LSD discussion - round 2

[CamB]

As a follow-on to this discussion on whether Quaife or ZF type diffs are better, I am looking for further opinions.

After an eye-opening ride in a Triumph TR7 with 300hp (and only about a 2m wheelbase...), I watched as the driver steered the car using the brakes and throttle as much or more than the steering wheel. I can't really envisage myself being this good, by the way.

He said the diff was 80%+ lockup, because he didn't like the diff slipping when it was least wanted. So this leads to my question.

Why would you want a diff with a lower lockup? I sorta see that you would end up with all the disadvantages (eg a car that wanted to push) without the advantages (steer it on the throttle). A 25% lockup sounds like it would be almost useless...

Help me learn (the BMW will get its lsd one day).

Cam

[Matt Smith]

My Beemer has 25% lockup Cam!
It didn't feel too useless the other week when my accellerator pedal mutated into the steering wheel around the infamous "Domain Hill Climb" course....

[CamB]

BBSers - does that sound like Matt wants to lend me the 325i for a vigorous weekend of LSD research (the legal kind)???

[Bill Douglas]

I would take that as "Matt talk" for lets cut lose on the track...

[A Quiet Boom]

The advantages of a lesser lockup diff would be as follows, less push in the corners (especially sharp ones), better fuel economy, less tire wear, better steering input and less stress on the driveline. The main disadvantage of course being less traction. On courses with shallow bends and high HP cars 100% lockup (spools) are often used as they are in drag cars, I believe the 935 even used a spool on some tracks, however spools are just about useless for anything else. Percent lockup is typically related to the spring pressure on the clutch packs in standard limited slip diffs. To my knowledge the best type of diff is the torsen because it is actually torque sensitive and provides power to the wheel with the most traction. Locking type diffs (not limited slip) i.e. "Detroit Locker" are often used in NASCAR and other types of racing but can be really scary when the diff goes in and out of lock during a corner. This leaves us with two obvious choices that I'm aware of, limited slip or torsen, the torsen like I said is probably the best but it's costly and not availble for all cars. Sometimes limited slip diffs can be shimmed for more lockup or have a clutch pack removed from each side for less lockup but it really depends on the manufacturer, I've also seen some Ford racers add clutches for more lockup with the factory Traction-Lok diff. Also remember that generally speaking independent suspension needs less lockup than a live axle since it's not subject to twisting in the chassis from the engine torque.

Cam, I'm pretty good mechanically and have a solid understanding of driveline improvements if you could provide me with links to the Quaiffe and ZF diffs I'll read up on them and give you my opinion.

[Jack Olsen]

I have 80% lock on my (Guard, Quaiffe-type) differential. The disadvantage is in very tight turns, where the lockup can produce massive understeer.

The nice thing is your ability to steer, correct, and save the car with the throttle. It becomes a very nice tool to work with.

(Harder to spin the tires when you want to show off, though.)

[CamB]

Thanks for the answers guys.

To foster further discussion, lets keep going on this. First up, useful links are:

www.quaife.co.uk
www.quaifeamerica.com
http://www.pelicanparts.com/catalog/SuperCat/911M/POR_911M_SUSAXL_pg3.htm#item8
http://www.nosubstitute.org/articles/LSD_Velocity_Feb01/

Secondly, Jack - I thought the Quaife style (torson) was infinitely variable - I didn't realise they got to a certain level then became unlocked as well. Also, I thought they were pretty effective at ruling out push/understeer. (sigh) I get more confused.

Back to the basics of my knowledge. I understand that if a diff is labelled 25% slip, it is a reference to a percentage of peak torque before it slips. Thus, it appears that a 25% diff in one car may not be the same as in another, and further (and more importantly) may not be the same following modifications. For instance, it appears to me that a 25% diff in an early car might be pretty easily overwhelmed by a big engine transplant.

Moreover, I have started thinking about what it means for a diff to be at a certain lockup and for that lockup to be surpassed. I initially believed (ie my first post above) that if you exceed the torque level implied by the lock-up that the diff would go completely open. However on reflection, it seems to me more likely that it is still transmitting this maximum torque to the inside wheel. Therefore, a 25% diff is useful, as is a 50% or 80%.

