RE: Ask the expert: All you want to know about differentials

[V8 FOU]

Captain Muppet said:

V8mate said:

Strawman said:

Dagnut said:

3) Do sheep get really heavy when it rains?

They get heavier, but not as heavy as you'd expect due to water repelling oils in the wool.

Surely the fact that they shrink compensates for the additonal weight?

Do Welsh sheep have more limited slip than New Zealand?

[kourgath]

jatinder said:

I've heard you can knacker your diff, if just messing about in the snow?

Is this true or rubbish? I heard it the Integra forums.

I think rubbish, but still just wondering….

Edited by jatinder on Thursday 23 February 15:11

True - you can knacker a normal open diff with messing in snow. I broke my LR front diff by letting it spin for too long (off-road event setting up), this spun the oil out, the planet gears then heat welded themselves to the cross shaft and bits then fell out. The open yoke allows the oil out much quicker than the closed yoke (centre casting) of LSD and in LR case 4-pin diffs.
Open yoke can be seen here
http://www.ashcroft-transmissions.co.uk/images/upl...
and here
http://lh4.ggpht.com/_csUOTOJI4tw/St8avK4EmCI/AAAA...
from a Honda CRX
http://i912.photobucket.com/albums/ac329/lsvtecarl...
http://www.tlarengineering.com/images/garage/honda...
closed centre as example for 4-pin, LSD, locking etc.
http://www.ashcroft-transmissions.co.uk/images/upl...

HTH
G

[cheadle hulme]

kedaha said:

This might make me sound stupid, but for a long time I found it hard to actually fully grasp how and diffs do.

A 1930's video helped me figure it out
http://www.youtube.com/watch?v=K4JhruinbWc&lis...

Thanks, the other youtube animations are good too.

[Kawasicki]

rhinochopig said:

Kawasicki said:

jagfan2 said:

Marf said:

Kozy said:

For a FWD racing car, is a helical or plate diff a better option?

Do Quaifes/Torsens/helicals work under braking?

Helicals require both wheels to be loaded in order to work, i.e. if one wheel is in the air, then no power is being transmitted to either.

Plate type diffs do not have this little foible.

Plate diffs have the same issue, but get over it with preload (a base level of locking). Torsen did develop a preloaded unit for racing, but as its less tunable and heavier, everyone stuck with plate diffs

Torsens are generally better for road cars as they are more predictable, and dont wear out as quickly. And you can easily fix the unloading issue by applying brake torque to the lightly loaded issue when it starts to spin up again. Audi us this principal a lot, and have taken it further with brake torque vectoring, as have Porsche

No one has asked anything about Torque vectoring diffs, this is the future and where the clever stuff happens

Edited by jagfan2 on Thursday 23 February 22:49

From my experience it works great for sub limit agility, but (so far) I've been unimpressed with the results of torque vectoring on limit driving.

For limit driving how is it logical to add to the vehicles kinetic energy? You've gone in too fast? Here is some more speed(energy)!

Linking active steer seamlessly to the DSC package is, for me, where the clever stuff happens.

These chaps disagree.

Watch from about 5:50m http://www.youtube.com/watch?v=0YJ-9GE0Re0 or skip to the end to see their conclusions.

Torque vectoring might be better in a controlled track situation, I've no problem with that.

What I am talking about is different. Take a corner, where the maximum speed a vehicle could carry through is 60mph. Which would be better if you entered that corner at 61mph, DSC or torque vectoring?

[WeirdNeville]

Q: Can an electronic system acting on the brakes across an open differential ever be a true analogue of a mechanical LSD?

[anonymous-user]

Kawasicki said:

What I am talking about is different. Take a corner, where the maximum speed a vehicle could carry through is 60mph. Which would be better if you entered that corner at 61mph, DSC or torque vectoring?

neither, you'd be unable to make the turn in either case!


As DSC can actuate the wheel brakes it can generate yaw authority whilst decellerating, Active torque vectoring does the same whilst accelerating (but you can never generate the same positive wheel torque as negative (brakes are VERY powerful, even shoping cars can get from 100mpg to 0 in circa 5sec, which, when attempted the other way around takes a LOT of power ;-)

The significant advantage for an active torque vectoring system on a road car (and less so on a race car etc) is that you can adjust the yaw sensitivity of the chassis in near real time. With a conventional mechanical differenctial you have to choose a "locking" setting for positive and negative torque situations and stick with that. So at high speed you want a large negative torque (trailling throttle) lock to aid turn in stability, but now at low speed the chassis becomes difficult to turn in to slow speed corners. In positive throttle conditions the chassis dynamics team has to decide between drive and stability. A fast locking diff on positive torque will accelerate the vehicle the best, but comes at the expense of yaw stability (especially on ultra low Mu, cambered surfaces (i.e Snow/Ice covered B road etc).

