re: zexel paper (where's my specs?)

[Eaton Dave <dave.eaton@minedu.govt.nz>]

for what it's worth, i've included my comments on scott's (and jeff's)
comments.  there is clearly almost no interest in this topic (did pavlov
have a dog?), so this is probably of very little interest.  so, hit the
delete key.  anyway...

dave
'95 rs2
'90 ur-q

-----------------------------
> 
> Date: Thu, 11 Jun 1998 11:21:42 EDT
> From: QSHIPQ@aol.com
> 
> I had several opportunities to read the Zexel paper in depth over the
past
> couple days, and like Jeff, I don't see any 'new' ideas here, nor
> 'conflicting' information to Stan the torsen man's paper or his
discussion of
> the device with Jeff.  What I do see, is the funniest verbiage for
simple
> concepts.  I also think it to be somewhat incomplete, since it doesn't
address
> the main 'fooling' issue with the torsen, which is relative slip
angle.  The
> claim that the "driver will naturally react" to a torsen seems like
somewhat
> of an oxymoronic statement to me, but hey let's look at this in some
detail:
> 
[snip]

scott, you've said that the torsen doesn't understand relative slip
angles for months.  the engineers who wrote the paper disagree with you.
wherever the text discusses increased front or rear slip ratios, this
is, of course, always in relation to the other axle.

exercise for the reader.  count the number of times the words 'side
slip', 'lateral slip', and 'longitudional slip'  occur in this section
of the paper.  the number of times *changing* slip angles are discussed.


the torsen doesn't understand slip? errrr....

with regard to the chocholek paper, there is *nothing* in that paper of
the behaviour of the torsen in varied cornering manoeuvres, or details
about the characteristics of that diff.  most of these things are left
unstated.  what the paper does a good job off, is developing a
mathematical model of the diff.

this paper also doesn't deal exhaustively with the application of the
torsen as a centre diff, but at least it gives pages more information on
the diff in a centre and an axle application, than does the chocholek
paper which accords the centre torsen 1 paragraph.
> 
> >This centre differential's passive basic characteristic actually
turns out to
> >be an active dynamic characteristic, as the parameters dictating its
dynamic
> >torque distribution are actually anticipating the vehicle's loss of
> >directional  stability.  This allows a particularly efficient and
immediate corrective
> >action during transitory situations at the limit of lateral
acceleration.
> >This proactive operating mode does actually prevent an eventual loss
of stability
> >instead of correcting it after it's occurrence, as electronic systems
with
> >accelerometers and yaw rate sensors do.
> 
> BIG problem with the above, is that 'anticipating' relative slip angle
is OUT
> of the ability of the device, it can be fooled.  So I might take
exception
> with the "particularly efficient and immediate corrective action"
part, since
> the anticipated event, by definition really isn't happening.
> 

i fail to see where you get this understanding of the diff from?  it
*does* understand relative slip angle differences between the axles.  in
as much as slip angles influence torque reaction (grip), it has to.
longitudinal and lateral slip are a *function* of each other (the
friction circle).  for example, as the side slip angle increases, so the
torque reaction (i.e.. available grip) at the axle decreases, and so the
longitudinal slip [must] also increase.   you seem to have some problem
with this.  it's simple physics.

the diff detects this.  on both axles.  all the time.  by design.

anticipation?  in as much as the torsen responds to changing slip, yes,
it is anticipating.

> >1.1 INTER-AXLE DIFFERENTIAL: OPERATING MODES
> >An interaxle centre differential operates in the following 4 basic
modes
> shown in Figure 10.  In reverse, DRIVE becomes COAST and visa-versa.
> >MODE 1: Drive, Rear high axle torque
> >MODE 2: Drive, Front high axle torque
> >MODE 3: Coast, Rear high axle torque 
> >MODE 4: Coast, Front high axle torque
> >In DRIVE mode, a Torque Sensing differential will distribute the
higher
> driving torque to the axle that tends to turn slower than the other
one.
> >In COAST mode the higher braking torque will be distributed to the
axle which
> >tends to turn faster than the other one.
> 
> A problem arises in "coast" mode scenario.  A car that oversteers on
power,
> then gets throttle lifted has two unsettling characteristics in an
audi awd
> car.  1)  Since the coast mode distributes braking torque to the axle
turning
> faster you go back to 50/50 which is understeer, BUT 2) the lifting of
the
> back of the car could cause  "resulting in a rapid drop off in lateral
> adhesion capability at the rear" (see below, another fancy definition
of
> spin:).
> 
1) switching of braking torque to the front is *not* [necessarily]
understeer, as much as it is removal of oversteer.  no grip at the rear,
so torque to the front.  sounds like the best of a bad deal to me.