I can understand the negative with a Quaife in a throttle on/off situation with the diff going from closed to open.

So, can anyone confirm or deny any of this rambling...

[A Quiet Boom]

You're mostly correct, I don't think power levels really change the lockup of a limited slip much though except that higher power will find the wheel with less traction sooner and therefore seem to overpower the diff. You are right though about "lockup" a limited slip never goes completely open, it just slips against the clutch pressure until traction is regained or the wheels start turning at the same speed again. As for the Quaiffe torsen diff going in and out with the throttle I'll need to look into that a bit more, locker type diffs are very bad at this but I thought the torsen was much more refined. It is however sensing torque so I suppose it's possible that negative torque (off throttle, engine braking) could make it "unlock". I also don't see how a torsen type diff could be rated by percentage of lockup, standard lockers use dogtooth faced discs that are engaged to one another by spring pressure and torque which works in a cam-like manner on the diff center shaft to force the teeth together, in an of throttle situation the force comes and goes and the teeth "jump" which can be unsettling mid corner. Stiffer springs are used to offset this problem requiring a greater torque difference between the wheels to unlock the diff. Since the Quaiffe/torsen appears to be a completely geared setup I don't see how it could "slip" one wheel, however the gear sizes could be changed to allow different variances in wheel speed i.e. 80% would allow one wheel to turn at 80% the speed of the other and vice-versa. Perhaps Jack can fill in the details here since he actually owns one. I know I'm certainly interested in learning more about Quaiffe's setup.

[CamB]

Thanks - I feel much better now.

The part that is really killing me is that a nice second-hand BMW e21 diff is around US$200 or so, and to convert from 25% lockup to 50% is quite doable and not that expensive.

However, the US$70 freight to NZ and the concerns over whether it would ever arrive are worrying...

[jluetjen]

Cam et al;
I think that you need to keep in mind that different diff's do different things in different situations. What the "Best" diff will be is the one that does what you want it to in the situations which most concern you. It's a mistake to think that one type is bad while another is good. It's strictly a "horses for courses" type discussion.

Differential Modes to be considered:
* What does the diff do when accelerating? How fast does it do it? How hard does it do it?
* What does the diff do when deaccelerating? How fast does it do it? How hard does it do it?

Some examples (based on a 911, FWD cars will react somewhat differently). Keep in mind there are 4 classes of diff's: open, limited slip, torque sensing or biasing and locked (which is actually no longer a differential!)

Spool (solid diff): On acceleration the extra traction from the inside rear wheel tends to push the car into an understeering mode (pun intended). Obviously there is no transistion. On deacceleration, the outside rear wheel is normally the first to lose traction. With a spool the extra grip at the outside rear will tend to pull the car straight. Since the two rear wheels are locked together it is also harder to flatspot a rear tire with a locked rear end. 935's used them because it is very difficult to light up both rear wheels which was a problem with those high HP cars. Oval track cars use them because they have the car set-up with stagger (a larger rear wheel on the outside at the back) which helps the car through the turns. A good analogy is a styrofoam dixxy cup rolling on a table. One downside is that cars with spools are absolutely impossible to meneuver at low speeds or in the paddock area without tire dollys.

An open diff: This is hands down the best solution for a street car. As long as both rear wheels are solidly on the ground, it will adjust according to each wheels speed. It runs into problems if one wheel (Usually the inside wheel) starts to spin under acceleration since it will overapply torque to that wheel based on its speed. On deaccelleration it allows each rear wheel to do its own thing.

-- Limited Slip Differentials --
Salsbury, Plate or Clutchpack differentials: This is the OEM style diff for a 911. The amount that the diff will lock is based on four things: the cam angle, the spring pressure on the clutch pack, the clutch material and the order in which the clutches are put together. There are two ramp angles, one for acceleration and one for over-run (aka: coast). They are usually listed as numbers in tha order, like 60/30. The number represent the ramp angles and vary from 90 degrees (no locking affect) down to about 30 degrees (Very strong locking affect). BTW, no one uses a 90 degree ramp because they chip, the highest number that you will usually see is 85. A 60/30 diff will have moderate locking on acceleration (so as to not induce too much understeer) and a strong locking affect on overrun (to help keep the car straight under braking).

