Adam M

Hi all,

was wondering if anyone with an appreciation of chassis dynamics might know what effect changing from a viscoud coupled centre diff like on normal subarus, (or a plate type lsd centre like on type rs and some ras) to a mechanical type centre lsd such as a torsen or quaife.

I know thje centre diff on the type r lends a lot to the oversteeriness of this model.

I have to add, that I am not tinkering to change the handling, I have no option but to uprate the centre diff so have a choice of everything. Ironically, these are actually cheaper.

firefox

Most centre diffs are either plate or VC.

NEver seen a geared (Torsen/ATB) centre yet.. but there is always a first time.

Playing around with the centre will affect the lockup characteristics from front to back.

Can this can impact on turn in/out... depends if the car is used for lose surface or tarmac racing.

Have you been speaking to Mark ?

I know STI do a few options (mainly 12 or 20).. Cusco do a few too.

J.

dowser

Adam

While I was investigating the front diff I looked at options for the centre too. Quaife didn't make one, and the only alternatives I could find were the STi lsd's - two types were available, a 12kg/cm2 or 20kg/cm2 (I think kg/cm is right?!).

I chose not to in the end (but maybe should have gone for the 12....isn't hindsight wonderful) - mainly for cost and the comment from Firefox about the 20 being useful if my drive to work included a gravel stage

Let me know how you go, and if you find alternatives - having experienced the improvements from changing the front to Quaife atb, I'm keen on centre options too. Transfer of torque front/rear feels very slow now.

Richard

firefox

Richard,

Changing the front and rear with ATB's will show the biggest "gains". The changing the centre wont be as noticable.

Standard STI centres are 12.

J.

Adam M

Guys,

have to change as I am changing box, and dont want to use standard 22B box for anything.

J I have not been speaking to mark on this no.

Quaife is going in front no question, but there is going to be a torsen type centre available soon, there is no doubt of that.

Have been offered these 12 and 20 sti upgrades which i migth consider. Have accepted I am going to lost the active centre diff and dont mind. Cuscos are v expensive >£1000!. what do these ratings of 12 and 20 mean? keep hearing them but dont understand. Is it the torque difference which causes it to lock?

Car is entirely for road use.

I had thought the front rear torque distribution was mostly responsible for the over steering, and so maybe a quaife 50:50 will sort this. I dont like oversteer. Am told it locks diff under acceleration, but once you lift off, it free wheels, have mo idea what effect this will have on over/understeer but that is my main concern. That and can it handle the power?

Am at a total loss and no one can offer help, perhaps I should ask quaife?

firefox

The 12/20 is the rating of the lockup (VC fluid characteristics)

As for operation.. acceleration.. coasting...reverse... that is classed as 1 or 1.5 or 2 operation

You will see it listed with different diffs.. depends if they are clutch or gear...

Mark was asking me about centre diffs too

J.

MorayMackenzie

Adam,

The Type R cars have a clutch type rear lsd, IIRC, which, I suspect, will contribute more to the oversteer you refer too than the centre diff.

Moray

dowser

Thanks J.

But *another* £525 for the rear ATB.....ouch I'm well happy with the front though, a very good investment. And at least it'll be easier to fit, lol!

It just feels now that the transfer of power front to back is slow, rather than the rear being uncontrolled - the front is so well tied down now everything else feels sloppy.

Do you think changing the rear will clean this up leaving the centre as is (Euro 'hand-brake abused at 60mph' my00 item)? I guess until there's a gear based torque biasing device available for the centre then everything else is a compromise? What do the WRC's use (can they distribute power between all 4 wheels efficiently)?

Can't believe I'm already whinging - the front has made a big difference...and makes you realise how limited the OE stuff is.

Sorry Adam - a bit off topic.

Richard

Sigmund

This is quite interesting reading. But I think some of you on here don't fully understand what a torsen or Quaife ATB diff centre diff will do to the handling of your car.