2) any car that has power oversteer and then the operator lifts throttle
has a [big] problem.  would you rather be in a bmw m3 or a porker 911?
doubt it, for anyone who's btdt.

3) in this situation, the torsen is helping you.  by definition.  the
paper makes clear that torque is only sent to the rear in low-speed
manoeuvres.  at high speed power-lift, torque is sent to the front.
this is caused by the increased *relative* slip angle between the front
and the rear (there is greater slip at the rear, so torque is sent to
the front).  so the paper states that the diff is removing braking
torque from the rear and sending it to the front.  this leaves more
lateral grip at the rear and decreases the drop throttle oversteer.

i'll take that, thanks very much.

> >Vehicle stability is not a prioritary argument for operating in modes
1 and 3
> >however.  In case of too low a locking effect value in mode 1 (DRIVE,
rear
> >high axle torque), the adverse consequence would be to let the front
axle slip
> >ratio increase to high values.  This will cause increased
understeering at the
> >imit, which is a situation that can be easily controlled.
> 
> ...  Unless the fronts have reached their "saturation limit", in which
case
> you understeer right off the road...  Not so "easily controlled".
> Not sure I'm with the boys on this, if an audi has power up oversteer,
mode 1
> is a big problem, cuz here we started the turn at o power U, power up
O, now
> they claim U.  "Easily controlled"????  Given a low powered audi on a
high cf
> surface, maybe....
> 

lets keep our feet on the ground here, scott.

we know that at the end of the day, 4x0 = 0 and, awd or not, you're
going to visit the scenery if you're going fast enough, and given enough
provocation.

their point is quite right.  understeer is a much safer condition to
oversteer, which is why all manufacturers seem to make their cars
exhibit this characteristic in all but the most extreme cases.  the
solution to understeer *is* easy.  power off. if that doesn't help,
you've got to do some driving to get out.

> >In case of too low a locking effect in mode 3 (DRIVE: reverse driving
or
> >COAST: rear high axle torque), the consequences are not relevant as
far as vehicle
> >stability is concerned.  Vehicle dynamics are not involved in reverse
driving
> >and during a drop throttle manoeuvre, the front axle always tends to
be the
> >faster one (therefore corresponding to mode 4).
> 
> Not on power up oversteer, or if cf gets low enough.  The consequences
of axle
> braking on lift are significant, the torque is all to the rear due to
slip
> angle (O), then coasting creates a 50/50 trend, that's U to an audi
awd car.
> "Tends to be" may exactly not be.  Let's try this at Steamboat cf's
with a 90q
> 10v, for example.
> 

mmmmm....

the part of the paper which you're quoting from is dealing with "low
speed cornering".  power-up oversteer seems to me to be dealt with in
the following chapter appropriately labelled "high speed cornering"
mode 4, as they say.  which is coming right up...

> 
> >1.3 High Speed Cornering: Power Oversteering
> >During a cornering manoeuvre at low speed and low torque (Figure 11),
the
> >higher driving torque will be to the rear axle (kinematic condition:
front
> >axle turns faster than the rear axle).
> >When more input torque is added, vehicle speed rises, the rear axle
slip
> >ratio increases (elastic conditions catch up the kinematic
conditions) until the
> >rear axle reaches the speed of the front axle.  At that time there is
no
> >differentiation at the centre differential, which operates as a rigid
axle. 
> 
> OK, defined audi awd above.  Start with U, then power up O, then the
rear axle
> reaches the speed of the front now we have a "rigid axle" (50/50 dist)
which
> is U.  Here comes the bite scenario...
> 
scott, the way you throw "u"s and "o"s around, you'd think this was
playschool.  fact:  with the torsen, corner entry is not "u" as you
describe, certainly wrt the locked centre, because more torque is being
sent to the rear.  so your "u-o-u" is simply disengenious.  it's
actually "n-o" in the scenario that paper is discussing, as the torque
originally sent to the rear, is now being split equally.  "steady state"
as the paper says.  with the locker it is "u".

