Limited slip differentials vs electronic control of brakes?

[autogyro]

http://www.youtube.com/watch?v=RnR0F42E13Y

I dont usualy like to promote companies but this video shows some of the capabilities of torque vectoring.

[autogyro]

http://www.mitsubishi-motors.com/corpor ... 18e_03.pdf

Your choice fellas.
Which would you choose?
Based on different designs mainly dictated by patent restriction.

[JoaoB]

There is a half decent discussion around here about the diffs. I'Ve been meaning to update ever since I got hold of my Ferrari F2000 book from Peter Wright. There is a cutaway section drawing of the diff there. Granted its quite old now but the design looked to be an hydraulicly actuated limited slip diff. No ramps, only an epicyclical gear train to do the initial torque split (which based on my dogy calcs on a 10 hour flight do not even split the torque 50:50 left right) and then the locking is controlled by the hydraulic piston acting on the clutch pack.

I'm pretty sure its not torque vectoring because torque vectoring diffs have a step-up stage of speed and this was missing in the drawing.

As to what they have now, I'd love to know. This clause about using it traction control doesn't rule out torque vectoring in my eyes but I've not put much thought into it to be honest.

That is what I thought. To have active torque vectoring a E-diff need a step-up stage of speed for the diff outputs. Without them (or something similar) the only action possible is to try to equalise diff left/right output speeds.

There is a video on youtube about an Audi diff with "positive" torque vectoring. http://www.youtube.com/watch?v=ee-6Tvd0q34

[JoaoB]

I think you could do torque vectoring just with left- and right-side clutch engagement. Effectively lock (or partially lock) one half shaft to the input, force the load path. Hell, a Detroit Locker does "torque vectoring" by locking one or both output shafts while letting the other ratchet.

In any event, you don't need left/right torque vectoring to control under/over-steer balance on entry and exit. Can do it plenty with equal L/R clutch engagement. Hence my point of... why add the extra complexity? I mean, maybe it's a gain, maybe not.

Decreasing the speed difference between left and right output shafts is in fact torque vectoring but it works in a much more limited way then when you can actively speed up the outside wheel like you can with the set-up gear arrangement type... at least I believe so.

[langwadt]

The FIA control the technology used in F1 differentials.
AFAIK they are all electronicaly controlled mechanicaly operated torque vectoring designs.
They must not vector torque to achieve traction limiting from the complete axle.
This (as is the case in most powertrain regulations) eliminates all other types other than the FIA's choice.
Because of the limitations forced by regulation, the way the current diffs work adds to the problems achieving balanced braking, energy recovery and controlled rear tyre wear.

yeh I think the rules say it must work like a mechanical diff, and any kind of vectoring will probably be banned as 4 wheel
steering just like mclarens extra brake pedal

[Tim.Wright]

I think you could do torque vectoring just with left- and right-side clutch engagement. Effectively lock (or partially lock) one half shaft to the input, force the load path. Hell, a Detroit Locker does "torque vectoring" by locking one or both output shafts while letting the other ratchet

To me thats not torque vectoring because it can only increase the torque of the slower wheel and decrease the torque of the faster one. Any diff which uses clutch plates between the carrier and the output shaft can only add torque to the wheel when its going slower than the carrier (typical inside wheel condition). While the faster wheel is effectively braked by the coupling to the carrier (typical outside wheel condition).

To my eyes, true torque vectoring is able to bias torque to either wheel regardless of the rotational speeds and it does this by coupling the friction plates between the output shaft and a stepped up stage (instead of the carrier). That ensures that the clutch torque is always adding to the wheel torque, never subtracting. This way when you apply the left clutch, it increases the left wheel torque regardless of the wheel speeds

[autogyro]

To me thats not torque vectoring because it can only increase the torque of the slower wheel and decrease the torque of the faster one. Any diff which uses clutch plates between the carrier and the output shaft can only add torque to the wheel when its going slower than the carrier (typical inside wheel condition). While the faster wheel is effectively braked by the coupling to the carrier (typical outside wheel condition).

To my eyes, true torque vectoring is able to bias torque to either wheel regardless of the rotational speeds and it does this by coupling the friction plates between the output shaft and a stepped up stage (instead of the carrier). That ensures that the clutch torque is always adding to the wheel torque, never subtracting. This way when you apply the left clutch, it increases the left wheel torque regardless of the wheel speeds.

Exactly Tim, I think you might have missed my two previous posts.

[g-force_addict]

I think you could do torque vectoring just with left- and right-side clutch engagement. Effectively lock (or partially lock) one half shaft to the input, force the load path. Hell, a Detroit Locker does "torque vectoring" by locking one or both output shafts while letting the other ratchet

To me thats not torque vectoring because it can only increase the torque of the slower wheel and decrease the torque of the faster one. Any diff which uses clutch plates between the carrier and the output shaft can only add torque to the wheel when its going slower than the carrier (typical inside wheel condition). While the faster wheel is effectively braked by the coupling to the carrier (typical outside wheel condition).

To my eyes, true torque vectoring is able to bias torque to either wheel regardless of the rotational speeds and it does this by coupling the friction plates between the output shaft and a stepped up stage (instead of the carrier). That ensures that the clutch torque is always adding to the wheel torque, never subtracting. This way when you apply the left clutch, it increases the left wheel torque regardless of the wheel speeds

It makes sense.
For a limited slip diff -whether it's controlled mechanically, by the driver or electronically- it can only go from fully unlocked (open differential) to fully locked (kart-like axle) and anything in-between of course.

As for having a faster stage that can be coupled on demand to the left, right or none, I find this super interesting.
Has it been done before?
Could you please elaborate? Can you show us some examples?
Again, great post.

