Diff, rear end, pumpkin, salad mixer, locker, LSD: Some of the common names you heard for the differential. Well, not salad mixer, I made that up – but it begs the question: What is a differential, what does it do, and how do the different types work?
The differential, like so many automotive topics is often surrounded by an overabundance of generalized science, marketing bluster and old wisdom:
“The wheel with no traction gets all the power”, or “from the wheels that slip to the wheels with grip”
Everyone has certainly experienced a differential in action, or lack of action if you’ve ever been stuck in the snow or mud. This article aims to demystify the differential and how the simple and advanced versions work to get you down the road, track or trail.
This is going to be a long journey, so strap in. Hit the shortcuts to jump straight to what you want to read.
What is a differential?
A differential is, put in simplest terms, a solution to a problem of geometry. Put two wheels on the ends of a stick then ask them to trace a circle there 2 things you’ll note immediately. The first is that the outside wheel in the turn needs to travel a much greater distance than the inside one. Second, you find that if the wheels spin freely on the stick, that the inside one spins slowly and the outside one much faster. This is the differential problem – a difference of speed and distance.
If the wheels don’t spin freely on the stick you’ll see the problem this creates as the wheels are forced to travel the same speed, even though they are traveling different distances the result is the inside wheel slipping as it tries to speed up to the outside one, and the outside one binding and scraping as it tries to slow down to the inside one.
Now keep the wheels attached to the stick, but cut the stick in half. The wheels would be allowed to go their different distances and speeds and no binding occurs. The bit in the middle that allows this to happen and still accept input power from the engine is the differential.
How do they work?
The ring gear gets its power from the driveshaft and because its bolted to the differential housing on bearings they they rotate together. The differential housing is attached to 2 or 4 spider gears (top and bottom in my lovely hand drawn illustration above) which push with equal force against 2 side gears that are affixed to each axle shaft, and then to the wheels.
This type allows for power to both wheels and speed bias but no torque bias; I.e. different rates of speed, but always the same torque. This can be confusing when you hear phrases like “all the power goes to the wheel with no grip,” or with the observation of one wheel spinning freely on a stuck car — but with an open type differential, the torque is always split 50:50 — or if you reduce this ratio, 1:1. This figure is known as the torque bias ratio – it defines the relationship between one side of the axle to the other in terms of torque. With a 1:1 ratio, if one wheel can apply X amount of torque to the ground, the other side is limited to X amount of torque. The ratio is what defines an open differential.
Traction enters the equation
Newton’s 3rd law states that every action has an equal and opposite reaction. Follow the power from the engine where torque generated at the crank, to the transmission where it is multiplied by the gear ratios, to the final drive where it’s multiplied even more by the rear gear/pinion gear, then through the differential to each wheel in the axle set where the tire pushes against the ground.
Well, the ground pushes back with equal force and the vehicle moves. This pushback is called traction and it’s a function of weight on the tire and the friction between the tire and the road surface. The ground pushes back (opposite) and the resistance it provides can be called traction (equal). What this also means is that the torque your vehicle can transmit is equal to the pushback from the ground due to traction an no more.
Here’s an example: A stuck bolt might take a 2 foot breaker bar and a great deal of effort to break loose, but once its loose and spinning freely it doesn’t matter how strong you are, or what kind of lever you are using, it’s only going to take a small amount of force to turn.
The amount of torque that can be transmitted to the ground is equal to the resistance offered by the ground as traction. In an open differential, the torque bias ratio says that the good tire can only get what the slipping tire gets in terms of torque. Imagine you had another stuck bolt to break, but you could only use the force that it took to spin the loose nut to do it. Not much will happen.
This is why in snow, a vehicle with an open differential will appear to send all the power to one tire while the other gets nothing. In actual fact they are both getting the same amount of torque but there isn’t traction on one side and the limited torque available to the other side is insufficient.
That’s the problem, what’s the solution?
The differential locker is the easiest of the differentials to describe, because when it’s in a locked state there is, effectively, no differential at all. When unlocked they [typically] act like open differentials, which is ideal for street-ability. However, when locked these act as one solid shaft with no speed bias, but with bias ratio of 100:0, or in practice, infinite. With a locked differential even if one wheel is on glare ice or high up off the ground, the wheel with the best traction calls the shots for torque — not the one with the least traction as in an open differential. A locked differential can bias torque all the way up to the engine limit or the limit of traction on the good tire or tires.
