Re: A4 Electronic locking diffs... (kinda long)

[Andy Poling <andy@realbig.com>]

Jay sez:
> In a message dated 96-10-11 18:15:02 EDT, Andy writes:
> 
> > I guess, upon reflection, I can see that the EDL system can direct close to
> >  100% of the torque to a single tractive wheel (if the EDL can actually
> brake
> >  three spinning wheels as claimed).  
> 
> Is that really true?  If the EDL is braking a wheel, isn't the engine's
> torque still going towards spinning it??

That's what the differential is there for.

A traditional "open" differential (i.e. not a limited-slip, lockable, or
Torsen) will steer the torque in the easiest direction.  Under most
circumstances it's equally easy to send it in either direction so it splits
it between the two directions.  When a wheel is spinning due to lack of
traction, it becomes the easiest route for the torque so the open diff sends
the torque away from the tractive wheel to the spinning wheel.  Not optimal,
but this ancient design was simply designed to allow different rates of
rotation when the car is not traveling straight ahead.  This is what the A4Q
has in front and in back.  Audi combats the diff's stupidity by applying the
brake (via the ABS system) on a spinning wheel to make it a more difficult
route for the torque, forcing the differential to steer the torque away from
that wheel.

A locked diff (i.e. a lockable diff or, to a lesser degree, a limited slip
diff - LSD) will try to split the torque equally between the two
directions.  A locked diff will always split it equally.  A limited slip
will always *try* to split it equally, but with limited (pardon the pun)
success (e.g. maximum 20% torque steering on my Miata's silly putty/viscous
LSD) which still allows the majority of the torque to leak out the
non-tractive wheel.  An LSD also comes into play only after the wheel has
started spinning.  

An old-fashioned clutch-type LSD reacts to a significant speed differential
between the output shafts and tries to use a frictional clutch to even their
speed, steering the torque in the process.  The spring/centripetal-force
nature of it's actuation delays it's effectiveness since it requires a
significant difference in output shaft speeds.  It's engagement can be
abrupt, and tends to take place too late (the tire/pavement interface is
already in dynamic friction mode).

A modern viscous (sp?) type LSD uses a viscous fluid that also reacts to
speed differential between the output shafts by heating the fluid and
increasing it's viscosity which increasingly resists differences in output
shaft speeds.  It takes time, however for the fluid to heat up, resulting
once again in late engagement.  The fluid also cannot force the output
shafts to exactly the same speed, so it's effectiveness is limited.

A Torsen diff works black magic through the use of worm gears which can only
transmit torque in one direction.  It will always try to send the torque in
the most difficult direction.  That sounds screwy, but the result is
wonderful:  the torque automatically goes where the traction is best.  The
Torsen also has it's limits though - it can only steer about 80% of the
torque maximum (from what I've read).  It varies the torque split instantly,
smoothly and continuosly though, as opposed to the operation of other
LSD designs.

Now, upon reading what I've just written about the operation of Torsen
diffs, I can't help but wonder if the center Torsen diff on my A4Q won't end
up fighting with the EDL if, say, both wheels at one end lose traction.  I
think the result is that the Torsen will only come into play above 25mph
when the EDL is no longer active, but I don't know how far Audi goes with
the EDL braking.

I still think the ultimate solution would be Torsen diffs in all three
locations, front, center and rear.  Can anyone offer a reason why this is a
bad idea?

-Andy

'91 Miata: "ZIP"          '96 A4 Quattro: "ZOOM"       '84 Yam RZ350: "ZING"  
                        '88 Bronco II 4WD: "ZZZZZ"