I can't say for sure that the 65/35 Subaru centre diff is planetary but the Escort is. I'll sketch the setup at work Monday and scan/post it.
I did see a proper drawing for this somewhere once.......if only I had a better memory

cc

 

My understanding...

As per Sigmunds 1st post - a torsen centre diff. would make the car undriveable, with torque constantly changing between front and rear wheels. So you'd get constantly variable torque split - in theory you could never have only one axle spinning, only 4 wheels drifts.

Having experience of the front torsen, I don't think I'd like this controlling front/rear bias.

The 12 & 20kg/cm2 centres relate to the amount of pressure preloaded in the viscous coupling. Lets say that for a given scenario that 30kg/cm2 is required to lock the diff - the 12 centre needs a speed difference between the axles that will generate 18kg/cm within the plates/viscous fluid. This is more than the 20kg/cm. So the 20kg/cm unti locks with a lower difference of speed between the axles, basically.

Or am I talking out of my bottom?

Richard
BD - bog-off!

 

guys,

thye impreza system does allow for complete locked centre diff as the fron and rear diff ratios are the same.

I know this because I am in the process of changing them both.

I still see that if the centre diff (being palte type) is locked, then the input and output shaft dont just rotate at teh same speed, they are actually locked together by an outer lock up clutch, similar to that on an automatic gearbox. Effectively there is no centre diff. In which case, torque aside, the front and rear wheels, in a straight line at least, have to travel at the same speed.

I am not so sure that a centre torsen is as bad as people are describing here. The audi quattro had one and didnt seem to be so bad.

I do know that tehre is a serious gap in my knowledge on this, and that I am currently tempted to stay with what I know, and fit the same centre diff, in my new gearbox.

 

Adam,

I have the EC Certificate of Conformity for my (UK Spec) Impreza. It says 'final drive ratio' -- axle 1: 3.9, axle 2: 3.545 (it may be the other way round, but it's the two ratios referred to in the SIDC FAQ here and here. I agree that with the centre diff locked it is the same as a solid shaft connecting the front and rear diffs, but with a different final drive ratio although the wheels will rotate at the same speed (obviously) I assume the excess is taken up by rotation of the outside bit of the diff (i.e. the square housing bit if you make one out of lego):

So for a single turn of the (locked) propshaft, the wheels will want to rotate 3.9 times on one axle and 3.545 times on the other. I think this means that all four wheels turn 3.545 times, but on one diff the axles rotate at the same speed as the square housing bit and on the other the excess is taken up by rotation of the housing. However, the SIDC FAQ says the final drive ratio is 3.9:1 so I can't work that bit out.

[As an aside, a search on Yahoo! for 'lego differential' returned a page with a design for a Lego LSD ]

 

Again, I'm not sure I'm knowledgeable enough to contribute...

But... as I understand it, the whole point in the clutch diff type is that the shafts out of the centre LSD do rotate at different speeds (irrespective of what the final drives in either end are doing), otherwise there is no point in engaging/locking the clutches - it would have no effect on torque unless one set of wheels was already spinning (or moving at a different speed to the other. See point 1...)

Or did I missunderstand the comment?

Anyway, I believe this is why you must not lock a scooby adjustable diff in normal conditions - it results in lots of smoke, and a knackered drivetrain. Full lock up is only used in very slippery circumstances where traction is paramount - eg ice / snow, getting out of deep mud.

The standard UK diff is essentially a normal, open diff, with a viscous coupler between the two sides - such that it operates like a normal diff until slip happens, then the coupler resists the difference between the shaft speeds.

I would have thought (someone elses's comment) that adding a front LSD would increase the feeling of understeer - you are now directing torque to the inside front which would not have been there otherwise....

 

Sempers

The torsen at the front doesn't increase understeer, or the feel of it. It doesn't get rid of it either, to be honest. Just makes best use of available traction.

Slow in fast out will reliably give oversteer in the wet now - before you had to time it just right (ie; know the road ).