In an open differential, we split the shaft in half, in a sense, and the power flow looks something like:
Torque at the ring gear (TrG)/2. With a locked differential, the shaft isn’t split, it’s whole and the equation is TrG/1. Either tire gets as much of the ring gear torque as traction allows and demands.
There are basically 2 types of lockers:
- User selectable
- Auto locking
With a user selectable locker, the user asks for the differential to lock via a switch, button, or lever, and the axle locks the ring gear and axle shafts together, bypassing the differential. This can be done pneumatically, electronically, mechanically, or hydraulically.
These can be Air lockers, factory e-lockers, hydraulic lockers like ARB, Yukon, TJM and others, as well as factory installed electronic and hydraulic lockers (Toyota, Jeep, Mercedes, etc) use a design where mechanism locks the side gears to the case. With air lockers, an air fitting must be drilled, and a compressor fitted to supply air in order to actuate the system. With aftermarket E-lockers, a hole is still required for installation but no compressor is needed. Factory e-lockers have actuators built into the housing and presses on the locking collar by way of solenoid. Factory user selectable lockers typically offer the most benefits among full lockers with the fewest drawbacks followed by air and then by aftermarket e-lockers.
Harrop, Eaton and Yukon as well as others offer aftermarket e-lockers. In these aftermarket installations, the buyer doesn’t have the benefit of the axle being designed to have a solenoid push through the housing to actuate the locking collar, and so another mechanism is necessary. With pneumatic actuation there is enough pressure to lock internally, but in this case the signal is sent to an electromagnet that actuates the mechanism by requiring a pin to climb a ramp to provide the locking effect.
The advantages of this design is that it acts as open normally, fully locked when selected, and is able to be retrofitted to axles not designed for them. Because of the nature of the mechanics on these, they often have the disadvantage of needing forward motion to ramp into a locking position, which means that while they work in reverse, they can unlock between forward and reverse and can become unlocked without torque input.
Auto lockers, unlike user selectable, rely solely on mechanical action to engage and disengage. The most common types are known as Detroit or lunchbox lockers. These have a pin in the center of a case with gears that mesh with corresponding gears on each half of the axle shaft. Under low torque input the pin is centered and case gears slide past each other and provide differential action when needed in a ratchet type motion. When high torque is applied, the pin is forced against ramps on either side of the case splitting them apart causing the gears to mesh tightly, locking the shafts together. They are very effective, work in both forward and reverse, but are noisier and less predictable in street driving. Their main advantage is simplicity, ease of install, cost, and availability.
Another type is the NoSpin which is sort of like a Detroit locker in reverse where its locked all the time until speed bias is needed, at which time it pops out of lock to allow for differential action. These are good for very heavy applications, like commercial vehicles or for vehicles that are mostly driven off-road.
Another type is the Eaton Mlocker, commonly as the GM G80. Though not available as an aftermarket item, its common and unique enough that it’s worth including on this list. This design is an open differential most of the time, only locking when a speed difference between the axle shafts is great enough (120 rpm). At this point a weight is spun which forces it to split open with centrifugal force to engage a pawl that forces a locking collar to engage, disabling the differential.
This has the advantages of a Detroit locker with few of the drawbacks. The major issue with this type is that they can engage suddenly and violently with an over eager right foot and have known to break under such intense shock loads. As with all auto lockers, these lock and unlock as they please and will not offer the same level of predictability and control as a user selectable type.
The limited-slip differential, or LSD, is exactly what it sounds like – a differential that allows for limited slip. If an open differential is full slip, and a locker is no slip, the LSD is the middle ground – or to go back to the math we used above, something like TrG/1.5.
There are many different designs of LSD’s, and they operate very differently, each with their own advantages and disadvantages. They can be grouped into 3 basic categories.
In all cases the idea is simply to add friction between the 2 sides of the axle shafts to partially lock them together. Here a spring (blue) presses on the 2 side gears, against a clutch material in the housing (red) to increase the cooperation of the 2 side shafts. The benefit of a limited slip differential are the safe road manners of an open differential with the performance benefits of a locking differential… to a degree, as limited slips could also be described as limited locking.