> >From this neutral "steady state" condition, there are two possible
dynamic
> >behaviour evolutions:
> >a) The front axle slip ratio increases causing vehicle understeering.
> >The centre differential will react by biasing the surplus torque to
the rear
> >axle thereby correcting the understeering.  Should this correction be
> >insufficient, the driver will naturally react by releasing the
throttle which
> >will reduce the understeering.  This manoeuvre does not need driving
skills
> >and can be done by any driver.
> 
> Thanks for the vote of confidence.  Care to comment as we lower cf?
>  Ok, from the U-O-U we have above the possibility of a) which is U, OR
> 
> >b) The rear axle slip ratio increases causing vehicle oversteering.
> >The centre differential will react by biasing the surplus torque to
the front
> >axle, thereby correcting the oversteering (Figure 12).
> 
> ...  the possibility of b) which is O then U
> 

ok, playschool again.  here the torsen is reacting to the changing
relative slip angles (front to rear), and releaving the front axle
understeer condition by sending more torque to the rear.  a good thing?
sure, i'll take it.

> >In this situation the driver will require consistent assistance from
the
> >centre differential because his instinctive reaction will be a
throttle release that
> >will worsen the initial oversteering instead of correcting it.  We
know that
> >oversteering is extremely unstable.
> 
> Mild understatement here.  What we have is U-O-U-O-U or U-O-U-U.  They
say
> that the O is hard to control here, how bout 2 O's in the same turn.
Not sure
> the 3 U's in either scenario are any better, btdt.  Oversteering is
extremely
> unstable, yup sure is, especially when you are Understeering the other
times.
> And relative slip angle is also changing....
> 

this is just stupid scott.  do you seriously think that audi's with the
torsen react like this going around a high-speed corner?  or that, if
they did, drivers wouldn't notice?   seriously????

right, they don't. 

the paper states that during a high speed cornering manoeuvre, the car
will enter a steady-state, neutral condition.  good.  after this, you
will either get understeer or oversteer.  why?  because there are a
number of *dynamic* conditions in operation which are dependent upon
speed, corner radius, cf, steering input, throttle position, weight
transfer and torque available (among others).

the paper states this, and that if understeer results, then torque is
sent to the rear.  result: less understeer.  which is good.

or that, if oversteer results, torque is sent to the front.  result:
less oversteer, which is also good.

it's called throttle sensitivity and, with a good chassis (ur-q or
similar), it makes the car delightful in low-cf, or high torque
situations.  balanced on the throttle.  a little bit more throttle, and
you get push, button off, and the car returns to neutral.  or you get
oversteer, so lift slightly, and the rear tucks back in.  just as the
paper says.  and many road testers with the torsen ur-q have described.

just as anyone with a torsen ur-q knows.

but, don't take my word for it, try a torsen ur-q yourself.

u-o-u-o-u-o-u-o-u?  nope.

the paper says consistent (not "constant", jeff) assistance is required
from the diff during a high speed corner when oversteer results.  ok,
fine.  as anyone whoc has driven a torsen car in these circumstances
will tell you (certainly with my 2 chassis), the car is consistent
during this ragged edge corner.  so posted (in low cf conditions).  just
as the paper says.

> 
> > Indeed, the rear axle's side slip angle
> >rapidly increases (see Figure 13), reducing it's potential to
withstand the
> >driving torque and allowing increased longitudinal slip, resulting in
a rapid
> >drop off in lateral adhesion capability at the rear.
> 
> Hey let's just say SPIN next time boys.

actually they are trying to tell us what the torsen is doing to
*prevent* the natural consequences of these manoeuvres.... (high speed
cornering throttle lift).  care to try that in a 911?  an m3?
> 
> >The front axle's side slip angle will normally not increase, as
steering
> >correction always tends to keep the front wheels in the correct
trajectory. 
> >Therefore the centre differential must ensure enough torque biasing
> >capability to relieve the rear wheels from longitudinal driving
forces, even for
> >increased side slip angles causing a very high traction potential
difference between
> >front and rear axles.
> 
> Front slip angle "normally not increase"?  Steering correction at or
near the
> limit of adhesion sounds like an oxymoron to me.
> 

huh?  to countersteer on oversteer is one of the most normal, natural
and fastest reactions.  are you trying to tell me that on high speed
cornering oversteer you *don't* correct with the steering wheel?

[snip]
> 
> >1.4 DROP THROTTLE OVERSTEER
> >The most critical situation for vehicle stability during cornering is
the
> >COAST mode (power OFF), as this generally corresponds to a "panic"
manoeuvre
> >(vehicle corner entry speed too high, or sudden decrease of curve
radius).  In
> >addition, the dynamic weight shift further decreases the rear axle's
vertical loads,
> >reducing its potential adhesion capability.
> 
[snip]
> 
> The biggest problem here is that we are talking about going from
maximum Trg
> rear to minimum Trg rear.  According to this paper that is O going to
U.  All
> from just lifting the foot.  While at the limit of adhesion, that
certainly
> could give you an 'F1' as the unloading of the rear and the loading of
the
> front, combined with the torsen "braking of the faster axle" can
easily exceed
> the available traction at the front tires.  
> 

"o" going to "u" by just lifting the foot?  absolutely friggin right.
it's called cornering on the throttle.  the control is your right foot.

never resulted in a spin for me though.  it could i guess, if my
f1-meter is asleep.

