Quote from J.B. :

stuff

uh?

afaik, preload only affects the transition from unloaded to loaded state, and does not affect the maximum locking factor. am i wrong?

edit - nevermind, i got it. preload only affects the locking if set too high...

Quote from evilgeek :

edit - nevermind, i got it. preload only affects the locking if set too high...

Yep. It kicks in when the torque difference gets low. What you are setting is when exactly this happens.

BTW, maybe you could be bothered to edit the my quote in your above post as there were some errors in it.

So, more locking results in less torque to the outer wheel? I guess I'll just have to let that sink in and think over it again, because now I'm actually more confused than before

Quote from AndroidXP :

So, more locking results in less torque to the outer wheel? I guess I'll just have to let that sink in and think over it again, because now I'm actually more confused than before

i'm confused about the same thing, actually.

too low preload should cause the diff to slip initially, then lock up.

just right preload should let the diff work as intended.

too high preload should cause the diff to lock more than intended.

and more locking should mean more torque to the slower spinning (outside) wheel.... i think.

@J.B.
Let's imagin input torque at the diff is 0 Nm. Preload is 100 Nm. The car is going in a straight line.
Using your calculation for the case preload > input torque I'd get to a torque at the left wheel of 50 Nm and torque at the right wheel of -50 Nm.
This is obviously not true. What didn't I understand? This is the point why I assumed a clutch to operate, which requires a constant in the unit of Nms/rad.

Vain

Ohh, the joys of really quite simple things

Quote from Vain :

@J.B.
Let's imagin input torque at the diff is 0 Nm. Preload is 100 Nm. The car is going in a straight line.
Using your calculation for the case preload > input torque I'd get to a torque at the left wheel of 50 Nm and torque at the right wheel of -50 Nm.
This is obviously not true. What didn't I understand? This is the point why I assumed a clutch to operate, which requires a constant in the unit of Nms/rad.

Vain

viscous coupling is speed dependent (Nm/rad), but clutch packs are power dependent. i found this:

Quote :

Here there are a set of clutch plates attached to the inside of the housing just outside of each drive gear, and a set of clutch disks attached to the output shafts, but also still inside the housing. There are two thick pressure rings which attach to the inside of the housing near the spider gear shafts. The pressure rings push outwards against the clutch packs. The space between the pressure rings forms an opening that is wedge shaped, and the spider gear shafts themselves, form the wedge. As power is applied to the ring gear, and transmitted to the housing, and down into the pressure rings, they start to push on the spider gear shaft which (depending on traction conditions) wedges itself toward the end of the opening between pressure rings, which forces them away from each other, which applies pressure to the clutch packs, which locks both drive gears to the outside housing and under very high power the differential locks up nearly solid.

So this one is power sensitive. Note, there must also be some preload built in, for use in low traction conditions. This system relies on there being some resistance to the applied power to cause the pinion gear (spider gear) shaft to wedge open the pressure rings. If you started with no static clutch pressure, then under very low traction conditions (one side on ice), there won't be enough resistance to wedge the pinion shaft in place and it'll remain an open diff. As a result there's usually enough static load added to provide some constant torque split to both sides, and then if you hit the gas, it'll lock up the diff and you get torque down to whichever side can use it.

Just to clear one thing up: in normal cornering the outside wheel is the faster spinning wheel. The inside only gets faster if it gets overloaded and starts spinning.

So in normal cornering more torque is applied to the inner wheel and if you think about it you will notice that this difference in torque will try to rotate the car away from the corner which is the exact reason why LSD's help against oversteer.

http://members.aol.com/agspeed/LSD-lock.htm
http://www.houseofthud.com/differentials.htm#positraction (that's what you just quoted)
http://www.gordon-glasgow.org/lsdtech.html

@evilgeek:
That explains well how a LSD diff works under load, but I'm aware of that. I'm talking about: What kind of torque does the LSD diff exert under a given difference in wheel speed (cornering), with no input torque and a given preload setting.
Either the unit of the setting is wrong or I lack a fundamental understanding of what happens in that situation.

Vain

Quote from Vain :

@J.B.
Let's imagin input torque at the diff is 0 Nm. Preload is 100 Nm. The car is going in a straight line.
Using your calculation for the case preload > input torque I'd get to a torque at the left wheel of 50 Nm and torque at the right wheel of -50 Nm.
This is obviously not true. What didn't I understand? This is the point why I assumed a clutch to operate, which requires a constant in the unit of Nms/rad.

Vain

Just leave complicated stuff like driving in a straight line out of it will you?