I will try to explain my views on this issue here:

As standard the Subaru Impreza have a 50/50 torque split front/rear. With a limited slip diff / viscous coupling this torque split won't differ from that. The only thing that will happen if you get wheelspin at one axle is that the center diff will lock up when the speed difference between the axles is big enough. I think this is what some people refer to as torque transfer to the front....

When you install a torsen-type (torque-sensing) center diff on the other hand the torque split won't be costant. The diff will deliver more torque to the axle with more grip. This is the system that was used on earlier Audi Quattro cars. And it's not a good system for performance driving. You will have no control of the torque split and it might change in the middle of a fast turn. Those of you who have driven Audi Quattros on a slippery surface will know what I'm talking about.
The torsen center diff is useful in slow off-road aplications where absolute maximum traction is more important than handling.

The point is that I will strongly discourage anyone from installing a torsen type center diff in their Impreza.
I would much rather go for an uprated viscous coupler that locks up faster than the standard. This will cure the "slowness" of the transmission that some experience.

As a front or rear diff on the other hand a torsen-type diff should be a good thing. As some has already experienced.

Adam M

Thank you,

that was the kind of answer I was looking for.

it is not entirely true in my case in that the centre diff torque split on the type r is 65:35. I assume when this is fully locked, it then becomes 50:50 like a rigid propeller shaft would give. This means that the varying degrees of lock up will shift torque from 65:35 to 50:50.

But then this confuses me even more. The electronic centre diff is a electrohydraulic plate type lsd. In the full back setting it is fully open, in the full forward, it is full locked. Does this mean that when open, it doesn't act as an lsd, as it is set to open diff mode? ie. if manuallyI set the lock up to off, is it still capable of partially locking itself up when it detects slip, or does the manually selected open setting prevent that?

sorry if this confuses everyone, but I could not be more lost already.

Sigmund

I didn't know that the type R had a different torque split, but now I do. If it's 65rear/35front that's very close to the torque split on the Ford Escort Cosworth. I have been told that the more rearward torque-split is better for racing/driving on concrete/asphalt. The 50/50 might be more suited for gravel. But I don't think it makes that much difference unless in a race.

When you have adjustable lock on the center diff I would think that when it's open the center diff will work just like any other open diff and the wheels with the least grip would spin.
But when on full lock I would think that the torque split would still be 65/35. Only you will have a very limited slip. I.e. the speed of the wheels of the two axles must be nearly equal.
About this I'm not completely sure though.

The thing that makes this difficult is the realtionship between torque and rotation of the wheels. It's easy to mix things up.
I.e. fast rotation of one axle doesn't mean that it recieves a great deal of torque, as in when both wheels of the axle is off the ground.

Sigmund

If those of you that have installed front and/or rear torsen-type diffs could explain how the handling characteristics changed after installing them it might be easier for people to judge wether it's worth the money.

dowser

Thanks Sigmund

You perfectly explained what my poor brain was trying to get a handle on Seems I'll be buying the rear soon too.....

So the difference between 12 & 20kg/cm2 centres is that the 20 will lock sooner 'cos there's already more pressure to start with? Given the STi centres are available in 12 & 20, what's fitted to a Euro car?

With the Quaife ATB on the front, it feels like the front end is being sucked onto the road when you 'give gas' It considerably changes the handling characteristics. You can actually feel (through the steering wheel) the torque transferring between the wheels when cornering hard (it's v.quick). On bumpy straights the same applies - overtaking hard in 3rd highlights it quite well. On smooth roads, or if taking it easy, you don't notice it.

I haven't pushed to the limit in dry conditions yet, but I can now exit wet corners around 15-20% faster than pre-ATB diff....albeit slightly sideways

Richard

sempers

Ok, I know diddly squat about this, but am reading with interest - this link is probably old news, but may be relevant to the plate type discussion:

http://www.fortunecity.com/silverstone/lancia/58/technical_school/traction/tech_traction_4wd.htm

Particularly, the electro-hydraulic clutch types rely on there being a difference in rotational speed between the front and rear parts in normal, no-slip operation, so that locking up the clutches transfers torque.

Hope this is of help.