The spring pressure on the clutches (sometimes called preload) will affect how quickly the diff locks in both directions. A loose pressure (say what Porsche describes as 25% diff) will be fairly gentle when locking. A high pressure (an 80% Porsche diff) will tend to lock more abruptly and if the tension is high enough, act almost like a locked diff.

The clutch material affects the lock-up in a similar fashion to the spring pressure based on how grippy it is. The arrangement of the clutches will also affect how abruptly the clutch locks. If the clutches are arranged so that there is only 1 rubbing surface, there will be comparatively little locking affect. If the pack is arranged so that there are more rubbing faces, the locking affect will be greater.

-- Other types of limited slips --

The Cam & Pawl (aka: ZF style) and the Viscous differential (popular in Touring cars and sometimes F1 or CART). You'll rarely see these on 911's for various reasons.

-- Torque Sensing and Self Locking Diff's --

Weisman, Torsen and Quaife's. These will actually allocate more torque to the wheel which can accept more torque. I'm not going to discuss the Weismans since they are not that common in teh Porsche world. Each design has some slightly different features. Torsens and Quaifes are infinitely variable gear driven diff's that take advantage of the fact that worm gears can efficiently transmit torque in one direction only -- in the other direction they generate huge amounts of friction. They work real well allocating torque up to the point when one side loses traction. Once one side loses traction, these diff's will just spin that wheel. One feature of these diff's is that the function in one direction only, usually under acceleration. Under deacceleration they function as open diff's. Their affect tends to be instantanious. As soon as you start to apply torque to a Torque biasing diff, you will start to generate an understeering affect as a result of the increased torque available to the inside wheel.

Finally the "Detroit Locker". A Locker is a pretty blunt instrument. It locks solid when under accelleration and is open on over run. During long corners Lockers "ratchet" and lock/unlock repeatedly which can be disconcerting until you get used to it.

* 1st Hand Examples

I've raced cars with open, locked and Detroit Locker diff's and have come to the following conclusions:

Locked Diff's: A lot of us have used these on karts. They tend to push and hop in slow turns until you start to pick up speed and generate slip angles. Once I'm up to speed, I generally don't give it a lot of thought. They tend to be predictable and go where they're pointed. They're also very solid when braking. A nice feature for endurance racing is the limp-home feature if one of the rear axles brake (assuming that it doesn't pitch you into a wall first!)

Open Diff's: Surprisingly, most road race cars use open diff's since the SCCA requires them in FF, FC, FV, S2 and SR. If the car has a low center of gravity and a wide track or a lot of downforce, it really isn't a big deal since the wheels are rarely off of the ground. In a sedan or production car (including 911's), this is not the case and it is not uncommon to spin the rear wheel in hairpins. While hardly a fatal flaw, it is slow.

A Detroit Locker: I once raced one in an ITC car. It does feel a little funny when it ratchets around long corners (like the banking at NHIS) because the car feels like it's "walking" or alternating between slight understeer and slight oversteer. In the case of the hairpin, I found that it was difficult to balance the throttle since the ratcheting affect would get in the way. The best approach seemed to be total commitment - either off the throttle or full on and steer like crazy.

Any sort of diff except for an open diff will tend to induce understeer. This is for two reasons:

1) Increased thrust from the inside (lightly loaded) wheel will tend to push the car straight.
2) The increased forward thrust on the car will tend to increase squat and induce front end lift. When the front end comes up, it will lose traction as the tires become unweighed. This is good in a car which has oversteer dialed in, but bad for a car which understeers to begin with. As a result you may need to change your suspension set-up after you have replaced an open diff with something else.

Conclusions:

Which is "best" will depend on the car, the course and the rules. For track racing, I'd use a clutch style since other ajustable styles (like a viscous or combinations) are beyond my budget. I could play around with the clutch preload and stack-up to get the amount of locking affect desired. If the car had a fairly high power/weight ratio I might just go for a spool since they're cheap and fairly fool proof. My pit crew would hate me because they really are a drag to move around in the paddock. For soloing I'd consider a Quaife since they tend to be one-size fits all.

