

No I don't. You need to *think* about it a bit more! If your foot's on the floor exiting a corner where you need some limited slip action to keep the inside wheel from spinning, then you are LOSING drive thrust with the virtual brake-system lsd vs. a clutch-type. This condition is true for at least one corner at most tracks I go to.

Yup, I blew it there. But if you have a stability conrol system (effectively mandated on 2012+ cars), you have what you need for at least a basic system.

It is worth noting that instead of a rear sway bar, the McLaren has a Z-bar. Which to acts like a rear ANTI-anti-roll bar. I.e., the lion's share of lateral load transfer under cornering is happening up front. Which very high power/weight cars might need more than my puny Datsun or even punier S2000...

 

Of course the same rules of vehicle dynamics apply. Only in one case some kind of limited slip is clearly needed much more than in the other case.

You can appreciate that the "continuum" doesn't imply that what works for one configuration will necessarily work in another quite different configuration.

Do you understand that a lower c.g. height vs. track width, plus greater weight %age on the drive wheels implies less of a need for lsd?

What I said was this:
With a c.g. height on the order of 1/2 that of my similar power/weight 240Z with similar track width and wheelbase, and with a significantly greater %age of its static weight on the drive wheels, a FF *is* a totally different story as far as the need for a limited slip is concerned. Even allowing for its greater cornering g's (probably ~1.2x that of the Z on Hoosier A6s).

What point do you take issue with?

 

Again, it is a continuum because even for a given car it's not clear if a LSD is good or bad. For that mater how much limiting do we want? Lets go back to the Miata example. With sticky tires on a dry autocross track I certainly want the LSD. However, if I have hard tires in the rain I want a less agressive LSD. The Infinity Pro Series cars are allowed to use LSDs. Depending on the track they will adjust the ramp angles for the diff. In a sense, for the exact same car they have more or less limited slipping (for lack of a better word). Sometimes they run a one way diff that's limiting in engine braking but open in acceleration.

This of course is one of the thing that makes the choice of LSD hard for a manufacture. A Torsen diff can have a torque bias ratio upwards of I think 4:1 but in most cars the ratio is more like 2.2:1.

I'm not taking issue with the claim that changes in the design of a car make the LSD more or less beneficial. What I do want to make clear is there isn't a line when we say needed or not. That line is affected by fundamental design parameters (weight, wheelbase, power), as well as event specific parameters (track surface, weather, tires, track layout). Even once we decide some type of LSD will make us faster we still have to understand that one LSD isn't the same as the next and sometimes we want a mild diff (less locking torque) while other times we'll want an aggressive diff. The LSD in my car is great in the summer on dry pavement. I'm not fond of it on gravel nor on wet oily roads. It's all a trade off.

Perhaps the most important thing to get out of this thread is the simple understanding that a LSD is not a cure all. It comes with trade offs and compromises. I'm certainly not claiming they are always unnecessary or will never help. I just want to make sure it's clear that they don't always help and there are reasons why we don't always want the most aggressive LSD we could find for our car. Again, it about balance and it's on a continuum.

 

 

It seems that BMW wasn't to fond of the traditional torque sensing LSD either.
http://www.autos.ca/auto-tech/auto-t...ferential-lock

I didn't realize they had this but the BMW LSD is a delta velocity based design. At relatively low differences between the left and right wheel the diff behaves like a normal open diff. Presumably that any time you wouldn't have slip with an open diff, this diff basically behaves like an open diff. However, once a wheel gets free the pump inside the diff creates clamping force on a clutch pack. The advantage to this system is it would give you the transfer of power to the wheel with grip when one tire is actively slipping but could be tuned to avoid the understeer issues that can occur with a torque sensing diff due to the transfer of torque to the inside tire when cornering below the slip limit.

I don't know if these diffs are durable enough for track work. I suspect BMW picked them because they offer the poor weather traction advantages of a LSD (even with one wheel on ice) yet they don't negatively impact the car's handling feel when you are at something less than the handling limits.

 

You keep repeating this garbled understanding of how a torque-sensing, specifically a clutch-type LSD works.

You are confusing speed with torque.

