h a l f b a k e r y
Apply directly to forehead.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
The typical car is equipped with some device that
generates rotational force driving two of the wheels.
Because drivers are fundamentally indecisive, cars often
change direction*. As the net motion of a car is usually a
circle, the wheels on different sides of the car travel
To prevent embarrassing tire noises
and strange handling, cars are fitted with a differential
<link>. In most cases the problem is solved, it's pretty
robust, we can pack it with grease and retire to the pub
safe in the knowledge it will last decades.
However, should you get onto a surface with variable
traction, snow/ice being the usual culprits, then the wheel
with the least friction will spin around uselessly while the
wheel with the traction does nothing at all. What we need
is a way of allowing some differential movement, but when
there's a lot, a way of transferring some of the torque.
Several methods currently exist, the simplest is a clutch
<link>. These are somewhat uncivilized and wear out.
Torsen<link>, the main competitor here, is good but
expensive, so much so, that even most "4wd" cars stick
with open differentials. In radio-controlled racing cars
which are so light and ludicrously over powered that they
have essentially no traction most of the time solve the
problem by fitting enormous differentials and filling them
with variable viscous oils that act as a velocity-sensitive
damper. At full size, this would overheat and require too
So, how do we get the velocity-sensitive damping without
the mess? Enter the eddy-current brake <link>. If you try
and move a magnet adjacent to a conductor, you will
induce a current and produce a force. Examples of this
include the dropping a magnet down a pipe experiment
<link>. The principle is used as a power-independent brake
on roller coasters and on earlier generations of bullet
trains. The beauty is, that nothing wears out and it self
To incorporate this, we have the drive shafts fitted with a
thin disk, this is close to the differential case. The
differential case has powerful neodymium magnets in,
when the drive shaft spins quickly relative to the case (one
spinning wheel) the magnets will generate eddy currents in
the disk transferring torque to the case and therefore, over
to the other driveshaft/wheel. There, a relatively mild LSD
that should work forever and cost relatively little to make.
Secret bonus, if you ever drop a screw, it will likely be
found safe and secure on the diff casing.
*This is something of a scandal, investigation reveals car
journeys to be oscillatory in nature, with nested
oscillations of various periods. Within a 24hr period, cars
leave and return to the same location. Often, the ultimate
destination of a car may be the location from which it was
purchased. Significant energy savings might be afforded by
simply leaving the car stationary after purchase, and
having a bloody good think about where you actually want
to be, which is unlikely to be at a car dealership.
Limited Slip Differentials
[bs0u0155, Sep 21 2020]
Transferring Power through a Magnetic Couple
[xaviergisz, Sep 21 2020]
US5477093 - Permanent magnet coupling and transmission
[xaviergisz, Sep 21 2020]
It's a Knockout
Too highbrow for me [not_morrison_rm, Oct 01 2020]
||Not that wasteful; energy dissipation is algebraically related to speed of rotation, and most of the time the port and starboard shafts are only moving very slowly with respect to one another.
||It's when one wheel spins - something an LSD is meant to prevent - that there's significant differential (ha, ha) velocity.
||A bigger issue might be that the system has a tendency to alternately slip and then latch up, leading to snatching and making torque control problems worse.
||The best solution is probably to drive each wheel independently with a feedback-sensed electric motor. That's where a hybrid drivetrain wins; no need for a gearbox, and for a small vehicle a modestly sized engine (maybe in the 150 - 200kW range) and a small battery means that on startoff, the enitre output of the engine plus the entire battery surge capacity can be dumped into the motors (which will admittedly heat up, but can withstand brief overloads) until the vehicle hits that golden 150km/h moment.
||//alternately slip and then latch up,//
||The relationship between velocity delta is proportional, in a
very non-linear way. So there's not really a mechanism for
that, it's quite similar to a viscous coupling. A clutch-type is
more like that in behavior, since the transition from
engaged to slipping involves sticktion. There's likely a linear-
ish relationship between velocity delta and force once it's
slipping but un-sticking is a hill to get over, so there's the
propensity for oscillation.
||Yes, the possibility of oscillation is a major issue; just what you don't want, alternate drive/free run.
||The viscous coupling will have quite a nice smooth characteristic - it's the fluid heating issue with prolonged load that's problematic. There are external coolers for ATF (WKTE) but you'd have to couple it to a RWD diff with flexi pipes to keep the unsprung mass down, unless the vehicle has FIRS ...
||If the viscous diff is integrated with the gearbox in a FWD design you can patch in the cooler much more easily than for RWD.
||//it's the fluid heating issue with prolonged load that's
||hmm, place the drive shaft-coupled disks on the outside of
the casing, cool them like you might with inboard disk
||//Yes, a motor at each wheel// including the spare, of course!
||<Sean Connery voice>"The name's Current Limited Shlip
Differential, Eddy Current Limited Shlip Differential"</SCv>
This idea appears to add weight and complexity to
the car for little gain, solving a problem probably better
addressed with the 'motor at each wheel' proposal, so [+].
||//motor at each wheel solves everything neatly//
||This is such a good idea that it has met with universal
non-implementation in the automotive industry. There
have to be reasons. I'll have a stab at what they might be:
||Torque vectoring is limited: In a proper mechanical 4wd
system, 100% of the torque can be sent to one wheel,
handy while rock-crawling etc. in the motor per-wheel
situation it's 25% + any sort of temporary boost.
||Unsprung mass: If you want to put your motor on each
wheel it makes sense to build it right in, only then you
have a ~30kg mass on the end of your wishbone/swingarm
etc. which will ruin the ride/handling/grip and smash a
lot of wheels. The vibration will likely kill motors too. If
you move it inboard, then you already concede the need
for a transmission, so why not drive two wheels with one
motor and do your torque vectoring with an active diff?
||There aren't even any (fun) radio control cars with a
motor at each wheel.
||Liability: Electric motors are reliable, not infallible,
controllers less reliable. By quadrupling the motor &
controller count you add 6 additional points of failure.
With one motor and a transmission, a failed motor stops
the whole car. When one motor of 4 stops, it could easily
cause a dramatic change of direction or even for the car
to rapidly swap ends on a mild highway bend. If a car
stops, the car that runs into the back of it is liable.
If a car veers into the opposite lane, then that car/driver
is liable. In the case of a motor failure, the company
would end up with the bill. It's hard to think of a way to
make that situation fail in a safe way.
||<obligatory> Er, who is Eddy, then, exactly?
||He's the one currently making the differential.
||"Eddy's Currants", a homely brand of dried fruit ?
||No connection to Eddie Waring?
||See link "It's a knock out"
||No, thankyou. We are already too familiar with that particular execrable light entertainment feature, i.e. we are aware of its existence, which is actually too much information.