None of them are perfect solutions. In the professional world, the trick set-up is to combine 2 styles (usually a clutch type and a viscous coupling type) into one diff. This is neat since it can be ajusted on the fly by varying the pressure on the clutch packs and playing with the viscous coupling. This is how F1 drivers can adjust their diff settings on the fly. But a lot of them still use spools or normal clutch-style diff's.

A final point to consider is that Quaifes generate all of their heat (read absorb HP) when they are doing the torque biasing thing. Limited slips on the other hand generate less heat when they are locked up then when they are open. A good reference is Carroll Smith's "Drive to Win", specifically chapter 19: The Racing Differential. To quote Smith (Page 7-36):

"So which is best? Probably the viscous coupling. For those of us who do not have access to a developed viscous, as usual it's different horses for different courses. In the current environment, given a limited budget and a reasonable amount of HP, a steel cased Torsen would be my choice, simply because the unit is so effective, so benign in operation and so reliable - one less variable. Given a testing budget, I would do a back to back comparison between the Emco, the Xtrac Viscous and either the Xtrac, Salisbury or the Hewland Powr Flow.

I would not be at all surprised to find that I ended up running different diff's for different types of courses. For sure I would be tailoring the Salisbury (clutch type) or Powr Flow to the type of circuit. With a severely limited budget and/or an inexperienced driver, I would probably elect to use a cam and pawl - no surprises. As always with any question having to do with vehicle dynamics, I reserve the right to change my mind without notice"


.

[jluetjen]

A footnote: According to Carroll Smith: Apparently Dr Ferdinand Porsche was the first to arrive at a practical solution to the problem of wheelspin by fitting the 1934 Auto Union Grand Prix car with a ZF cam & pawl limited slip differential - which biases the torque towards the tire with the greatest traction.

[CamB]

Wow - great writeup. Since I am budget constrained, the BMW will either get:

a) nothing
b) 25% standard BMW diff
c) (b) + extra clucthes

Anyone coming to New Zealand in the near future who needs an extra 60lb of luggage?

[jluetjen]

Cam;
I don't know if you can do "extra" clutches, but I'm sure that you can change the arrangement of the clutches or the spring pressure.

Here's the deal with the clutches...

The clutch type diff has two clutch packs on either side of the gears, which are used to connect the case to the respective half shaft. The ramps push against the clutch packs as the diff is loaded in either direction. A common configuration is to have 6 plates (7 clutch surfaces) on either side. Half of the 6 of the plates may be attached to the inside of the case with splines, lets call these "C" plates. The other 3 plates are attached to the halfshaft via splines, lets call these "H" plates. In order to get the strongest lock-up, the plates would be arranged as follows (from inside to the outside of the pack): C-H-C-H-C-H. This will provide 7 contact surfaces. (Note that the inside surface is attached to the halfshaft and the outside surface is part of the case) Let's say that this is too stiff for you and you want to loosen it up. In that case you can arrange the plates as follows: H-H-H-C-C-C. Now there is only one contact surface between the two sets of clutches. This will have a much softer lock-up since that once surface doesn't provide as much total grip.

Alternatively, you could install stiffer springs or increase the preload. Both of these will increase the grip levels on the number of contact surfaces since the pressure on the surface(s) will be greater.

Unfortunately, with clutch-pack diff's, the clutches do wear, and so the preload will decrease over time. So they will act more and more like open diff's as they wear unless you adjust the spring pressure accordingly. How fast this degredation happens is a function of the pressures, clutch arrangment, clutch material and type of driving to which the diff is submitted.

I doubt that you will be able to get extra clutches to fit. Nor should you remove clutch plates since this will make your preload be essentially 0 unless you get some really big springs.

Once you get it installed, let us know what happens as well as the what you found to be the affects of any tuning that you do. The nice thing about a 911 is that the diff comes out pretty easily be after you remove the half-shafts. I don't know if it is as easy in a 2002.