[autogyro]

http://www.mitsubishi-motors.com/corpor ... 18e_03.pdf

Table 1 page 22 should give you an idea whith the schematics.

The types shown with brakes that slip and hold components to the diff housing would be considered traction control IMO.
The clutch types can also be used for traction control across the axle but it would depend on the controls used and the FIA opinion.

However I was told that some teams use a Ricardo system, braked or pure clutch I do not know.
The Mitsubishi sytem as developed (later type) would seem to give a finer control capability, mainly due to the increase in rpm (and reduction in torque) inside the primary bevel gear set.
This gives a better rpm and torque trade off between the primary diff and the counter (torque transfer) shaft.
The ricardo diff shown IMO will give less control and could be one reason that some teams were suffering from rear tyre wear, difficulties under braking and wheel lifting because of unpredictable steering effects last year.
However, I do not have access to enough team data on their systems and I am unsure as to how much downforce maskes and hinders diff control in all circumstances so this is more of a guess.

[autogyro]

Decreasing the speed difference between left and right output shafts is in fact torque vectoring but it works in a much more limited way then when you can actively speed up the outside wheel like you can with the set-up gear arrangement type... at least I believe so.

A limited slip differential LSD is what it says it is.
A differential design that prevents (or reduces) wheel spin (slip).
The diff itself does not vector torque, the wheel and tyre on the potentaily slipping side vectors the torque from itself to the other side.

A torque vectoring diff can also do this if it is designed to, however the torque vectoring is achieved by outside control over clutch packs and disc brakes.
There are other methods, hydraulic, or electro magnetic.

[JoaoB]

http://www.mitsubishi-motors.com/corpor ... 18e_03.pdf

Your choice fellas.
Which would you choose?
Based on different designs mainly dictated by patent restriction.

Interesting paper to study!

[JoaoB]

Decreasing the speed difference between left and right output shafts is in fact torque vectoring but it works in a much more limited way then when you can actively speed up the outside wheel like you can with the set-up gear arrangement type... at least I believe so.

A limited slip differential LSD is what it says it is.
A differential design that prevents (or reduces) wheel spin (slip).
The diff itself does not vector torque, the wheel and tyre on the potentaily slipping side vectors the torque from itself to the other side.

A torque vectoring diff can also do this if it is designed to, however the torque vectoring is achieved by outside control over clutch packs and disc brakes.
There are other methods, hydraulic, or electro magnetic.

Well, IMO anything that goes beyond an open diff, as it has an effect in the line the car want's to follow by means of changing torque or wheel speed on traction wheels, is torque vectoring in the open sense of the definition.
But I understand your point... we are talking about positive/active torque vectoring.

[autogyro]

Well, IMO anything that goes beyond an open diff, as it has an effect in the line the car want's to follow by means of changing torque or wheel speed on traction wheels is torque vectoring in the open sense of the definition.
But I understand your point... we are talking about positive/active torque vectoring.

Limited slip differentials are operated by the wheel and tyre on the side of the car where wheelspin (breaking of traction) occurs.
The sudden increase in drive shaft speed on the slipping side operates the mechanism that locks that shaft converting the diff from open to spool.
Ramp LSDs and gear ramp LSDs do this more suddenly and less progresively than plated diffs.
Unfortunately plated diffs will still allow wheel spin on one side at higher torque inputs from the drive train (competition use etc) unless the friction and steel discs in the disc packs are held together with a fairly high static torque setting.
The result is a reasonably controlled vectoring of torque from the slipping wheel to the other wheel to maintain drive.
The problem with this is that during tight cornering without wheelspin, the diff is prevented from operating properly as an open diff and the clutch pack transferes the torque loading on its plates to the tyres which increases tyre wear and increases the torque loss in the powertrain. It can also produce cronic understeer on rwd.
Wedge or gear wedged LSDs tend to be unpredictable and upset handling especialy on a steering driven axle.
My choice for the 4x4 range rover we used to cross the sahara was a plated rear dif set at 120 ft ibs and a jack knight 'locker' diff on the front axle which was only used in deep sand with the transfer box locker.
I have used both plated and roller lock diffs in racing minis, both cause problems.
The roller lock tends to throw the car sidewise on engagement and the plated less suddenly increases oversteer.
I believe some of the gear wedge type diffs are better at least in heavy cars although I used an open diff in my drag mustang because the huge tyres gave better cross axle traction.

[riff_raff]

To me thats not torque vectoring because it can only increase the torque of the slower wheel and decrease the torque of the faster one. Any diff which uses clutch plates between the carrier and the output shaft can only add torque to the wheel when its going slower than the carrier (typical inside wheel condition). While the faster wheel is effectively braked by the coupling to the carrier (typical outside wheel condition).

To my eyes, true torque vectoring is able to bias torque to either wheel regardless of the rotational speeds and it does this by coupling the friction plates between the output shaft and a stepped up stage (instead of the carrier). That ensures that the clutch torque is always adding to the wheel torque, never subtracting. This way when you apply the left clutch, it increases the left wheel torque regardless of the wheel speeds

I would describe any system that employs a friction clutch as "torque limiting" rather than "torque vectoring". Using a friction clutch is inherently inefficient, since the clutch slippage dissipates energy rather than transferring it.

[autogyro]

I would describe any system that employs a friction clutch as "torque limiting" rather than "torque vectoring". Using a friction clutch is inherently inefficient, since the clutch slippage dissipates energy rather than transferring it.

An LSD is torque limiting but it also vectors torque.
It vectors the lost torque from the slipping wheel to the gripping wheel, if it did not it would be an open diff and the torque would be lost resulting in no drive and powertrain over reving.

However maintaining drive in wheel slip conditions with an LSD comes at a high price in handling, car control and lost torque from heat.

An LSD of any type is a crude devise.