When the inside wheel presents less resistance to torque input, the clutch pack on that side loosens. The pack on the other side binds up because it is now pushing more of the car's mass. As a result, more torque is transmitted to the outer axle. It doesn't need the inside wheel to start smoking, it just needs a difference in resistance to the apllied torque. That is why it is called a "torque-sensing" device.

And no, I don't need to post up math homework on a public forum.
#1, I have a car that needs and has a torque-sensing clutch-type LSD and I can observe how it behaves in corners with the thrtottle applied. Do you? Because if you don't, you should stop telling me your paper math trumps my observations of the physical world.

And I know my car benefits from the LSD because I've driven similar vehicles without one, and the difference is easy to notice.

#2, Your math only accounts for a limited set of variables. It is a model, not the thing that defines reality. Models are developed based on observation of the physical world, not the other way around. It is YOUR problem if you cannot understand how something works. Take our observations and work up a mathematical model if you wish, but do not demand other people make one for you as proof they are not lying.

It is time for you to concede that your understanding of how a clutch-type LSD works is based entirely on theory and defer to those who have knowledge based on experience and direct observation in the physical worrld. You cannot discount the observations of multiple people merely because *you* cannot figure out the math. And it is intellectually dishonest to claim someone else is wrong because they cannot show you "the math" to your content. The math comes after the observation, one does not need to know how to demonstrate something as a mathematical model in order to observe it.

I can tell you the moon orbits the earth based purely on observation. I can prove the point without working out a mathematical model of orbital mechanics. In fact, you cannot work out a model of orbital mechanics without the observation. So claiming something doesn't or can't happen because there is no model is faulty logic.

 

Two things - that article incorrectly described the behavior of a torque-sensing LSD. And it went on to describe the BMW design as a speed-sensing differential.

OK, 3 things That article is reporting what BMW touts as an improvement over other designs. Of course it will talk up the benefits, ignore the negatives, and was written with zero motivation to accurately describe the other options. In other words, it's clear and obvious PR, not a peer-reviewed mechanical engineering white paper.

 

Nope, I have that part right. With any LSD (even the BMW one) torque is transferred from the wheel which is spinning faster to the wheel which is spinning slower. The only question is how much and what determines how much. With a fixed torque spring clutch the transfer is always the same. With a Torsen it's a function of input torque to the diff. With a speed sensing diff it's a function of relative velocities. No mater HOW the locking torque is determined, with any conventional LSD the torque transfer is from the faster wheel TO the slower wheel. That means when you are turning and neither wheel is slipping the slower inside tire gets more torque. My math shows this. If you disagree you need to show why my math is wrong.

You read where I mention speed and you think I'm not talking about a torque sensing diff. That is not the case. The magnitude of the locking torque is torque sensing. So the magnitude will be bigger if the input torque is higher. The direction of the torque is related to relative velocities. The clutches for the slower wheel will apply torque in the same direction as the diff housing. They are trying to pull the slower wheel up to speed. The clutches on the faster wheel are acting against the rotation of that wheel. They are trying to slow the wheel.

If you don't see this, if you still think I am confusing velocities with torques then you fundamentally don't understand my arguments and thus are not in a position to say I am wrong. We can try to come to an understanding on this point and I can try to help you understand what I'm trying to say. Don't claim I'm wrong if you don't understand my argument. If you don't understand it you can't say why it is wrong.

I showed why this is wrong mathematically. Where does the binding force come from? In some it comes from the spider gear axles wedging two collars apart. In that case the clamping forces of the two clutches are opposing each other thus must be equal.

In cases where the spreading force of the spider gears load up the clutches individually we have to look at the left and right spreading forces. That's easy, we have already shown that the spider gear applies EQUAL force to each output gear. Since that force is equal, the clamping force on each clutch must be equal thus the locking torque must be equal. Again if you disagree explain the math.

I have shown a mathematical proof. Either my math is wrong or I am correct. If I am wrong you should be able to show WHERE my analysis is flawed. Show me the detailed step where the mistake was made. If you can't, well I've shown a complete and sound theory. You have not.

I do. That is why I started thinking about this. I wanted to understand WHY the car would understeer then quickly transition to oversteer. I hadn't noticed the same behavior in a nearly identical car with an open diff.

I have also seen the benefits based on side by side drives. I have seen the drawbacks. Notice I never said don't get a LSD, only I have said they have drawbacks and most people don't understand the tradeoff.

