A device which increases or decreases the ratio of relative rotation between its input and output shafts is order to trade speed for power through gearing or torque conversion.
There are three types of transmissions:

Automatic Transmission - The Automatic Transmission has automatic gear shifting and an automatic clutch mechanism.
Manual Transmission - The Manual Transmission, often referred to as a "STICK SHIFT", the gears are changed by hand with the gear shift lever in conjunction with a clutch pedal.
Continuously Variable Transmission - A Continuously Variable Transmission provides automatic, seamless, shifting, which can not be felt. A Driver can shift between an Automatic Transmission and Manual Transmission while driving.

2WD - Two-Wheel Drive

This is by far the most common type of drive train in any car today. The engine drives the gearbox which sends its output to an open differential either on the front or rear axle, which in turn drives those wheels.

4WD - Four-Wheel Drive

Also known as part-time all-wheel drive, this system has an open differential on the front and rear axle and a transfer case on the output from the gearbox. 4WD is normally driving the rear axle with the front axle only coming into play in 4WD mode. The transfer case is the device that splits the torque between the front and rear axles. It has a selectable internal differential or viscous coupling to allow the front and rear drives to turn at different speeds if need be. Some trucks and SUVs have a selector with 2H, 4H and 4L on it - it looks like a second gear shift. This is actually controlling how the front and rear outputs of the transfer case get locked together. In 2H mode (2-wheel drive, high), it disconnects the front output completely and only drives the rear axle. In 4H mode (4-wheel drive, high), it engages the front output via the viscous coupling so that the axles can turn at different speeds, and now sends torque to both open differentials. In 4L mode (4-wheel drive, low) it engages a second set of reduction gears and locks the front and rear axles together so they must spin at the same speed.
Locking Hubs - On older 4WD systems, the front wheels could only be engaged to the transfer case by locking hubs. Essentially the transfer case was always sending torque to the front driveshaft and had no viscous coupling. To get into 4WD mode, the driver had to stop and get out, and lock the front wheels to the axles so they could be driven. In newer 4WD systems, the lockable hubs are still present on some models, but are designed more for mechanical sympathy and fuel economy than anything else. With the hubs unlocked, the whole front part of the drive system isn't being dragged along for the ride, which causes mechanical wear and a drop in fuel economy.

AWD - All-Wheel Drive Type 1

All-Wheel Drive or full-time 4WD. The simplest form has two open differentials - one on each axle - and a viscous coupling between. The engine drives the gearbox which drives two output shafts. One goes to the front open differential and the other goes to the viscous coupling, the output of which is connected to the rear open differential. Under normal conditions, this type of AWD system functions exactly like a 2WD car, driving only the front axle.The front wheels turn at a certain rate, and the rear wheels are dragged along for the ride. Both halves of the viscous coupling are spinning at the same speed so no torque is sent to the rear axle. If the front wheels begin to slip and spin, the input to the front of the viscous coupling begins to spin faster than the rear and because of its torque-converter-like design, this causes the rear output to want to speed up. At this point, the drive train is now transferring torque to the rear axle and the car starts to function in AWD mode.

AWD - All-Wheel Drive Type 2

Very similar to the type 1 AWD, it replaces the viscous coupling with a Torsen differential, and replaces the open differentials front and rear with either Torsen or limited-slip differentials. This is the only true all-wheel-drive system because it will always drive all four wheels.

Most every automobile has a differential. Differentials basically allow two wheels on the same axle to turn at different rates. Two-wheel-drive automobiles do not have a differential on each axle. Four-wheel-drive (all-wheel-drive vehicles) have differentials on each axle. For all-wheel-drive, there is often a third differential in the driveline from the front to the rear of the vehicle, to allow the entire front and rear axles to spin at different speeds to each other.

Open Differentials

They are the most common, and they supply the same amount of torque to each output. Open differentials have a few essential components. The input pinion gear is the gear that is driven from the drive train, typically the output shaft from the transmission. It drives the ring gear which, being larger, is what gives that final gear reduction I mentioned. Attached to the ring gear is the cage, containing two captive pinion gears that are intermeshed with the two output pinion gears, one connected to each axle. The captive pinions are free to rotate. As the input pinion spins, it meshes with the ring gear. The ring gear spins, spinning the cage and the two captive pinions. When the vehicle is traveling in a straight line, neither drive pinion is trying to spin any differently from the other, so the captive pinions don't spin and the turning of the ring gear is translated directly to both drive pinions. These are connected to the drive shafts to the wheels, so, the ring gear spins the wheels at the same speed that it is turning. When the vehicle starts to turn a corner, one of the wheels is going to want to spin more quickly than the other. At this point, the captive pinions come into play, allowing the two drive pinions to spin at slightly different speeds whilst still transmitting torque to them.