In a limited slip, the torque bias ratio is usually in the 2-6:1 range. This means the differential has the ability to send 2 to 6 times the slipping wheel torque to the non-slipping wheel as the slipping one. This doesn’t sound like a lot, but remember that when a tire slips, its traction potential drops off a cliff. Preventing tire slip by balancing out tire loads not only gives you the additional torque bias, but prevents a single tire from being overpowered and going into a state of extreme low traction potential. Basically, not slipping to begin with is the best way to maximize torque output.
An example of this usefulness comes through in the math. If a slipping tire can only support 100 lbs-ft of traction, a limited slip will multiply that by the TBR (torque bias ratio) for the other side. Lets say the TBR is 4:1. With 100 lbs-ft going to the slipping tire, 100 x 4 = 400 lbs-ft can go to the tire with good traction, this 400 plus the 100 add up for a total of 500 lbs-ft with a limited slip with this TBR, versus only 200 lbs-ft (100 lbs-ft to each side) with an open differential. On a high grip surface like a track you can see how this would easily add up to more than enough torque to prevent the dreaded one wheel burnout.
With a limited slip, cornering safety and comfort is minimally affected and corner exit traction is enhanced. Depending on the type, high speed stability and deceleration stability is improved. Also, unlike a locked differential, if a shaft breaks power is interrupted, which reduces the potential for a yaw motion that could be dangerous at high speed. This makes them safer for street and track driving compared to something like a spool, which is a locker that doesn’t disengage.
Friction plate types, like posi types, address the problem by adding clutch type material to add friction between the 2 half shafts. The basic examples use a preload spring to provide a fixed pressure on the plates at all times. This works, but has negative effects on differential action, as there is always friction added — even when the limited slip function isn’t needed.
More advanced versions of this design have friction plates that are either mechanically, electronically, or hydraulically controlled to vary the amount of pressure on the plates, from minimal mechanical preload to fully locked. These systems are often highly sophisticated and not likely found on the aftermarket, but as factory installed units on high performance models. A good example of this type would be a Jeep Quadra-Drive that uses a hydraulic pump called a gerotor that is driven by the differences in axle speeds to pressurize a mechanism that presses on the clutches to progressively bind up to provide increasing locking effect, all the way to full lock. In Quadra-Drive II the gerotor is replaced by an electric motor and sensor to accomplish the same effect with a higher degree of control.
These types are speed biasing and torque biasing. A clutch type LSD is requires additional maintenance and special fluids to perform though many of them are also highly tunable in addition to being rebuildable.
Known more commonly by their brand name Torsen, these differentials got their name as a portmanteau of the words TORque and SENsing. These differentials are different from standard differentials as they not only bias speed and torque, they are mechanically torque sensing and aren’t bound up statically like a clutch type LSD is.
These devices are based around the principle of the worm drive (pictured above) where a worm gear drives a worm wheel, transferring motion at 90 degrees. The key is that the worm gear can drive the worm wheel, but NOT vise versa.
Instead of a spider gear/side gear configuration in most differentials, the 2 axle shafts each have worm gears that mesh with a series of worm wheels within the case. As the case rotates, the worm wheels bind against the worm gears on the axle shafts, pulling them along. Connected to the ends of the worm wheels, attached to the case are a set of standard 1:1 gears. When the vehicle turns, one worm gear is spinning forward at the exact speed the other side is spinning backwards (relatively speaking), and because a worm gear can spin a worm wheel, the speed is transferred across the gears to the other axle shaft with no bind.
When one tire has less traction, the side with good traction pushes against its worm wheel harder than the other side, forcing it against the case. That friction generates the locking action, similar to a clutch type. With this design, any torque imbalance is corrected (up to the bias ratio) without having a permanent preload, and without wearing on friction material or without affecting differential action.
Because these react to torque, there is no need for electronic control, and there is no delay of the limited slip action, meaning their reaction time is basically instant. Additional clutch preload can be added to this type to increase the torque bias ratio of this design to full, as is the case on the Driver Controlled Center Differential (DCCD) on the Subaru STI. More modern types, like the Type 2, 2R and 3 simplify the design and reduce internal friction, but work on the same basic principles. Any “helical” LSD design is based on these principles.
The other type in this same category is the AP Suretrac which is still used heavily in SxS applications, and uses geometric blocks instead of gears to transmit torque. It has been used in some Subaru models in the past, but is generally a rarity in cars today.