But seriously, when left/right torque differences are very low I would imagine that the diff indeed behaves like a fully locked diff due to the friction plates not slipping but being under static friction. I think as soon as you try to go round a corner and the plates begin to slip you will notice there is a torque difference between the wheels i.e. the car isn't as willing to turn in as it would be without an LSD. There really could be -50 at one wheel and +50 at the other. As long as the forces caused by the steering front tyres are bigger the car will still go round the corner, just with less oversteer.

So maybe we have three states?

1) locked when the forces applied to the wheels don't cause the clutch plates to go from static to dynamic friction and start slipping

2) in the "preload" state: clutch is slipping and the torque difference between the wheels is constant

3) in the "normal" state: clutch is slipping and the amount of torque difference is defined by delta_torque = locking_factor * input_torque

I guess it's either that or I have gone completely bonkers trying to understand this stuff. :Eyecrazy:

Quote from J.B. :

Just to clear one thing up: in normal cornering the outside wheel is the faster spinning wheel. The inside only gets faster if it gets overloaded and starts spinning.

So in normal cornering more torque is applied to the inner wheel and if you think about it you will notice that this difference in torque will try to rotate the car away from the corner which is the exact reason why LSD's help against oversteer.

ah yes. i was getting ahead of myself and thinking about cases where the inside wheel has lost traction.

Quote from Vain :

@evilgeek:
That explains well how a LSD diff works under load, but I'm aware of that. I'm talking about: What kind of torque does the LSD diff exert under a given difference in wheel speed (cornering), with no input torque and a given preload setting.
Either the unit of the setting is wrong or I lack a fundamental understanding of what happens in that situation.

Vain

i don't know how scawen has programmed it. it could be the spring rate, but it's more likely that it's the breakaway torque value, which is the amount of torque you would have to apply to one wheel to make it turn if the other wheel was held stationary.

I read about that in those linked articles.
So the preloaded LS diff behaves like a locked diff until the torque difference reaches preload?

Vain

And then the torque difference stays at the preload value until the input torque reaches a certain level? Make sense? Where's Todd?

Quote from J.B. :

Probably this assumption is wrong for a clutch pack diff although I can't find any information on this. In fact it's correct for a viscous clutch which has Nms/rad as its unit in LFS.

i just found out its the wrong unit in the german translation

but anyway to clear it up its better to write it as
that way its pretty clear that the unit is actually
so in other words its the amount of torque that the diff can transfer as a function of the difference in rotation speed of both wheels

wheras the preload is just the _minimum_ amount of torque the diff can transfer at all times

i hope those images work

edit:

Quote from J.B. :

It' really quite simple.

An example for accelerating in a corner with 40% locking:

1. Find out how much torque is coming from the engine. For example the FOX has a max torque of 221 Nm so I'll just pick the case where I'm using about half throttle to accelerate --> torque_engine = 100 Nm

2. Multiply this torque with the gear ratio and final drive ratio to get the torque at the diff. Example: torque_diff = 100 Nm * 1.6 * 4 = 640 Nm

3. use

delta_torque = locking_factor * torque_diff

to find what the torque difference between the two wheels is

--> delta_torque = 0.4 * 640 Nm = 256 Nm

Now if this number is bigger than the preload then
--> torque_inner = 640/2+256/2 = 448 Nm
--> torque_outer = 640/2-256/2 = 192 Nm

If the preload is set to more than 256 Nm, say 300 Nm then
--> torque_inner = 640/2+300/2 = 470 Nm
--> torque_outer = 640/2-300/2 = 170 Nm
--> more locking, less oversteer than before

I hope that clears up some confusion.

it should be noted that what you calculated isnt necessarily the actual difference in torque at the wheels in that distuation so your numbers for torque_inner and torque_outer arent necessarily correct
if im not mistaken what it actually is is the _maximum_ difference the diff can sustain without diff slip

Quote from Vain :

So the preloaded LS diff behaves like a locked diff until the torque difference reaches preload?

if im not mistaken all lsds will behave like a locked diff to the point where the difference in torque attracted by the wheel overcomes the amount of torque the diff is able to transfer in that situation

This is very appropriate timing, I've just been explaining diffs and their effects in my project report. Of course I haven't had to cover pre-load since it's not in my analyser.

If you can let me know when everyone is in agreement and then I can incorporate pre-load into my setup analyser sometime in the next few weeks. I'm thinking I'm going to have to add quite a bit, maybe a graph of torque that would be in a V shape but begin to look more U shaped with high pre-load settings?