Limited-Slip Differentials

Sometimes known as "positraction" it still has all the components of an open differential but there is two crucial extra elements. The first are spring pressure plates which are a pair of springs and pressure plates nestled in the cage between the two drive pinions. These push the drive pinions outwards where the second extra element comes into play - clutch packs. The backside of the drive pinions have friction material on them which presses against clutch plates built into the cage. This means that the clutch is always going to try to behave as if the car was moving in a straight line by attempting to make both output pinions spin at the same speed as the ring gear and cage. However, when a car with a limited-slip differential goes into a corner, there are enough forces at play that the drive pinions begin to slip against the clutch material, thus allowing them to turn at different speeds again. The stiffness of the spring pack coupled with the friction of the clutch pack together determine the amount of torque required to overcome the clutch.

Torsen Differentials

Torsen differentials are a derivative of open differentials. They derive their name from their function - Torque-Sensing. When the torque going to both outputs is the same, a Torsen differential essentially works just like an open differential. The change comes when the torque going to each output begins to change, for example as a result of a slippery road surface under one wheel. When this happens, what's known as an Invex gear train begins to bind together. The Invex gear train is designed with a torque bias ratio in mind that determines the ratio of torque that it can split between the outputs as the gear train begins to bind together. Torsen differentials are normally used in-line between the front and rear drives for performance all-wheel-drive vehicles, to split the torque between the front and rear axles, rather than the left and right wheels.

Locking Differentials

Locking differentials are another derivative of open differentials but with an electronic, pnuematic or hydraulic actuation system that locks the two drive pinions together as if they were a solid axle. This is for use in serious off-roading, where a vehicle will spend a lot of time with one wheel per axle in the air. By locking the differential, it behaves like a solid axle and both wheels are spun together.

The clutch is what enables you to change gears, and sit at traffic lights without stopping the engine. You need a clutch because your engine is running all the time which means the crank is spinning all the time. You need someway to disconnect this constantly-spinning crank from the gearbox, both to allow you to stand still as well as to allow you to change gears. The clutch is composed of three basic elements; the flywheel, the pressure plate and the clutch plate(s).

The clutch pedal is connected either mechanically or hydraulically to a fork mechanism which loops around the throw-out bearing. When you press on the clutch, the fork pushes on the throw-out bearing and it slides along the lay shaft putting pressure on the innermost edges of the diaphragm springs. These in turn pivot on their pivot points against the inside of the clutch cover, pulling the pressure plate away from the back of the clutch plates. This release of pressure allows the clutch plates to disengage from the flywheel. The flywheel keeps spinning on the end of the engine crank but it no longer drives the gearbox because the clutch plates aren't pressed up against it.
As you start to release the clutch pedal, pressure is released on the throw-out bearing and the diaphragm springs begin to push the pressure plate back against the back of the clutch plates, in turn pushing them against the flywheel again. Springs inside the clutch plate absorb the initial shock of the clutch touching the flywheel and as you take your foot off the clutch pedal completely, the clutch is firmly pressed against it. The friction material on the clutch plate is what grips the back of the flywheel and causes the input shaft of the gearbox to spin at the same speed.

Burning Clutch - This is when you hold the clutch pedal in a position such that the clutch plate is not totally engaged against the back of the flywheel. At this point, the flywheel is spinning and brushing past the friction material which heats it up in much the same was as brake pads heat up when pressed against a spinning brake rotor. This can also happen unintentionally if you rest your foot on the clutch pedal in the course of normal driving.

Slipping Clutch - This is a clutch that has a mechanical problem. Either the diaphragm spring has weakened and can't apply enough pressure, or more likely the friction material is wearing down on the clutch plates. In either case, the clutch is not properly engaging against the flywheel and under heavy load, like accelerating in a high gear or up a hill, the clutch will disengage slightly and spin at a different rate to the flywheel. You'll feel this as a loss of power, or you'll see it as the revs in the engine go up but you don't accelerate.