- Mark.

Sigmund

Good to hear that I can explain so people understand! Thanks.

I'd like to have the front Quaife ATB on my car too (Escort Cosworth). The car has limited slip front and center diffs which makes it feel like it's a bit short on traction on the front wheels. A Quaife ATB would fix that I think.

That link is really good. Have been looking for something like that for a while.

Adam M

Nice site, but I found it a bit too basic, and it didnt address what happens with the efcet of the diffs on the handling.

It does seem to suggest that the torsen is a great choice, but doesnt mention it much as a centre diff application. 50:50 torque split apparently gives the car neutral handling which I really like the idea of, although it has been pointed out to me that neutral does not me neither over nor understeering, but a mixture, the variety of which can dangerously catch you out.

It does seem to make the point that rear torque bias yields over steer and front torque bias yields understeer.

So in essence a centre torsen will give me immediate acting variable degrees of lock up withotu any change in torque distribution, btu at the same time, reduce the tendancy to oversteer.

It does still sound like a good prospect to me.

carl

I'm not convinced at all. ISTR Simon once mentioned on the (now defunct, I presume) drivingtechniques website that the diff control on a Type R/RA does not alter the front/rear torque bias, only the amount of lockup on the centre diff. I presume the torque bias is simply a function of the final drive ratios on the front and rear diffs.

However, what I don't understand is this. With a viscous coupling, consider a situation where the 'rear' propshaft is rotating faster than the 'front' one. The differences in rotational velocity will cause the viscous coupling to start to lock (actually the fluid will start to drag the slower-moving shaft around to match the faster-moving one). But it takes torque to accelerate the shaft, so surely this is a redistribution of torque, i.e. altering the torque bias? Hence locking up the centre diff does alter the front/rear torque bias? The same would apply if you had a Torsen diff in the middle.

Also, if you changed to a 50/50 torque split with a viscous coupling wouldn't you essentially have the drivetrain from a UK Impreza?

Edited to say I think this is an appropriate Torsen centre -- "The T-3 is currently offered in a Twin Differential configuration which has an open differential surrounded by a Torsen differential! This versatile unit is ideally suited for center applications and is designed for those hard to package all wheel drive systems. The design of the Type C allows for a torque split other than 50:50 with a majority of available torque going to either the front or rear axle. "

[Edited by carl - 2/22/2002 5:16:32 PM]

Blow Dog

geeks.

carl

I guess if wanting to understand how things work makes me a geek, then that's what I am.

Still, it was good enough for Dick Feynman.

Sigmund

Adam ; you're right about that page beeing at bit to basic for our purpose but to get to the more technincal explanations of how a torsen-type diff works we need lots of mathematics and equations.
Still it could have delved deeper into the handling issues.

But a center torsen diff WILL have a variable torque split. That's the whole point of it as it is torque-sensing. More torque to the wheel or axle which has more grip.

Regarding the non-variable torque split of either front or rear biased, or neutral (50/50). I agree with your ideas of rearward bias giving more of a tendency to oversteer etc.
But I think suspension setup and weight distribution has a lot more influence on handling than the torque-split.

Carl: On a car with a limited slip center diff: When the rear propshaft rotates faster than the front propshaft it will do so for a reason. The reason might be that the rear wheels have less grip than the front wheels and as the torque split is invariable this will lead to wheelspin as the tires can not provide as much traction as the tires at the front. So the torque split is not altered.

Hope this has made things a bit clearer.
If not somebody will surely buy a torsen-type center diff and experience the strange handling traits this will bring along.
I think torsen front and rear diffs + a uprated center viscous coupler will be the best for improved handling.
I can't see what's wrong with the adjustable center diff on the Type R/RA. It might not be possible to install in other imprezas though. It will give you exactly the lock-up you want.

Blow Dog: I think you should keep comments like that to yourself. This is a great thread. It's exactly what I love about the scoobynet.bbs.