Basically, as long as you can enter the corner without understeering on no throttle, there's no way putting your foot down will induce it anymore. I'm starting to play around on constant throttle corners now to see what happens at break-away point - initial response suggests a dab of throttle will bring the front back in line, but need to play more. I like it

Richard

 

LOL @ carcl, thanks for the description of how the diff works.

Except that at constant speed, in a straightline, the box bit in the lego will be spinning all the time, at the bevel gears should be fast in rotation with it. Only when there is a speed difference between your front and rear final drives will the bevel gears be rotating relative to the housing.

BTW. in your car your front/rear final drives may differ, but in mine they do not. Weel not to my knowledge anyway.

The crown wheel and pinion on both the diffs which connect directly into the centre diff output shafts, have the same number of teeth.

I can only presume any difference in final drive would come fromt he drop gears on the centre diff, but to my knowledge these are all 1:1 ratio.

I know for a fact that the centre lockup clutch makes no difference in a straigth line and doesnot make any noises whatsoever when locked shut on a straight road.

If there were a front/rear final drive difference then surely it would appear in such conditions and make a lot of noise whilst creating vast bills for me at the same time.

But it doesnt happen.

This can only suggest two things.

1. The diff doesnt fully lock and the final drives can be different.

2. The diff does fully lock and the final drives are the same.

 

OK, let's go for (2). In which case I think we're decided that in your car the torque bias is 65:35 rear:front with the diff control wound all the way back (open diff) and 50:50 with it wound all the way forward (locked). It must be like this because you have the same final drive ratios on the front and the rear. You want to achieve a 50:50 torque split without the centre diff locked -- it sounds like you need something akin to a viscous coupling. A viscous coupling's normal state must be with both the shafts rotating at the same speed, otherwise the fluid would start to get hot and there would be loads of transmission loss. In which case it appears that a viscous coupling works the other way round to a Type-R/RA plate-type diff, in that it's normal state is fully locked whereas a plate-type's normal state is open?

Back to the original question -- which of these states does a helical-gear (Torsen/Quaife ATB) diff most closely represent?

 

btw, I have just found out that your car has a final drive ratio of 3.9 fron and rear.


It is only teh rear axle that rotates 3.54 time per turn of the propeller shaft, but there is a transfer box on the back of the gearbox with a 1.1:1 ratio, so that both sides of the centre diff rotate at the same speed!

 

What idiot designed that then?

BTW I think I'm going to settle this argument by buying a Lego Technic diff kit and building a car with three diffs. However, I've got a suspicion they don't do a Lego Torsen.

 

Guys,

to avoid confusion, there is a 1.1:1 reduction in the transfer casing of a UK Spec Impreza Turbo 2000 (rear output from centre diff) and that means that the actual final drive of the second axle is the same as the first.... 1.1*3.545 is 3.9 (well, nearly, but 3.545 is actually 39:11 which is 3.5454545454). If there was such a HUGE difference in final drives front and rear then the centre diff would boil the oil in a couple of miles, if that... it would just blow up.

Also, the torque split cannot be anything other than 50:50 with the type of diff fitted (it is a normal open diff with a viscous coupling). In order to alter the torque bias one would have to somwhow make it "preferable" for one shaft to take more torque than the other; this can be easily achieved with a planetary type diff as used in the Type R. Additionally the Type R has a viscous coupling between the two outputs on the centre diff, which is surrounded by a magnetic field coil. When the field coil is energised it causes the plates in the VC to move closer together, increasing the coupling coefficient until finally it locks (then the torque split will be 50:50 as it has no choice but to be).

This torque plit is theoretical; in other words it's what would happen if the car were accelerated along a perfectly smooth strip of high-grip tarmac, where it is impossible for a wheel to spin, therefore it puts down any and all torque it receives from the diff. In the real world, if a wheel lifts off the ground, it cannot put down any torque so the torque split at that point would be 100:0.... this is a dynamic torque split; please understand that the wheels will always take as much torque as they are given by the diff, until such a point as they start to slip. At that particular point, the different types of diff will have make massive difference.... for example a VC diff NEEDS slip on one wheel to begin to lock the wheels together.... a TORSEN on the other hand can tranfer torque to the wheel with most grip even before the other one starts to slip....