Just like a manual gearbox, an automatic gearbox needs a method of decoupling the constantly-spinning engine from the gearbox components. To do this it uses a torque converter which is a viscous fluid coupling which has the ability to multiply the torque from the engine 2 or 3 times in certain conditions. When the engine is spinning far faster than the gearbox, the whole design allows the torque from the flywheel to be multiplied. As the car gets up to speed, the multiplication factor drops until it becomes 1x once everything is in motion and the impeller and turbine are moving at almost the same speed.

A torque converter consists of three basic elements.

Impeller - The impeller is attached to the torque converter housing which itself is attached to the engine flywheel and is basically a centrifugal pump. As the flywheel spins, so does the impeller and the vanes take the fluid from the central part of the torque converter and fling it to the outside creating a pumping action. The fluid then circulates around the outer edge of the torque converter and back into the turbine.

Turbine - The turbine is basically the opposite of the impeller - it's like a ship's propeller in that the fluid passing through it causes it to spin. The turbine is connected to the input shaft of the gearbox via a splined shaft so as the turbine spins, so does the input shaft to the gearbox. The fluid passes through the turbine from the outside towards the inside. Finally, as the fluid reaches the central core, it passes through the stator.

Stator - The stator is designed to help redirect the flow into the inner vanes of the impeller. With this mechanism, the fluid is constantly being circulated.

When the engine is idling, the fluid is pumping around without a lot of force and the amount of torque on the turbine is minimal. As you accelerate, the impeller speeds up and creates larger forces on the turbine which in turn spins more quickly and with more torque. Because it's connected to the input shaft of the gearbox, this feeds more rotational speed and torque into the gearbox and the car starts to move forwards. It's because of this viscous liquid coupling that automatic gearboxes have a certain amount of 'slop' in them - the engine can rev up and down without the car actually changing speed too much. It's also the reason automatics are less fuel efficient because the torque converter uses up energy from the engine simply in its design by spinning the hydraulic fluid.

For sportier vehicles or those with specialized needs, some torque converters include a hydraulic clutch. Once the car is moving and in top gear, the clutch engages and locks the turbine to the impeller. Once that happens, the whole torque converter spins as one and the viscous coupling becomes redundant - effectively the gearbox now behaves like a manual because the engine flywheel is connected directly to the gearbox input shaft. By locking all the components together, it makes the car as fuel efficient as a manual when in top gear because the energy that was being used up in the viscous coupling is no longer required. It also means instantaneous throttle response - you push the accelerator and the car accelerates instantly just as with a manual.

With the clutch engaged, the lay shaft is always turning. All the helical gears on the lay shaft are permanently attached to it so they all turn at the same rate. They mesh with a series of gears on the output shaft that are mounted on slip rings so they actually spin around the output shaft without turning it. The dog gears are mounted to the output shaft on a splined section which allows them to slide back and forth. When you move the gear stick, a series of mechanical pushrod connections move the various selector forks, sliding the dog gears back and forth.

When the gear stick is moved to select fourth gear, the selector fork slides backwards. This slides the dog gear backwards on the splined shaft and the dog teeth engage with the teeth on the front of the helical fourth gear. This locks it to the dog gear which itself is locked to the output shaft with the splines. When the clutch is let out and the engine drives the lay shaft, all the gears turn as before but now the second helical gear is locked to the output shaft.

Grinding Gears - In the above example, to engage fourth gear, the dog gear is disengaged from the third helical gear and slides backwards to engage with the fourth helical gear. This is why you need a clutch and it's also the cause of the grinding noise from a gearbox when someone is cocking up their gear change. The grinding noise is the sound of the teeth on the dog gears skipping across the dog teeth of the helical output gears and not managing to engage properly. This typically happens when the clutch is let out too soon and the gearbox is attempting to engage at the same time as it's trying to drive. In older cars, it's the reason you needed to do something called double-clutching.

Reverse Gear - Typically, there will be three gears that mesh together at one point in the gearbox instead of the customary two. There will be a gear each on the lay shaft and output shaft, but there will be a small gear in between them called the idler gear. The inclusion of this extra mini gear causes the last helical gear on the output shaft to spin in the opposite direction to all the others. The principle of engaging reverse is the same as for any other gear - a dog gear is slid into place with a selector fork. Because the reverse gear is spinning in the opposite direction, when you let the clutch out, the gearbox output shaft spins the other way - in reverse.

Automatic gearboxes don't use a clutch they use a torque converter. In a manual gearbox, the dog gears lock and unlock different sets of helical gears to the output shaft in order to give the various gear ratios. In an automatic gearbox, the planetary gearset produces all the different gear ratios in one go and with only one set of gears.