Cosie Convert

I recently bought a gearbox which had a plate type front LSD in it, I thought that maybe it would help with the handling, so I fitted it.
I'm not convinced that it has helped I seem to suffer even more power understeer than I did with the open diff. I have since read that plate type fronts are really only any use on loose surfaces, is this true ? Or do I just need to drive it harder and induce some oversteer ?
Would it be worthwhile setting the suspension so the car is a bit more oversteering as standard and use the front diff for stability under power ? Or should I just take it out ?

cc

carl

I wonder if there's a good reason for using viscous couplings and plate-type centre diffs. Both would seem to allow a great deal of slip where necessary (i.e. they're based on friction), whereas it seems to me that any sort of constant mesh gears arrangement would allow less slip. Maybe it's more important to be able to have more front-to-rear slip than it is side-to-side slip?

(PS: What's the point of understanding quantum physics if you can't understand how a simple diff works? )

[Edited by carl - 2/22/2002 10:48:54 PM]

Blow Dog

Was a joke. Don't let it kill you.

Adam M

Sigmund, carl, bring on the maths. I would love a chance to get my mind around some really heavy stuff. Havent had any real difficult stuff since university!

I think sdb is wrong about the adjustable centre diffs. He has to be.

They are 65:35 when open and they have to be 50:50 when closed because they are essentially a rigid shaft when completely locked. IN which case the degree of locking must therefore adjust the torque split.

so if both type r and uk car have plate type lsds which when fully open only differ in terms of their torque split, then all I need is a UK or sti uprated centre diff to give me safe handling that I know will work.

So can someone help me understand the whole 20 or 12 Nm uprated centre diff. Are these torque ratings purely for the purpose of defining which conditions the diff is best suited to?

I would therefore assume it is nothing to do with the torque handling capability.

Sigmund

Although the diff on the type R/RA makes for a close to stiff axle (or at least it will feel that way since there is no slip between the front and rear)I don't think that necesaryily means that the torque split is 50/50. As I have said before rotation and torque is not the same thing.

I have a real good pdf document about torsen diffs. How shall I bring it on here ?

carl

Again I disagree with you Adam.

I thought as you do, that with the shafts locked the torque split must be 50:50. But then consider the front and rear diffs. If they have different ratios then you will achieve a 65:35 torque split with the input shafts rotating at the same speed, but the output shafts on the front one rotating faster than those on the back (or is it the other way round -- I can never wor it out). If it was as you say, in normal operation there would always be slip on the centre diff (as one of the shafts would always be rotating quicker than the other) which can't be right. In normal operation (i.e. going in a straight line) it *must* be that all the shafts are rotating at the same speed.

Sigmund

You're onto something there Carl.
It can't be the gear ratios that makes up the torque-split. It must be some other arrangement which I'll admit I don't know about.
It could be something like a torsen diff with worm wheels and gears and stuff that makes the torque split..

Cosie Convert

Depending on the transfer gears in the back of the box (std UK 1:1)you could have different front and rear diff ratios, different shaft speeds and a non 50:50 split although the centre diff would still be running without slip.
The take off point from the centre diff also dictates the torque split. On the escort cossie box the rear shaft drives from the outer annular gear and the front shaft from the planetary gear assembly. The sun gear in the middle is the input from the engine.
The relative radius of the gears from the sun gear dictates the torque split. The viscous clutch (lsd) is between the annular ring and planetary gear carrier.

Sigmund

Cosie Convert:

That was a rather good explanation, and I understand the main idea of how the torque split is achieved now.
But I wish that you could make some sort of drawing of the system so that it might be a bit easier to understand.
Both the Escort Cossie system and the Impreza system(s): both the 50/50 and the 65/35 split ones.
You could just draw it on paperand scan it or you could draw it in paint or something. very crude drawings would be ok. I use AutoCad at work so I could make a nicer drawing in that and post it on here.
Please ?

carl

I must admit, I wasn't aware it was a sun and planets arrangement -- I had assumed the viscous coupling was just a pair of drive plates separated by a viscous liquid. I think drawings are definitely required -- it might be easier if there was a cut-away drawing on the web that could be annotated.