That's not necessarily the LSD causing the change in behavior. Tires can cause that - instead of a gradual change in grip, it only holds on to a point then lets go completely. Different tires will demonstrate differences in how they react approaching the limits of adhesion.

Car setup can cause that as well. For instance, the rear suspension can bind as the load builds, then slip, causing a sudden unweighting of the inside tire. Or wheel camber can change as the inside rises until the contact patch is no longer square to the road surface and you lose half your traction over a very small amount of travel.

A RWD car will understeer as rear traction overcomes front, until the point rear traction drops. If this happens suddenly, there are a number of possible explanations. Just because this doesn't happen in a different car with an open diff says nothing about the behavior of an LSD. It could very well be the behavior of the specific LSD you have installed combined with the rest of the car setup, including tires. It certainly doesn't mean "LSD's cause understeer and snap over-steer."

Any more than "I drove my friend's truck and it understeers, and my car doesn't do that. We have different tires, so truck tires make vehicles understeer."

You are finding a way to explain the behavior of numerous components in terms of one single element. Sometimes that's OK, from an "all else being equal" stand. But all else is rarely equal, and I'd expect a car needs to be setup for the particular differential you are using. IOW, if you are using an open diff and swap in an LSD of some type, you need to make some changes to other aspects of the car in order to get the most benefit. Same as swapping different types of tires will demand changes in car setup to extract the most benefit from the change in rubber. You wouldn't just say "softer race tires make my car handle weird, so soft race tires make cars handle weird."

 

Great discussion fellas.

On the extreme talk of the FF, as was already mentioned and has to be reinforced....the centre of gravity is very low, the lower the centre of gravity the less weight shift will occur even under higher and higher lateral g forces....meaning you really do not need an LSD. However....take formula 1 cars as another example....they also have extremely low centre of gravity and have the stickiest of the ickiest tires around. However they have more than enough torque to overwhelm the tires in the lower speed corners...these cars not only come with mechanical LSD's they also typically run very strong settings to try and equalize tire speed as much as possible. these cars are not set up to be fun to handle, they are set up to be as fast as possible.

I definitely understand what Rockville is saying in the terms of handling...maybe I didn't say it in the best of words when I said that LSD's make a car more difficult to drive...just that you have to drive the car differently. My first car was a Datsun B210 (37 horsepower, RWD, wheel barrow tires, open diff). This car had so much body roll that even with its low power was easy to light up the inside rear tire going around corners....this car was impossible to spin completely out of control on a dry road with even the most daring hoonage. Even though it was super easy to light up the inside rear, I could never get the outside rear to break out. Great handling fun car to drive...not very fast.

My last car was a regular Mazda 3 sedan, steering feel was a little numb because it was a hybrid steering system, part pump, part electric. Open differential though, also meant that when beating on the car it was incredible safe...it did of course understeer when pushed to the limits but it was very controllable, again only one tire would spin a lot. Now I have a Speed 3 normal power steering, 1 way LSD, lowered, stickier tires, and a lot more power....the car has tons more available grip and is significantly faster in the corners...however when applying too much throttle the car can easily break both front tires loose in a low speed corner which literally will pull the car out of line and it is actually down right scary the couple times that has happened to me. I would say I get more feedback from the Speed 3 and it is a sportier more fun car to drive overall...however the overall handling feel is somewhat unpredictable...this is not simply because of the extra power...its really more because of the LSD.

I say this because I did go back and read the McLaren thread...I can understand McLaren not implementing an LSD to just make a safer car to drive at the limits. The car has like 600 horsepower and its fairly light...it comes with street tires....I bet they will put a mechanical LSD in the GT car. Maybe they just don't want to see McLarens getting wrecked and they found the car was easier to drive with an open differential.

I think in reality LSD's when set up properly make just about any kind of car faster, other than maybe some extreme examples...but I also agree that their is no perfect system...blame it on physics, because of G forces you will never have a perfectly loaded car.

 

Nunco,
I think we need to go over a few fundimentals. All of my arguments are based on these points. If you disagree with any point please say why. I think you are having a hard time wrapping intuition and previous knowledge around the points I have made.