A quick primer on how planetary gearsets work

Any planetary gearset has three main components. The sun gear, the planet gears (and their carrier) and the ring gear. Any one of these three components can be locked in place, but more importantly, any one can be the input or the output drive. Locking any two of them at the same time will always produce a 1:1 gear ratio. So how the hell does that work? One set of gears for every ratio you need? The work of the Devil? Time to get the old brain massager out again. For this example I'll talk about a planetary gearset with a ring gear that has 75 teeth and a sun gear that has 25 teeth. The following table shows how sending the input to one set of gear and locking another set can give a wide variety of gear ratios.

Input	Ouput	Locked gears	Calculation	Resulting ratio
Sun	Planet Carrier	Ring	1+(Ring/Sun)	4:1
Planet Carrier	Ring	Sun	1/(1+(Sun/Ring))	0.75:1
Sun	Ring	Planet Carrier	-Ring/Sun	-3:1 (ie. reverse)

So that table basically has one reverse and two forward gears. Need more gears? Add more planetary gearsets with different numbers of teeth and link them together. Make the ouptut of one become the input of another and you can start to multiply up the number of gears available to you. The image below shows an example planetary gearset with the planet carrier in cutaway.

Again, something like this works much better in motion. Below I've rendered an animation showing a planetary gearset in motion. In this example, the blue ring gear is locked. The input is the yellow sun gear and the output is the planet carrier. The planetary gears are the green ones, and the planet carrier is semi-transparent so you can see what's going on inside. This shows clearly how the input to the sun gear can be geared down - in this case by a ratio of 2.7:1.

Compound planetary gearsets In reality, automatic gearboxes typically use one or more compound planetary gearsets instead of chaining regular gearsets together. They look just like a regular planetary gearset from the outside, but inside there are two sun gears and two sets of intermeshing planet gears. There is still only one ring gear though. With a single compound gearset, the number of ratios available increases to 4 forward ratios and one reverse. The image below shows an example compound planetary gearset again with the planet carrier in cutaway. In my example, the planet gears are arranged as inner and outer planets. The inner ones are shorter and only engage the small sun gear and the outer planet gears. They in turn engage the larger sun gear at the bottom and the outermost ring gear. Another configuration would be to have the two sets of planet gears next to each other but slightly staggered so that only one set meshes with the ring gear. Would you believe there are people paid to come up with this stuff? Makes you wonder if you shouldn't just accept that an automatic gearbox simply works and that you don't want to know why.

I could now go on to explain to you how all the different ratios get selected but if I did, I'd lose most readers at this point and all the typing and fine imagery in the rest of the page would go to waste. For the sake of a working example, I will explain the first two gears though.
Looking at the image below, When first gear is engaged, the smaller sun gear (green) is driven from the torque converter. The planet carrier (red) tries to spin the opposite direction but because of a one-way clutch system, it locks in place which forces the ring gear (blue) to turn instead. The ring gear becomes the output from the gearbox in this case and there you have first gear. The catch is that because of the design of the compound gearset, the direction of rotation of the output shaft ought to be opposite to that of the input shaft, but it isn't. This is because the first set of planet gears engages the second set and it's the second set that turns the ring gear. Doing this reverses the direction of rotation, thus making it now the same as the input shaft.
Moving swiftly along, when second gear is engaged the input is again the small sun gear but this time the ring gear is held in place by a band and the output becomes the planet carrier.

Locking planetary gearset components

If you've got this far, congratulations, you're doing better than I did the first time I had automatics explained to me. You might now be wondering how the clutches and bands I've mentioned above actually work. Bands are literally that - they're a band wrapped around the outside of the ring gear and when tightened, they lock the ring gear in place. Bands are actuated by a lever or pivot connected to a small hydraulic piston in the gearbox housing. The image below shows how a band might work in the example I've been building up. The actuator piston actually sits in a small cylinder inside the hydraulic distributor (see later) which is built into the gearbox case. You can see the band wraps around the ring gear and when the piston is pushed down, it tightens the band and clamps the ring gear into place, locking it to the gearbox case.

The clutches are a little more complex and are used to perform functions such as locking the sun gears to the turbine or input shaft. Automatic transmission clutches are a lot like the motorbike basket clutches mentioned higher up the page. They consist of a series of pressure and friction plates with splines on the inside and outside. These are compressed by hydraulic fluid fed through channels in the various shafts to a clutch piston. Clutch springs make sure the clutch piston releases when hydraulic pressure is reduced. The example below shows how a clutch system might work to lock the ring gear to the output shaft.