As no manufacturer can control who drives their cars, and how skillfull those drivers might be, generally all road cars are set up biased for stability as a primary aim.


Now with an active system suddenly you can decide on a dynamic level of yaw control (be that damping or deliberatly peturbing the chassis on positive torque). With the inputs from Handwheel angle sensor and a yaw sensor, the control system now knows if the car is going where the driver wants it too. if it is then that's great, the system can aim for maximum traction as it's primary goal. But if it isn't then the system can return to the "stability" first calibration.


The downside of course for expert drivers is that in effect there is no passive setting of the yaw authority, it changes with conditions. For drivers skillfull enough to be able to control (and deliberately leverage) non linear control inputs, having a chassis that changes under them can be disconcerting.

For example, in a fully 3 diff active WRC car, you must not use any significant countersteering lock ! (in effect the system now thinks you want to turn left in the middle of a big drift in a rhd corner, which is bad ;-) Except, when you are beyond the yaw authority of the system you have to use countersteering to augment the systems capability to return the front of the car to, er, infront, of the back (if you see what i mean).
Hence you have a car where initally you deliberately don't react as a driver to yaw instablility, then, at some point depending on a million factors such as road surface, camber, friction, speed/momentum, etc etc, you must now react.

Learning that point is difficult, and why a lot of privateer drivers who got into full fat WRC machinery went a lot slower initally untill they had adapted to those changing responses

Edited by anonymous-user on Saturday 25th February 11:42

[peteO]

Dagnut said:

3) Do sheep get really heavy when it rains?

ha ha ha ha ha ha ha... that actually made me 'lol' as the kids say!.. or even 'rofl'

[Kawasicki]

Max_Torque said:

Kawasicki said:

What I am talking about is different. Take a corner, where the maximum speed a vehicle could carry through is 60mph. Which would be better if you entered that corner at 61mph, DSC or torque vectoring?

neither, you'd be unable to make the turn in either case!

If calibrated correctly a stability control system would recognize in the corner entry phase (before the required peak lat acc was reached)that the vehicle was nearing the absolute limit. If the system brakes, then it is conceivable that the vehicle could slow to 60mph before it slides off the corner. Adding to the speed however is a disaster, so using torque vectoring alone in my opinion is not logical.

Now, give me a few minutes to read the rest of your post!

[anonymous-user]

Kawasicki said:

Max_Torque said:

Kawasicki said:

What I am talking about is different. Take a corner, where the maximum speed a vehicle could carry through is 60mph. Which would be better if you entered that corner at 61mph, DSC or torque vectoring?

neither, you'd be unable to make the turn in either case!

If calibrated correctly a stability control system would recognize in the corner entry phase (before the required peak lat acc was reached)that the vehicle was nearing the absolute limit. If the system brakes, then it is conceivable that the vehicle could slow to 60mph before it slides off the corner. Adding to the speed however is a disaster, so using torque vectoring alone in my opinion is not logical.

Now, give me a few minutes to read the rest of your post!

But you said the maximum speed the car could carry through is 60mph, so 61 = crash ! No DSC system can break the laws of physics. If you mean't that 60mph was the maximum apex speed, and the road was wide enough to allow the car to scrub off that extra 1mph without hitting anything then yes, DSC is the one to have ;-)

Currently of course the systems are only using Handwheel angle as the driver directional demand, so it is still up to the driver to do the correct thing (and why so many people just drive straight into the thing they hit as in the "heat of the moment" they forget to steer.........)

The next-big-thing in active chassis dynamics in undoubtedly "holostic trajectory planning" where a large number of sensors, some operating outside of human capacity (think IR sensors to see at night, or even roads embedded with corner ID/info tags that relay the severity and surface conditions of the next bend to the oncomming car etc) all add information to the dynamics controller so it can preempt the future !!


In conbination with peer-to-peer inter-car networking it's an interesting time to be in vehicle development ;-)

[Kawasicki]

Max_Torque said:

Kawasicki said:

What I am talking about is different. Take a corner, where the maximum speed a vehicle could carry through is 60mph. Which would be better if you entered that corner at 61mph, DSC or torque vectoring?