The Fundimentals:
1. The spider gears in an open or LSD ALWAYS deliver equal torque (in steady state conditions which is just about always).
2. All traditional LSDs (ie not externally controlled, e-LSD etc) apply some sort of locking torque to the outputs of the wheels. That locking torque can be a fixed value or variable. In the case of a variable system it could be a function of input torque (Torsen) or a function of the differences in left and right wheel speeds (Viscous). That means the MAGNITUDE of the locking torque will be a function of either delta wheel velocity or torque in.
3. The torque to the wheels is always of the form T_output=1/2 T_diff +- T_locking (or 1/2 T_locking depending on how you express T_locking). Locking toque refers to the torque in the diff that tries to make the output shafts and the housing spin at the same speed. It is easy to see how this works with a simply clutch diff. More complex diffs such as a Torsen have different parts but work the same way.
4. The DIRECTION of T_locking is always such that it tries to make the output shaft and diff housing spin at the same speed. That means if the output is faster the locking torque will try to slow the wheel. If the output is slower the locking torque will try to speed up the wheel. This means the wheel that spins slower will see an increase in torque as it gets 1/2 the diff torque (ie the spider gear torque) + the help from the clutchs. The faster wheel will see a decrease in torque as the clutch torque will be opposed to the spider torque.

If you want to disagree you need to start by proving one of those claims untrue. Those 4 points are the foundations for my argument. You haven’t shown any of them to be wrong and you seriously challenged your own credibility by questioning #1.

See point #4, when a wheel spins slower than the diff housing the DIRECTION of the torque added to the output is WITH the spider gear torque. Think about it, if the diff housing is going FASTER than the wheel will the clutch at that wheel try to speed the wheel up or slow it down? When the wheel is going faster than the diff housing it is subtracted. This is very simple math and the basis for all I have said. If you disagree SAY WHY. Don't claim I'm wrong if you can't explain why my math and physics are wrong.

Your sandy patch example actually perfectly fits with my description. Both wheels start off at the same speed. Once the wheel hits the sand it has less resistance and starts to slip. When it slips under power it will go FASTER than the wheel on pavement. I've said torque transfers from the faster to the slower wheel. In this case that means torque transfers from the wheel in the sand to the one on the pavement. The AMOUNT of torque is dependent on the input torque if we are dealing with a torque sensing diff (see point #2). The direction of transfer is ALWAYS dependent on the speed of the wheel with respect to the diff. My math proves that. The torque transfer doesn't occur because the diff knows how much traction each wheel has, it happens because under a given load one wheel, the one with limited grip, starts to speed up.

You are confusing the total torque applied to the output with the torque applied by the spider gear. The spider gears ALWAYS behave like an open diff. Physics tells us if the forces on the two outputs APPLIED BY THE SPIDER isn't equal the spider MUST be speeding up. That means the spider MUST apply equal force to both outputs. Any difference in torque at the output shafts comes from the addition or subtraction of torque related to the clutches (or other anti-slip mechanism). That again holds with my statements. The wheel in the sand slips thus momentarily speeds up thus a transfer of torque to the slower wheel on the pavement.

I'm not sure which part I would be disputing or not but your last sentence is wrong, the torque split is ALWAYS with respect to relative speeds. Even a very small delta in wheel speeds will result in transferring torque to the slower wheel. Think about it, any relative motion will result in the clutches moving which means torque was needed to move the clutches. That’s how the LSD works.
This means you don't understand what happens. The MAGNITUDES of the clutch torques are equal but their DIRECTIONS are different. Did you ever take a statics class? Because the DIRECTION of the torque is different one ADDS to the spider's applied torque one subtracts. Net result is one wheel sees 0.5T_diff-T_clutch. The other sees 0.5T_diff+T_clutch. In one case we add, in the other we subtract. That means the torques will NOT be equal. I would suggest you reread my first post and look at the pictures again.

Of course the clutches allow for a range of lock up. With a torque sensing diff the magnitude of the locking torque is proportional to the torque into the diff. That I think is consistent with what you are claiming.

True.

Not exactly true. The distribution of torque is ONLY dependent on which wheel is going faster. However, the wheel with less grip slips thus goes faster (at least in a straight line) thus the wheel that slips gets less torque but only because it's the faster wheel.

Since you want to claim torque goes to the wheel with grip, let me ask you a simple question… How does the diff know which wheel has the grip? I have a simple answer. The wheel with insufficient grip will slip. When it slips it will go FASTER than the other wheel.

That is not true. Speed sensitive LSDs have a locking torque curve that is based on relative wheel velocity. The shape of that curve can be something like a viscous diff where it slopes up then levels off or like the BMW system it can use relative wheel speeds to drive a pump which locks the clutch. That setup can be made to be basically open for small wheel speed differences and totally locked for large ones.
That is just plain wrong. Your lack of understanding doesn’t make that statement a true summary of what I am claiming.
In this case it was the LSD. Tires are part of the interplay but no, this was clearly the LSD.