PS: 2002's have always been a favorite of mine too, but second to a 911.

[Bill Verburg]

Nice write up John! I would only like to emphasize that the ZF type will help in braking situations the torque-sensing(Quaife) will not.

[ChrisBennet]

At some point Porsche went to a "lockup on over run" type diff. Greg Brown wrote a nice article on diffs in the Feb '01 Velocity Magazine
Here's something Greg posted to the Porshephiles mailing list in 1996:

From: GregBBRD@aol.com
To: winders@apple.com, PORSTURBO@aol.com
Subject: Re: LSD

Scott:

So glad that you have such faith in the Quaife. However, I think that we
should get a few things straight.

Porsche was given a Quaife to test by Ron Mosier at Autotech (he was the
original importer to this country) some 8 years ago. The did test it, and
returned it's lifeless, worn out, burned out remains to him. They said that
it was not up to their standards, thanks.

You say "just because the factory uses it does not make it superior." This
is one serious nearsighted statement. Porsche is certainly aware of the
products on the market. They do not live in caves. They use a clutch style
limited in the GT-2 Turbo, the Evolution Car, the New 993 Cup RSRs, all of
the European Cup Cars, the 94 RSRs, and all of the street vehicles. Now,
Porsche is in the business to win races. This sells cars. This makes them
money. This makes them happy. And you think that if there was something
better out there, they wouldn't be using it? Hello! Anyone home?

The new generation of differentials that change the locking effect were
created as a direct result of the early European Cup Races. They were not
(repeat not) used to "get around the understeer issue I mentioned earlier."
When you drive a 911 into the corners real deep (Which ABS and trail braking
techniques allow you to do), the rear end on a 911 tries to step out.
(Oversteer) The increased locking effect of the "trailing throttle lockers",
which is what the factory called these units in the beginning, increases the
locking effect and drives both rear tires at the same rate. This keeps the
rear end going in a straight line and helps with the oversteer under braking
problem. Of course, the more the locking effect, the more horsepower needed
when accelerating through a corner. (Anyone tried to push a race car with a
locked rear end with the wheels turned?) So, as you get back on the
throttle, the locking effect decreases, which is what is desired.

So now, lets review. Race Car. Corner. Don't want to do all my braking in
a straight line, because the 20 guys behind me are going to go past. Need to
brake when the car is turning. (Trail braking) ABS lets me do this.
Differential helps me do this. O.K. exit to corner. Inside tire does not
spin. The 20 guys behind me are still there.

Well, but what does the Quaife do for you in this situation? Well, gosh,
gee, you get to watch the 20 guys who were behind you drive away from you.

Greg Brown

-Chris

[A Quiet Boom]

"(Anyone tried to push a race car with a
locked rear end with the wheels turned?)"

LOL, I can park my dragcar (with spool) on a hill with the wheels turned in neutral. Have I tried to push it? Yes but it takes about 3-4 people if the wheels are turned.

As for extra clutches in a rear diff, I was speaking only from experience with Ford limited slips and using worn clutches which are thinner. It can be done in a Ford, I didn't say that it could be done with others but it's something to think about.

[CamB]

John - you're probably right - I will be getting terminology mixed up...

I tried checking again - this is the best I found and I am still confused. In the 2002 world, everyone talks about it but noone has done it...:

http://bimmer.roadfly.org/2002/messages/archive/msgsy2000w35/51882.html

Given the circumstances (ie, no money right now), I should be able to eventually provide a good descrption of the change because I will get an LSD one day and will have had plenty of practice driving in anger without...

Either that or I'll find a way of making a Nissan one fit - dime a dozen here in NZ.

[CamB]

Found another one:

http://bimmer.roadfly.org/2002/messages/archive/msgsy2001w13/91438.html

Cam

[jluetjen]

Hi Cam;
If the diff has spacers (thus replicating the HHHCCC arrangment), then you are right, replaceing them with clutches will tighten it up. Now your comment makes sense.

'Boom - That is definitely one of those "tricks" that you'd never learn by taking apart a new LSD. If you can fit extra clutches by using disks which are really worn, I guess that will work. How long does one of those last, especially in a road race application?