The automatic gearbox hydraulic system - how it changes gears.

You've got the idea by now that hydraulics are used a lot in an automatic gearbox. They're used to pressurise the piston plate for the clutches and they're used to move the band-activation pistons up and down. In the good old days, the routing of the hydraulic fluid in the system was controlled by mechanical shift valves linked to the throttle valve on one side and the governor (see later) on the other. Those days are on the way out now and generally speaking, when you move the gear stick, you're doing nothing more than giving an input to the engine management system or engine control unit (ECU) indicating what gear you'd like to be in. The ECU then looks at engine speed, speed across the ground, current gearbox configuration and position of the gear selector and decides what the best action is. It signals solenoid shift valves inside the hydraulic system to open and close appropriately and the gearbox then changes gears as necessary.
But how does the gearbox know to go up gears when you're speeding up, and down when you're slowing down? Well there's a device called the governor attached to the output shaft of the gearbox. It's a centrifugal sensor connected into the hydraulic circuit. The faster you're going, the faster the governer spins and the more open the valve in it becomes. That in turn allows the pressure of the hydraulic circuit to rise, which then applies more pressure to different components, pistons and clutch activators and lets the gearbox shift up at the right speeds. Again, in modern cars, all this information is fed through the ECU which also takes another input from a throttle sensor or more usually a vacuum modulator. These devices allow the ECU to know how hard the engine is working - something else that's critical to how the gearbox operates. It's these inputs that can sense the sudden need for more power so that when you stuff the accelerator to the floor, the gearbox can downshift. The ECU sees a relatively sedate output shaft speed from the governor but a sudden and dramatic increase in vacuum pressure in the engine intake manifold. This is the key to dump the gearbox down a gear to get more power and quick.
Limiting gear selection. Most gearbox selectors have a '1' and '2' position. When you select one of these positions you're inhibitting the gearbox's ability to pick any gear higher than that. In a mechanical system it locks off certain portions of the hydraulic system physically so the gearbox simply cannot provide hydraulic pressure to the selector components. In a modern electronic gearbox, again you're simply telling the ECU "don't select anything higher than this". The ECU will then simply not ever send commands to open the solenoid valves to activate higher gears.
The pump. It's probably no surprise to you that all this hydraulic trickery needs some sort of pressure to work and that comes from the hydraulic pump. This is normally located in the cover of the gearbox housing itself and it draws fluid from the gearbox sump to feed the gearbox hydraulic system, the fluid cooler (basically a small radiator) and the torque converter. The pump itself is a typically a rotary displacement pump that uses the difference in pressure between the spinning centre lobe and the outer housing to suck fluid in on one side and expel it on the other.
For the uninitiated or the morbidly curious, the image below shows a highly simplified example of the rats nest of hydraulic routes in a gearbox housing. The hydraulic lines are effectively cast in the metal because doing it with rubber hoses and clamps would be so complicated and take up so much space that it would be uneconomical and unreliable to do in mass production.

If you've owned a VW or Audi in the last few years it might have come with a TipTronic® gearbox. To you, the driver, it looks like a regular automatic gearbox but with with an H-gate for the gearshift. In normal operation, you use the gearbox just like an automatic, putting it in 'D' for Drive and just letting it go about its business. But if you click the gearstick over into the H-gate it becomes a discrete automatic, meaning you can then click it fowards and backwards like a sequential gearchange. In this mode you are basically telling the gearbox when you want it to shift rather than allowing it to shift for you. When you click it forwards for example, you're indicating a desire to go up a gear. The ECU looks at the engine speed, road speed, torque and load and if all the planets align, it shifts up by activating the relevant solenoid valves in the automatic hydraulic system.
Most TipTronic® designs do have a certain amount of idiot-proofing though, and if you try to rev the tits off the engine in first, it will override you and automatically shift up to second to save the engine. These types of gearboxes often have steering-wheel shifters either as buttons or triggers on the steering wheel (like the Mazda MX-5) or paddle-shifters. TipTronic® is actually a design from Porsche and they simply license it to other vendors, typically German manufacturers. Because it was one of the first designs to come to the mass market, this type of discrete automatic automatic gearbox is now often referred to as TipTronic® even if it isn't one of the VW/Audi/Porsche ones. Here's a non-comprehensive list of some of the manufacturers and their TipTronic® type shifts:
Acura: Sequential SportShift. Audi: Tiptronic, Multitronic (CVT). BMW: Steptronic. Chrysler/Dodge: AutoStick. Citroën: Sensodrive. Ford (Australia): Sequential Sports Shift. Honda: iShift, S-matic, MultiMatic. Hyundai: Shiftronic, H-Matic. Infiniti: Manual Shift Mode. Jaguar: Bosch® Mechatronic. Lexus: E-Shift. Mazda: Sport AT. Mercedes-Benz: TouchShift. MG-Rover: Steptronic. Mitsubishi: INVECS, INVECS II, Sportronic, Tiptronic. Nissan: Tiptronic. Vauxhall/Opel: Easytronic. Peugeot: 2Tronic. Pontiac: TAPshift. Saab : Sentronic. Subaru: Sportshift (system developed and name used under license from Prodrive Ltd.). Smart : Softip. Volkswagen : Tiptronic. Volvo: Geartronic