Shotglass - that locked diff until overcome thing doesn't seem right to me. I need to build a mini LSD and play with the driveshafts in my hand, it's the only way...

I found some videos on preload:
http://www.youtube.com/watch?v ... ;mode=related&search=
http://www.youtube.com/watch?v ... ;mode=related&search=

Quote from J.B. :

So maybe we have three states?

1) locked when the forces applied to the wheels don't cause the clutch plates to go from static to dynamic friction and start slipping

2) in the "preload" state: clutch is slipping and the torque difference between the wheels is constant

3) in the "normal" state: clutch is slipping and the amount of torque difference is defined by delta_torque = locking_factor * input_torque

I guess it's either that or I have gone completely bonkers trying to understand this stuff. :Eyecrazy:

Thats the way I see it

1 and 2 are both under preload conditions without any or minimal dynamic torque transfer from the driveshaft. i.e. off throttle, brake and or gearbox in nuteral. 1 discribes straight line conditions or mild cornering were the difference in torque between the two wheels (generated by the momentum of the car and different paths the outer wheels are traveling) is not enough to overcome the preloaded clutch plate and cause them to slip. 2 deiscribes the same cornering were the forces are enough to cause the plates to slip.

3 is the normal operation of the diff (which is now only in effect above the preload setting) and can still be split into power and coast so there is 4 phases I guess. Also for the dynamic torque transfer there must be two states were under certain conditions it give a fully locked characteristic and a semi locked (slipping plates) also no?

So that would make it how many states of operation

@ Bob in my imagination I'd think the curve would look like a cross between a V and a U but with a flat bottom

Edit: One thing about the LSD that still bothers me though is the RPM drop you get when you trail brake into a corner with low preload settings on FWD cars. Even under conditions were I'm fairly certain that one wheel isn't locking this seems to happen. This was discussed in another thread some time ago:
http://www.lfsforum.net/showthread.php?t=5398
But I'm still not convinced? Hmm I'll do some more testing later this week and see. Could it be when the diff is working under normal dynamic conditions but is going from wheel locked state to clutch slipping state that has a bug? Or is this normal operation of a 1.5 or 2 way LSD on FWD? I did read that for FWD a 1 way LSD is best IRL, maybe that was the reason? Anyone with RL experience in this here?

Quote from Glenn67 :

Thats the way I see it

1 and 2 are both under preload conditions without any or minimal dynamic torque transfer from the driveshaft. i.e. off throttle, brake and or gearbox in nuteral. 1 discribes straight line conditions or mild cornering were the difference in torque between the two wheels (generated by the momentum of the car and different paths the outer wheels are traveling) is not enough to overcome the preloaded clutch plate and cause them to slip. 2 deiscribes the same cornering were the forces are enough to cause the plates to slip.

3 is the normal operation of the diff (which is now only in effect above the preload setting) and can still be split into power and coast so there is 4 phases I guess. Also for the dynamic torque transfer there must be two states were under certain conditions it give a fully locked characteristic and a semi locked (slipping plates) also no?

So that would make it how many states of operation

Great! I understood every word and what you say is 100% in agreement with my understanding of the subject, including the things I hadn't thought of before like the case where the diff is locked while not in preload mode. To me this is an indication that our reasoning has been sound. It's been a nice discussion, now let's hope someone with RL experience drops by. Oh, and BTW that's 8 states of operation. :eek:

Another thing I thought of: with an open diff any friction in the gears would pretty much have the same effect as preload. Would real cars have enough friction in the diff to influence the handling of the car towards understeer? If so I guess sims would need to add preload to open diffs as well.

Woah that's a lot of input for my small brain

But one thing I didn't find out is this:

How do you setup the "new" diff option?

Let's say I've set the power side on 55% and the coast side on 75% (on an XRT).
Now I've read that a good range for preload is 400-600nm but there must be some regulas that you can rely on? (this range works fine but it's the fine-tuning that is bothering me)

The next person who says"Its perfectly simple" then fails to explain this in simple non jargon language is getting a punch in the throat.

Wouldn't any clutch preload setting greater then the torque outputed by the transmission just make it a locked differential?

Yes, isn't that the conclusion we came to?

LRB_Aly - I'm not able to try out the W9/W10 test patches yet so I can't comment, but that's a massively locked diff for the old XRT. Personally I think 40/25 is plenty, diff locking just adds unnecessary understeer, IMO the less you can get away with the better (so long as you can put the power (or rather, torque) down effectively, of course).

From my searches I've found out that the M3 has a 25%/25% with 45NM preload diff.
If anyone has more info from the real cars, please post.