You mentioned things that could be contributing factors if nothing else was known. In this case I knew the basic car and the suspension and tires. The specific issue was a symptom of the LSD combined with tires that needed to be replaced. However, I want to be 100% clear, the fundamental source of what I felt was the LSD. Since you don’t know how the LSD operates I can understand why you would question this.
I agree. The factory LSD was not a good match for the rock hard, cheap tires the previous owner put on the car. Replacing the tires greatly improved things. All the stuff you are talking about retuning, the reason why that retuning is needed is because the LSD behaves the way I’ve claimed! You seem to understand the need but not the cause.

Anyway, you are very vocal in your opinion but your opinion is not based on an understanding of how things work. You have perhaps tuned a car and found that it was better after a change but since you don’t know what’s going on under the skin (or in the axle) you can’t fundamentally understand why those changes work or fail. If you believe my descriptions are wrong then tell me why my 4 points are wrong. If you can’t then you don’t know enough about the subject to make a good case for your view.

 

I think you are almost right. I think a BMW style speed sensing LSD will make any car faster than the same car with an open diff IF the diff weight isn't an issue. The BMW system has almost no downsides. It has the nice open diff behavior when wheels aren't slipping yet will stop a peg leg turn. The BMW diff does handle the one wheel on ice conditions and handles a large range in tire types (sticky or rock hard). However, I don't think it would be as fast as a properly tuned torque sensing diff and I believe some people have expressed durability concerns when the diff is used for track driving. I don't see this type of diff resulting in a gotcha type situation that you see with torque sensing diffs surprisings an owner with a spin.

A torque sensing LSD is a different mater. I think it will make most street cars faster in most conditions BUT it isn't a one size fits all. The correct tune for a car when equipped with sticky tires and a dry road is likely too aggressive for an oily, wet track. The diff that works for the oily, wet track is likely not aggressive enough to make one much faster than an open diff.

 

You continually insist that the torque from the clutch stacks on each axle are equal and therefore impart the same amount of torque to the axle. This is the basis for your insistence that torque-sensing lsds operate by resisting speed differences.

You need to accept that the friction of the clutch stacks is not equal at all times. When one wheel is on a surface that provides less traction, the stack on that side will loosen and deliver less of the available torque to that wheel. There doesn't need to be any difference in speed from one side to the other, merely a difference in traction.

The effect is as if the system is resisting differences in wheel speeds, but how it works is different than a speed-sensing differential.

Torsen literature makes it clear that no wheelspin is necessary for it to distribute torque asymetrically. The fact that you claim otherwise carries no weight. How can you claim you learned how these things "really work" at the knee of a Torsen engineer, then claim that the papers from that company are lies?

There are a number of descriptions of lsd's on the internet, and not all are accurate. There is disagreement as to what defines "torque-sensing," and disagreement as to what design is best for specific uses. I do not think it is necessary for you and I to agree, but I do think you need to accept that you may not be right in this case. It's not a static system that is always in equilibrium, despite your insistence. You can read the information I already posted, or continue to claim you are right and everybody else, including the folks who actually manufacture the Torsen LSD, are wrong. Your call entirely.

 

I insist that the torque from the stacks are equal because they are! I have even proven it even though you don’t seem to want to dispute the proof. Where I think you are confused is I’m not saying the torque to each wheel is equal EVEN when the clamping torques are equal. That’s because the clamping torques are ALWAYS opposed to each other when there is relative wheel motion.

Having reread your post I think you are stuck on the static case where both wheels are spinning at the same speed.
It is clear that you missed something that I requoted below. Perhaps for you I need to say the torques are equal when there is relative wheel movement. When the wheels are spinning at the exact same speed the torque to each wheel is 1/2 the diff input torque +/- up to what ever the clutch torque is. This is only true in the static case. This is the case when you have a box on the ground. If the static and sliding CofF's are the same and it takes say 10 lbs of force to slide box then the force of friction between the box and ground is 10 lbs when we are moving it (the equivalent to the wheels moving at different speed case I have been arguing) and between 0 and 10 lbs when the box is static. That is the special static case you are arguing. I've been saying the box's force of friction is 10 lb while you have been saying it's variable. We are both right but you are only right when the applied force doesn't move the box. I am right in the case where the box is actually moving and I actually already addressed the non-moving case as well.