> >1.5 CONCLUSIONS
> >The ideal centre differential TBR layout in the 4 operating modes is
a
> >function of vehicle dimensions (wheel base, track width, centre of
gravity height,
> >etc.), suspension elasto-kinematic design (stiffness front/rear,
angular
> >variations, etc) and engine torque characteristics for given road
conditions.
> 
> So engine torque and center torsen could be good for dry cement
pavement cf,
> and not for anything less.  Given road conditions is a HUGE variable.
> 

actually, rephrase that last sentence in terms of chassis dynamics, and
you're getting close.

> >Therefore the ideal design characteristics for a centre differential
can be
> >determined after a great deal of subjective vehicle tests.  The
optimisation
> >will be a compromise between different set-ups, depending on the
surface
> >conditions (dry asphalt, wet asphalt, snow, ice, etc).
> 
> And given that every chassis optimisation between the v8 and the
20vUrq are
> different (but the torsen the same), audi sure did compromise some.
And we
> haven't even addressed the surface conditions (cf).  When we do
decrease cf,
> we can see that ALL the 1-4 modes become more significant to chassis
dynamics
> while turning.

yup, modes 1-4 operate regardless of awd, 2wd, fwd, rwd, porker, beemer,
audi.  open centre, locked centre, and torsen.  and your point is?

> 
> All in all, an interesting and very technical report.  Spin is a word
that
> these authors did not want to use.  "Exceeding the load and slip
limits" in
> the front of an audi, you go right off the outside of the turn.
"Exceeding
> the vertical load and increasing slip angles" at the rear, is a spin.
Both
> concepts kinda buried in verbiage.  But definitely addressed.  
> 
> This paper doesn't address how a Torsen can be fooled in a turn.  If
it did,
> it might be more complete.  It might also directly conflict with the
statement
> that the Torsen is:  "Anicipating loss of directional stability..."
That's
> just not something the device can do consistently in a turn.

if the paper agrees with you, you'd think it was ok? mmmmm....

so the zexel engineers  just kind of over-looked the bite?  

these are the same engineers who talk about the consequences of panic
throttle lift whilst cornering at high speed?  of high speed cornering
oversteer?  of high speed cornering understeer?  who talk of manouevers
at the limit of adhesion?

these are the same engineers who, it is apparent in the rest of the
paper, actually went out onto skid pans and took measurements of torque
splits?  who describe traction control systems hunting.  with
measurements to show this.  just plain overlooked it?

nope, don't think so.  however, i will say that it is clear that the
paper is not an exhaustive treatment of the torsen centre.  it is a
discussion of the application of traction control systems to torsen
cars.

other points:
1) it is good to see that we no longer have this bite @70% bullsh*t.

2) it is good to see that there are no more comments about "the torsen
wasn't designed for a centre application".

3) it is good to see that we have a clear description of the importance
of the chassis dynamics to the effective operation of the torsen.  as
well as just the bias ratio.

4) it is good to see that the paper deals with the centre torsen's
reaction to varying (front/rear) slip angles.

scott, you time and again make statements about "u-o-u"-type cornering.
as though this is an intrinsic part of just the torsen.  it's not, and i
can't believe that you think that it is.

for instance, lets look at the locker accelerating through a turn.
using your own hypothesis:

slow corner entry:
front axle wants to be "faster" than the rear.  result: fixed axle
speeds causes slip difference and wind-up torque between front and rear
axles. causing major understeer.  the tighter the corner, the more
understeer. result:"u".  torsen is "n" at this point with torque going
to the rear.

mid-corner (medium speed):
axles at the same "speed", so no slip difference. result:"n" for
neutral.  torsen is also "n"

corner exit (high speed): rear axle wants to be "faster" than the front,
causing another slip difference between the axles, except this time the
opposite (from the slow speed corner).  so now, the wind-up torque in
the system produces an oversteering moment.  result: "o".  at high
speed, this causes loss of torque reaction at the rear (i.e. more slip),
meaning more torque at the front (less slip).  result: back to "u",
until the rear slip angles recover.  this will cause increased slip at
the front, resulting in less torque reaction etc....

so we have: "u-n-o-u-o-u" in the same turn.  with the locker.  the
torsen is "n-n-o-n".