My math fully defines the function inside of the diff. My model also fits my real world observations as well as those of a former Indy car engineer who worked on the 1980s Torsen diffs.

I wouldn't concede that because it is completely untrue. The math I showed came first from observing real life behavior of a car then asking what would cause that behavior. I'm sorry you don't like the explanations I presented but really I'm not sure you have shown enough knowledge of the material to say I am wrong. I had to try to show you that an open diff sends the same torque to each wheel. That is a very fundamental aspect of an open diff. If open diff fundamentals aren't understood then the extra complexity added by a LSD mechanisms isn't likely to be understood.

But claiming my model is wrong without an explanation is also wrong. There are two ways to show that my model is wrong. One, is to explain where one of the details is wrong. This is a very robust way of showing that I am wrong. You have said you aren't willing to do that. The second would be to show how my model doesn't account for real world behavior. That is a harder proof because you must now explain how the results I claim couldn't be true based on real world observations. The problem with that is the real world observation I and others have made support my model.

It is clear I haven't changed your view. However, you haven't shown WHY I'm wrong. As best I can tell you don't understand how what I have said would affect handling but you have it in your head that what I have said doesn't jive with something so it must be wrong. I would suggest you spend more time thinking about the problem. Again, you thought that an open diff wouldn't send equal torque to each wheel. If you were wrong about such a fundamental point isn't it possible you are wrong about a more complex subject?

 

I'm only posting it to show BMW has decided that a speed sensing, not torque sensing LSD is their choice for some of their cars.

 

You are predicating things on untrue statements. The spider gears transmit the same torque to each wheel regardless of wheel speed differences. Fine. The clutches transmit torque based on how tightly they are bound. This torque comes from the carrier as it spins. When the car is going straight and one wheel is on a sandy patch, for instance, it offers less resistance to the forward rotation of the carrier and spider gears, so the internal clutch-plate stack loosens on that side and binds on the other side. Both wheels are still going the same speed, but the torque on the side with traction has increased. Based the increased resistance to rotation relative to the loose side, or to put it another way, based on the ability to apply torque to the wheel.

The spider gears are pushing on both axles, but there is less resistance on one side, so it isn't driven against the side of the carrier. The other side has more traction, so there is more resistance to axle rotation, and therefore the spider gears push it against the carrier with more force. The carrier is rotating, one wheel has less resistance to rotating, and in an open diff it would begin to spin faster than the other wheel, forcing differential action. The binding clutches on the side with traction serve to limit the differential action.

I see nothing in what you wrote that disputes with this so far. You might argue with the mechanics of the operation, but you have repeatedly insisted the slower wheel always gets more torque with clutch-type LSD's. I'm merely saying that in straight-line operation, the slower wheel may well get more torque, but it isn't relative wheel speeds that drive the torque split.


Now if you make the assumption that the clutch-plate stacks always have the same pressure side to side, then why are the stacks even there? You can use a locker to endure locked behavior under acceleration, i.e. any increase in torque being delivered to the carrier. There is no need for the clutch-plate stack.

Where the stacks differ in operation to the locker is the ability to provide a range of "lock-up," in that the ability to deliver a differential action ranges from full open to completely locked. If both wheels have the same amount of traction, there is no functional difference between an open and LSD of the clutch type.

On the other hand, with a viscous or the BMW type you linked, it is only when wheel speed differences exist that differential action is resisted. And with speed-sensing LSD's, the distribution of torque from side to side is dependent on the difference in wheel speed, side to side.

With torque-sensing LSD's, the difference in torque distribution is dependent on the ability of the wheel to apply that torque.

Speed-sensitive LSD's do not bind harder as the average speed between sides increases. Clutch-type LSD's do bind harder as total torque is increased. So it is incorrect to say that the terms "speed-sensing" and "torque-sensing" refers simply to the input at the carrier that motivates lock-up.

If clutch-type LSD's operated as you insist, they would be called "speed-sensing" as your description claims that clutch-type LSD's resist differences in wheel speed only.

They are called "torque-sensing" because wheel speed differences side to side have no effect on their operation. They split torque based on the difference in the ability of each side to apply torque to the wheel.