As for the even clutch torques. This does assume relative motion. You aren't being clear if you agree about the relative motion case.
Basically the things I’m about to say should be easy for even an undergrad mech engineer to understand.
Here is the ramp pack from a LSD:
http://photos.motoiq.com/photos/585301708_obtmA-M.jpg
This is a common type. Note that the shaft the spiders spin on sits in a wedge pocket. When the spiders feel resistance they wedge the two out halves apart. Each half is pressed against one of the two clutches. When the ramp pack expands one have pushes on the left clutch, one half pushes on the right clutch. Since the spider axles float inside of the two halves that means the forces on each clutch pack are equal! When you lift a 2 ton car with a jack how much force does the jack apply to the bottom of the car… 2 tons. How much force does the jack apply to the ground… 2 tons.
GUESS WHAT? The ramp pack is trapped between two clutches the same way a jack is between the ground and car. The ramp pack pushes equally on both sides thus… wait for it… the forces on the clutches are EQUAL. If you disagree explain why.

Now I know you are having trouble with this one but… even with a clutch pack diff that uses the spreading of the gear teeth the clutch loads are equal. This one is escaping you because either you don’t understand kinematics or haven’t tried to figure this out.
1. What loads the clutches? Answer, the spreading force between the spider gear and the output gear teeth.
2. What is the spreading load proportional to? Answer, the torque the spider gear applies to the output gear.
3. What do we know about the force the spider gear applies to the two sets of output gear teeth? Answer, the forces are equal.
4. What is the clutch clamping force proportional to? Answer, the spreading force between the spider and output gear.
5. What does this mean? Answer, the clamping forces are proportional to the spreading forces which are proportional to the torques applied by the spider gears which are always equal. That means the clutch torques are EQUAL.
Again, if you disagree explain why?
So I would say you need to accept that the resistive torque (friction if you want to call it that) of the clutch stacks are ALWAYS equal. My claims above are absolute for steady state conditions. If you disagree and you have any mechanical sense you should be able to show where I am wrong. I suspect you don’t understand the material hence you have never gone after the technicalities of my arguments.

You claim when one wheel is on a low traction surface the clutch on that side relaxes. Well EXPLAIN HOW THAT HAPPENS. You made the claim, PROVE IT. How does the diff know which wheel has less traction? Again you made the claim prove it. My explanation shows how the wheel with less traction sees a reduction in torque. Yours seems to need magic or only is true in the case where the wheels are spinning at the same speed.
Funny, I seem to be the only one doing any explaining. Tell me how that works. How does the diff know which wheel has more traction? How does it know to transfer torque? How does it know to load up the clutches? I explain all of it on page 1. Since you seem to know that I’m wrong tell us how it works.

The Torsen literature doesn't dispute what I'm claiming. When you hit the gas you load the diff. Just like a clutch diff you have created a frictional binding force between various parts which would otherwise move with respect to one another. So long as the difference in torque on either wheel is LESS than the torque of the locking system (T_clutch of both clutches) the load on either wheel can be within a range. So if we have the equivalent of 20 ft-lb of clutch torque on each output and say apply 100 ft-lb to the diff then the torque applied by either wheel will be 1/2 of 100 +/- up to 20 ft-lb or between 30 to 60 ft-lb. I already addressed this in my first post. In this instance the diff would be behaving like a spring loaded clutch pack diff with a 20ft-lb per side clutch torque. This also plays right into what I said about cornering. With this same condition the slow wheel gets 50+20 ft lb while the fast one gets 50-20. So now we can see the torque going to the slower wheel. Perhaps this is what was confusing you. I didn't reiterate this portion of my claims. However, this was within original claims and does not apply when we are cornering and only applies when, in this case, the wheel with limited grip can still apply 30 ft-lb to the road.
On April 3rd I said the following:
So basically I agreed with the Torsen engineers. The Torsen marketers simply didn't tell you that what they said was also true of a good clutch diff and only true when both wheels are spinning the exact same speed.

I know there are a number of LSD descriptions and I’ve read many that are wrong… what little you have offered included… mine is not wrong. You certainly haven’t shown it to be wrong and I really doubt you have the understanding to do so. I have addressed the non-static case and explained the degree to which the non-static case affects my explanations. If you disagree, SHOW ME WHY. You have yet to post anything of merit that I haven’t already shown to be wrong or shown to be in agreement with. If you feel I haven’t addressed a technical point that you made show me. Better yet, if you understand the material it should be easy for you to show why my explanations and MATH are wrong.