For example, in a fully 3 diff active WRC car, you must not use any significant countersteering lock ! (in effect the system now thinks you want to turn left in the middle of a big drift in a rhd corner, which is bad ;-) Except, when you are beyond the yaw authority of the system you have to use countersteering to augment the systems capability to return the front of the car to, er, infront, of the back (if you see what i mean). Hence you have a car where initally you deliberately don't react as a driver to yaw instablility, then, at some point depending on a million factors such as road surface, camber, friction, speed/momentum, etc etc, you must not react.

Learning that point is difficult, and why a lot of privateer drivers who got into full fat WRC machinery went a lot slower initally untill they had adapted to those changing responses

Thanks for the info. There are obviously many parallels with DSC systems, as both systems are comparing driver requests to estimated path. The opposite lock comment is also very familiar as that is also used by DSC to fire up oversteer control which really kills speed. That's sort of my point, I can only imagine how much I would brick it if I got a shot of torque propelling me to the right in a long fast left hand bend.

[Kawasicki]

Max_Torque said:

Kawasicki said:

Max_Torque said:

Kawasicki said:

What I am talking about is different. Take a corner, where the maximum speed a vehicle could carry through is 60mph. Which would be better if you entered that corner at 61mph, DSC or torque vectoring?

neither, you'd be unable to make the turn in either case!

If calibrated correctly a stability control system would recognize in the corner entry phase (before the required peak lat acc was reached)that the vehicle was nearing the absolute limit. If the system brakes, then it is conceivable that the vehicle could slow to 60mph before it slides off the corner. Adding to the speed however is a disaster, so using torque vectoring alone in my opinion is not logical.

Now, give me a few minutes to read the rest of your post!

But you said the maximum speed the car could carry through is 60mph, so 61 = crash ! No DSC system can break the laws of physics. If you mean't that 60mph was the maximum apex speed, and the road was wide enough to allow the car to scrub off that extra 1mph without hitting anything then yes, DSC is the one to have ;-)

Currently of course the systems are only using Handwheel angle as the driver directional demand, so it is still up to the driver to do the correct thing (and why so many people just drive straight into the thing they hit as in the "heat of the moment" they forget to steer.........)

The next-big-thing in active chassis dynamics in undoubtedly "holostic trajectory planning" where a large number of sensors, some operating outside of human capacity (think IR sensors to see at night, or even roads embedded with corner ID/info tags that relay the severity and surface conditions of the next bend to the oncomming car etc) all add information to the dynamics controller so it can preempt the future !!


In conbination with peer-to-peer inter-car networking it's an interesting time to be in vehicle development ;-)

Yes, I meant apex speed. Thanks for the discussion. Besides the tech, the future looks a little boring.

[anonymous-user]

Kawasicki said:

Yes, I meant apex speed. Thanks for the discussion. Besides the tech, the future looks a little boring.

Just as long as there is one of these:



we'll be fine ;-)

[Kozy]

Marf said:

Kozy said:

For a FWD racing car, is a helical or plate diff a better option?

Do Quaifes/Torsens/helicals work under braking?

Helicals require both wheels to be loaded in order to work, i.e. if one wheel is in the air, then no power is being transmitted to either.

Plate type diffs do not have this little foible.

That's all very well. But since plates effectively lock rather than bias the torque, surely they will tend towards under steer and also lose the torque vectoring effect that the helicals have, felt as the car steering itself into the corner?

This is outweighed by the pros on a RWD car, but on a FWD, increased understeer and steering effort must surely be detrimental. With a heavy front bias on weight distribution, is unloading the wheel an issue like it is on a RWD?

[Marf]

Kozy said:

Marf said:

Kozy said:

For a FWD racing car, is a helical or plate diff a better option?

Do Quaifes/Torsens/helicals work under braking?

Helicals require both wheels to be loaded in order to work, i.e. if one wheel is in the air, then no power is being transmitted to either.

Plate type diffs do not have this little foible.

That's all very well. But since plates effectively lock rather than bias the torque, surely they will tend towards under steer and also lose the torque vectoring effect that the helicals have, felt as the car steering itself into the corner?

I can only speak from experience of driving the same car with both a torsen and plate type diff and say that both felt pretty similar during cornering (entry, apex and exit), though the plate type could be felt locking and unlocking markedly when adjusting the throttle, whereas the torsen was smooth as butter.

[cure]

So where did the answers end up? Does anyone know? Searching has not helped me so far. Thanks!