These work by using a fluid to transmit force. In the case of a limited slip a dilatant, or non-Newtonian fluid (fancy talk for a fluid that gets thicker with stress) is used inside a case as either as the sole coupler, as was the case with the original VW Syncro van’s 4wd system, or as a friction adder on top of an open differential.
The idea is that one side of the axle is attached to the case, which rotates a series of outer plates and the other side rotates a series of inner plates. If there is no speed difference between the 2 side they rotate together, as one side speeds up relative to the other, the fluid sheers and begins to thicken. This causes the plates start to bind together, forcing the speed difference to go away as the unit locks together with friction from the plates.
Oobleck, that strange corn starch and water mess you may have played with as a kid, is a dilatant fluid. You can hit it with a hammer and it firms up, or let it sit and turns back to fluid. These are typically found inside a standard open differential to reduce to stresses on the fluid as the spider gears carry the force from regular driving and the fluid coupling only comes into effect with slip.
These are very rare outside of OE applications and are more or less being phased out. Most commonly found on Subaru’s LSD’s that aren’t helical and as mentioned on older AWD vehicles with no center differential. These are effective as limited slip devices but can be abused easily and they fail completely with no option to rebuild.
A few other types worth mentioning are twin clutch systems like the GKN Twinster, as found in the Focus RS, and Honda VTM/SH-AWD, which aren’t technically differentials at all. Instead, they at 2 shafts connected to the power only via clutches that lock and unlock to bias both speed and torque. These can have advantages, like being able to perform center and rear differential duties in one unit (called an RDU, rear drive unit). These can also perform true torque vectoring, where instead of torque being biased across an axle, all the torque is simply sent to the axle shaft that is requesting it, with as little as no torque being sent to the other shaft.
These also can come with small planetary gear sets on each shaft to actually overdrive the engaged shaft to induce yaw in the vehicle, as is the case with the Honda/Acura units. These are harder to quantify with a torque bias ratio as they technically are limited slip defined by the force required to overcome the clutch slip on a single shaft. In the case of Honda, for example, they claim a full 70% of engine torque can be sent to a single rear wheels, which would act as a locked shaft up to the point where the clutch slips, or more accurately a very high bias limited slip.
Also important and worth mentioning, is what is NOT an LSD, such as a pseudo locker or pseudo torque vectoring: brake based traction adders. These go by many names like Jeep BLD, or Toyota ATRAC but they all work on the same principle and they are all NOT limited slip devices or lockers.
These systems are popular because programming is light weight, relatively cheap and highly tunable, and can be accomplished using existing ABS brake hardware — which means manufactures don’t have to invest in expensive hardware. They all operate as an open differential with the added trick of fooling the slipping tire into thinking its not slipping.
As mentioned before, open differentials have a 1:1 bias and unless you add a limited slip function BETWEEN the axle shafts, it will always behave that way. Brakes, being on the outside of the differential and on separate shafts, don’t alter the TBR of the open differential. What these systems do is brake a slipping wheel, simulating high traction on that wheel, which doesn’t limit the 50% share the other gripping wheel is entitled to under ideal circumstance.
These systems will always be bound by TrG/2 and CANNOT bias torque across the axle; In other words, no more than 50% of the torque at the ring gear can ever go to a single wheel. This is more than sufficient to provide movement in vehicles with high ring gear torque, such as vehicles with low range gearing; as even with good traction on both tires, the TrG in low range is more than the traction potential can handle anyway before spinning both tires.
In single speed systems or ones without a lot of torque, these system will struggle compared to a locker or limited slip, as up to half of the precious little torque available to the ring gear will be absorbed into brakes and not into the ground. Some road biased systems like those on the Subaru WRX are characterized as torque vectoring, when they in fact are only using brakes to drag the inside wheel to induce yaw.
These systems are advantageous as they can be programmed for different terrains or driving styles and offer the smooth drivability, automatic operation, and low maintenance of an open differential — however they are not torque biasing, not torque vectoring, and are inherently reactive in nature.
Whew. That was a lot to go through!
If you’ve made it here, I congratulate you on your stamina. I’ve always been a bit of a nerd when it comes to this stuff, probably as far back as getting a full Torsen brochure and sticker from my dad when I was only 8 years old. He tried to explain it to me then, but it took another 20 years or so before it really clicked. Hopefully I’ve sped up your journey down this path considerably.