I challenge you to show where my explanation is wrong. You say clutch loads won’t be equal then explain the torque flow starting from the ring gear. You challenged me to do this. I did it and have posted it. You have never said why my description is wrong. If you can’t explain WHY I’m wrong then why should we assume you are right and I am wrong? I’ve found plenty of people who don’t know how these things work but you sure seem the most determined to prove that you don’t understand. Since you disagree why don't you find a detailed description including power flow diagrams... or you can just show where my description breaks down in detail. I repeat this challenge because I don't believe you have the knowledge to do so.
Basically that beautiful, instant, perfect torque transfer only occurs when the diff ISN'T doing it's other important job, allowing the wheels to spin at different speeds!

BTW, the guy who designed the Torsens used in the old March Indy cars from the 1980s DOES agree with me.

 

I've explained it repeatedly. You ignore it. You've not responded to one thing I've written except to write, "You're wrong," and simply retype the same nonsense you've been typing all along. It's like you don't even process the things I've written before you start typing again.

I started with the torque sensing design using clutches and plates in a straight line because that is the simplest mode of operation. If you cannot grasp their operation in a straight line, how could you claim to know how they work?

Why do you refuse to participate in a discussion? You simply refuse to consider what anyone else has to say. You write retardedly long posts expounding your point of view, sprinkled with bits of other people's posts, then ask me to show you why I disagree. Um, wasn't I showing why I disagree when you were going off on your ego trip?

You can still maintain your personal sense of infallibility. Just cut the college boy crap. It's wildly obvious you just got your degree and want everyone to know. God forbid they let you graduate without knowing absolutely everything in the universe.

I'm not the only one telling you that your theories are not entirely accurate. However, you are the only one arrogant enough to be condescending and insulting. And that does imply your understanding of the subject is incomplete, and you are sensitive about it.

 

You haven't explained it. If you have tell me which post. I have taken most of your posts apart paragraph by paragraph. Tell me which posts I failed to deconstruct or which deconstruction was wrong.

I told you that in a straight line the clutches will allow the diff to transfer up to T_clutch to either wheel when neither wheel is slipping. I agree with you on that point and said as much on April 3rd. Why would you say otherwise? I also said the moment the difference in grip exceeds 1/2 T_diff - T_clutch (eg one wheel on sand with lots of power) you end up with wheel slip. At that point the wheels aren't spinning at the same speed. Do you disagree?

If you want to say things are different then discuss how they are different. I've offered math and diagrams, you have not. You suggested I look at the torque flows. I did exactly that. Now tell me where those torque flows are wrong.

I maintain the personal sense of infallibility because the math is on my side. You have not shown why I am wrong. Please don't lecture me on being insulting. This was started off when you insisted I didn't know what I was talking about but you didn't offer a proper explanation. BTW, no I did not just get my degrees. I've had them for quite a while. What I did was take a commonly misunderstood device and reduce it to principles. You are having trouble understanding or at least communicating why you think my principles are wrong.

It is your turn to show your knowledge. You need to explain things to the same level of detail I have or at least show where my level of detail is flawed. Again, since you didn't know how an open diff worked forgive me for questioning your knowledge of a LSD diff.

Nunco,
Even in this post you have said I'm wrong but only in vague terms. You have not shown WHERE my claims are flawed. It is now your turn to tell us how these things work. If you are willing to tell us and answer my detailed questions then I will happily admit we had an internet disagreement and a failure to communicate.

 

From this earlier post of yours. The first part basically represents the case when the differences in traction are small enough (which is the torque bias ratio) that both wheels are spinning at the same speed. In this case you would be correct that both clutch torques are in the same direction. This was the case I acknowledged on 4-3. Where you are wrong is that the clamping force on each clutch pack will be the same as I showed before. Since we are dealing with a static friction case the resistive torque of each clutche is simply known to be less than T_clutch. The actual value is unknown but the direction, as you indicate will be additive with T_spider. In this special case T_spider will be less than 1/2 T_diff because we can't have more torque leaving the diff than enters the diff. This again is the special case where a LSD actually behaves exactly like a spool. This case does not apply when the wheel speeds are different and we don't want this to happen when we turn. You would be correct in saying I didn't make this part of my argument clear enough because I have been concentrating on the cases where we have relative wheel speed. This is perhaps the source of your confusion with regards to what I have been saying.

This part I discussed I think on the 3rd on the first page of the thread. I don't have time to restate my claims but I will later today and you are then free to point out why you agree or disagree.

 

Might I suggest pistols at dawn?