"Care to Comment" -> differential t...

QSHIPQ at aol.com QSHIPQ at aol.com
Tue Apr 9 20:56:08 EDT 2002

[ Picked text/plain from multipart/alternative ]
Please feel free to either send Corey's address directly or get him on the
torsen list.  For now, I'll respond to the post to all lists BUT the main
list because T*rsen is banned there.   If I overextend my welcome nerding on
the S car list, please give me fair warning....
In a message dated 4/9/02 5:27:50 PM Central Daylight Time,
randrews at sbcglobal.net writes:

>Here's one response to that Scott...I'd like to see what you have to say...
>"First of all, you've got a couple things wrong. You are assuming the
traction of >the rear gets exceeded first....which isn't hardly true.

I don't assume that at all (a better statement would be "which isn't ALWAYS
true").  What I do understand, is that the torsen changes engine torque
allocation based on exactly two inputs of which it can't differentiate:
Turning radius and traction.

>I will agree that when you transition from trail-braking turn-in, to
on-throttle, the >lightest corner of the car is the inside rear, and there a
chance that you can spool >up that wheel because of the lack of grip. That
can be overcome with technique of >throttle application, and with suspension

I don't understand that argument at all.  The torsen acts based on turning
radius or traction, that's all.  Power up oversteer isn't a suspension
problem, it's a torque allocation problem.

>Once you get past that point, however, you begin to transfer weight back to
the >rear of the car, therefore making the tractibility of the rear higher
than the front. >Especially when you get to full throttle, trying to get to
the apex, then tracking out >past the apex, the tractibility of the rear is
much greater than the front.

You are making traction arguments and turning radius arguments as a single
property.  They are independent variables that *can* affect each other.

>How do I know? How about countless amounts of data and seat time from 42 >
autocrosses in 2001?

The problem I had with going from autocrossing to 'on track' driving, is that
they are different animals in terms of handling evaluation.  And effects of U
and O are usually at a slower speed, and they don't usually become a problem
until you track them.  YMMV

>When we finally got a high bias Torsen to put in the car, the benefit was >
immediately noticeable. Now that the power-on understeer has been greatly >
reduced, we can apex earlier and get on the throttle earlier, which mean
higher >corner exit speeds.

Again, basic understanding of the device is necessary.  IF you use a higher
bias ratio you potentially shift more torque fore and aft.  AT the limit of
adhesion shifting 60% of engine torque (4:1 BR) sounds....  Overwhelming and

>How about 4 wheel speed sensors on the Stasis A4 touring car? I've seen
their >data, and the advantages of the higher bias Torsen were very apparent.

I don't buy into the argument that a wheel sensor will give you the data you
are looking for for wrt the torsen, necessarily.  Remember, the torsen is
TORQUE SENSING, which means by the time you measure wheel speed differences,
torque has shifted already.

>You could see the high bias Torsen doing its work, transfering power back
and >forth as traction levels were changing (which including heavy throttle
at apex, >bouncing over berms, and heavy throttle at track out). With just
the 2:1 Torsen in >their car, the car always had some amount of wheelspin in
the FRONT of the car, >which resulted in on-throttle understeer.

Again, I'm not convinced you understand WHAT the device does while turning.
IF the torsen was transferring power back and forth, with a 4:1 BR, that
means when TRACTION is exceeded, then 60% of the torque goes to the front
wheels, which results in power on U.  Once that condition causes the front
wheels to spin faster than the rears (traction and turning argument), 60% of
the power goes to the rears = power on oversteer.  The net effect of this
oscillation is that A) you aren't at the limit of adhesion or B) you have to
take a corrective action > steering/throttle/brake to change it, all
resulting in a slower corner.  One could accept your O argument if you could
keep the rear bias all the time, but that's not what you are claiming with a
4:1 BR.

>You bring up a good point about locking the diffs, but if you really think
about it, >you'd be limiting yourself.

A thorough review of audi racing would contradict that statement.  Solid
center diffs were found in just about every rally car, and still found in
racing audis today, rally or track.  A front weight biased car U's under
throttle, why fight it.  You aren't limiting yourself unless you try to make
a front weight bias car try to O under certain conditions.  The torsen IS the
limiting problem.

>A biasing system is dynamic, and can handle changing situations.

Only optimally in a straight line, the torsen follows weight distribution
EXACTLY.  The PROBLEM with a torsen, is it doesn't know the difference
between a traction argument and a turn.  This is WELL documented.

> Also, the "championship" winning S4's (the Champion S4's) are running a
viscous >center (which is also very expensive!). Although a viscous center
doesn't seem >like the best solution for road racing (widely used for rally),
it is working for them >because they've made the car bias most of the power
to the rear of the car at all >times.

Interesting conflict to Dave Lawson's interviews posted last year Corey. And
if I remember the rules correctly (if memory serves from my search last
year), diffs are NOT free, but can be welded, which means VC's are not

Before you data log playing with the Bias Ratio, I'd suggest a thorough read
of WHAT the device is and isn't capable of.   Then I'd suggest a simple
comparo of lap times (on a TRACK) between the torsen high bias ratio car and
baseline (welded center).  I think you might be surprised (I'm not). The
problem with racing torsens comes down to predictability.  Too many variables
affect it's torque allocation.


Scott Justusson

More information about the 200q20v mailing list