Transmission dynamics a FAR more complex in real life than the simple "50:50" or "34:66" figures that are often discussed. Like I said, it only ever splits the torque in the prescribed fashion in perfect cnditions; in others the balance shifts, and it's the way that the diffs control this shift which determines whether or not the car is still effectively using the remaining grip that it has.

Hope this helps,

Pat.

 

Thanks for the explanation, pat. But can you go a bit further into this bit:How is this 'easily achieved'?

 

Blowdog has a car that does all the geeky stuff for him, so he can stick to doing whatever it is that he does.

 

Lego Torsen diff:
http://www.ecf.utoronto.ca/~stehlik/torsen.html

Pat:
I think your idea of 100/0 torque split is wrong for some cases:
When we have a 4wd system with one axle with zero traction and one axle with good traction then with these different diffs:
Open planetary diff (normal diff): Torque split is 50/50 as it always is in this case. If you got no grip on one axle then you have no torque going there and hence no torque to the other axle.
Normal diff with viscous coupling: After a speed difference between the two axles has built up you will have the diff lock up and give equal speed on both axles but since there is no grip on one axle only the axle with grip can hold torque... might be close to a 0/100 split here.
A torsen center diff will spread torque in a ratio up to 1:6 or something which means that you must have some traction on both axles to get any torque to the wheels at all, and the max difference would be 83/17 (with a torsen diff with 1:6 ratio, could be 1:4 and others too).

After this explanation I'm not too sure myself either (about how bad a torsen center diff is), but theory is one thing and driving is something completely different. I would still have a viscous coupling....

 

.

You bet your life he can.

 

Carl,

by "easily achieved" I mean that you can just buy a Type R centre diff and install it. If you want electronic control that will just take a little longer, but you don't "need" it because you will get the 66:34 split anyway....

Also, a planetary configuration is not extremely complex, so it's "easy" to manufacture, and it works quite well. Because of the differing radii, there will always be more torque to one ouput than the other (if this were not the case then the shafts would turn at differing speeds).

Sigmund,

The 100:0 was an extreme example to illustrate that it IS possible that a diff that we would normally call a 50:50 does not necessarily give a 50:50 distribution all the time. It does not mean that 100:0 is likely to happen very often, just that it *can* happen.

The normal open diff is NOT a planetary diff, only the Type R (and 22B and some RAs etc) have a planetary centre diff. Planetary diffs do not normally have a 50:50 split, see above, differing radii cause different torque distribution when both outputs are in sync... I agree that with one set of wheels spinning there would be no torque to either were it not for the VC... lucky that Subaru remembered to put one in, eh? But, since it IS there (in all standard Subaru centre diffs) you would get 100:0 split, all torque would go (by means of the VC) to the wheels with grip.

The trouble with a VC is that there MUST be a speed difference between the outputs in order for it to "lock", a clutch type unit does not require this, nor does a TORSEN.

Agreed on the TORSEN's requirement for torque at one wheel. Quick tip.... if you get stuck somewhere and have no grip on one wheel (say it's in the air) then apply the brake lightly. This will give the wheel in the air "grip" (actually, it just requires torque to turn it, but that's all that counts). This way, with (say) 4:1 range TORSEN you can still get 3x(braking effort) through the wheel with grip.

Personally, I'de prefer to have an active transmission... that way the electronics can control the locking of the diffs using strain gauges, rotation sensors (eg ABS sensors), steering angle sensor etc. Hmmm. Food for thought

Cheers,

Pat.

 

OK, now I understand the 66/33 torque split from the planetary gears -- I think it was mentioned earlier in relation to the Cosworth diff. The drive for one shaft will be taken from the sun and the other from either the planets or annulus.

Not so sure I would want fully active electronic diffs -- mid-corner how would you (the driver) be able to predict its response and drive accordingly? I can just imagine halfway round a gravelly bend wondering which end's going to go first