Viscous LSD’s have a pretty bad reputation among anyone who does real performance driving, but Subaru owners, who always have viscous centers and often rears are quick to point out that viscous LSD technology has evolved drastically in recent years. So much so, that they are now standard equipment in lots of ultra-high performance applications (like the Bugatti Veyron). I was doubtful but curious…
A viscous LSD is composed of alternately arranged inner and outer plates which spin in opposite directions, and are separated by a silicone based oil sealed inside the differential case. You probably knew that. The common misconception is that when the speed between the inner and outer plates increases, the oil inside heats up, expands and forces the whole unit to lock up. That’s what I had thought, but it’s only partially true. Like any oil, the silicone based oil inside actually loses viscosity as it heats up… which actually reduces the torque required to spin the plates. The real intention is that the viscous LSD gets it’s lock up power from the shearing of the oil against the plates inside the unit. The faster the plates spin, the greater the shearing, the stronger the lock up… except that the faster the plates spin, the greater the heat that is generated and the less viscous the oil inside becomes. So you end up with a non-linear relationship between lock up and speed differential.
This is quite simple in theory, and perhaps relatively comparable to a 2-way LSD (viscous works under accel or braking of course) with a minimal initial torque and a stupidly long shallow cam. However, while a clutch LSD can easily and predictably reach 100% lock up, a viscous can be considered to work against itself. I.E. as it locks up, the speed difference between the plates decreases and the diff opens back up. As this happens though, temperatures increase, and if as a driver we keep feeding the diff power and creating the plate speed difference, temperature and pressure will increase so much that the plates themselves might start flexing and even contact one another causing the unit to lock up much harder than it would under normal use. When you hear about viscous “humping” or STA (self-torque amplification) that is what’s going on.
As a drifter, humping is certainly something to celebrate, and even in rally or rough surfaces when speed is the main priority this is a tolerable side effect, but in other situations it can be less than desirable as the differential essentially locks. Furthermore, while in normal operation a viscous diff might experiences negligible wear, as soon as the plates are contacting one another wear becomes a real concern and that wear not only affects the ability to hump, but also to lock up in normal use. Another possibility is that the plates will deform to such a degree that they, and the input and output shafts in turn, become permanently locked together.
The real issue with viscous LSD’s though, is their inability to provide substantial torque transfer in anything but humping mode. The stock GDA viscous for instance, transfers 4kgfm at 100rpm or 29ftlbs at 100rpm. In contrast, the Cusco clutch type LSD for the same application might have an initial torque of as high as 100ftlbs and under full load might require 300 ftlbs or even more to break lock. For the viscous diff to transfer this much torque would theoretically require a several thousand rpm differential between plates… certainly any driver would give up and go home long before roasting rubber that badly, although humping would likely take over at some point and salvage the drive somewhat. Perhaps this is an extreme example, but even 100rpm differential is notable, representing a speed difference side to side of 12km/h on an axle with 25″ tires. And this is why we cannot make a super stiff viscous LSD with tightly packed plates and frozen molasses oil: because it would activate intrusively all the time. Consider that when turning a corner the outside wheels spin faster than the inside, and the front axle usually spins faster than the rear. So even at neutral throttle a super stiff viscous could activate, and over time temperature and pressure could build, activating humping and causing drivability issues. Note that the clutch diff in this situation would only be as stiff as its initial torque.
And that exactly is why viscous LSD’s are no good for real performance. Not because the diff itself requires time to heat up and work (as many would say), but because energy is wasted first creating a speed differential between input and output shafts before sufficient locking force can be provided. When seconds are of importance, nevermind hundredths of seconds, the ability of a clutch LSD to react in sync with your right foot is a tremendous advantage. Of course, for running groceries around and getting an axle off a patch of ice a viscous LSD might be the best option: it should never wear out, it’s operation is indiscernible, and it’s cheap to manufacture.
But these are not the only highlights of viscous technology. You’ll noticed that I’ve neglected to mention Torsen type LSD’s in this post. It’s a comparison for another time, but one that does well to highlight some interesting advantages of viscous differentials.