Despite the name, these are actually an advanced type of manual gearbox. It's better to refer to them as clutchless manual gearboxes because that more accurately describes what they are. Semi-automatics do not use planetary gearsets and torque converters; they use layshafts, output shafts, clutches and selector forks just like a manual. They come in three flavours, all of which have the same internal mechanisms. Two of those use the familiar paddle-shifters or up-down gearstick for changing gears. (This begins to explain why you cannot simply look at a gearstick or paddle-shifter and tell what the gearbox is. Up/down gearsticks or paddleshifters can both control sequential manual, clutchless manual or TipTronic® type gearboxes.) The third type has a pure manual gearstick. None of the three types have a clutch pedal though so how do they work? Well in the case of the first type, when you click the gearstick up or down, or press one of the paddleshifters, a hydromechanical system disengages the clutch and then moves the gearbox selector forks into the position for the next gear before re-engaging the clutch. Because the system takes inputs from load- and torque-sensors as well as road speed, throttle position and engine demand sensors, and because it's all computer controlled, it can shift more quickly and more smoothly that you or I ever could.
The third type uses the same hydromechanical system underneath but has additional sensors coupled to the gearstick. With this type, the action of moving the gearstick out of the gate for one of the gears (for example pulling it back from first) passes a hall effect sensor which tells the clutch to disengage. When you push the gearstick into the gate for the new gear, another hall effect sensor detects the final position of the gearstick and tells the clutch to re-engage. Effectively it's identical to driving a manual car only without a clutch pedal.
Clutchless manual gearboxes have appeared under many different names such as Saxomat and Olymat (Fiat 1800, Saab 93, some BMWs and Opels).

DSG / DCT Gearboxes - what, why and how?

How does this sound? A manual gearbox that's always in two gears at the same time. Sounds impossible, right? Scroll back up to the top of the page and look at how a manual gearbox works - how can this happen? Enter stage left the dual clutch transmission (DCT) or direct-shift gearbox (DSG). Two different names for essentially the same design. The most famous / common of these currently is the DSG as fitted to the Audi TT and some of the newer VW Golfs. The DSG is licensed technology from BorgWarner, which despite sounding like a horrible accident between a Star Trek character and a large movie studio, is an automotive parts supplier known until this point for its automatic gearboxes.
The principle is really simple even if the engineering is really complex. The idea is that when you're going up through the gears, increasing in speed, one clutch has the current gear engaged and a second clutch has the next gear up pre-engaged ready to use in the blink of an eye. Technically, that's not even true because a DSG can shift gears in 8 milliseconds. At 400 milliseconds it takes you 50 times longer than that to blink. That in essence is the key benefit to the DSG - blisteringly fast gearchanges. Plus, because one clutch engages as the other one disengages, the time that the gearbox is not driven under power is minimised.
So how does this work? Well a DSG gearbox has one layshaft like a normal gearbox, but two output shafts that mesh to a third shaft which goes to the differential. One output shaft has 1st, 3rd and 5th gears on it whilst the other has 2nd, 4th and 6th. The layshaft is actually two shafts one inside the other connected to two concentric 4-plate basket-type clutches at the end. In first gear, one clutch is engaged and the central layshaft is connected to the engine. Selector forks have the first dog-gear engaged with the first helical gear and the car is moving forwards. At the same time though, on the second output shaft, the second dog gear is already engaged with the second helical gear. Because the outer clutch on the layshaft is disengaged though, there is nothing driving this second gear and the outer layshaft is simply spinning freely. At the point when the gearbox needs to shift up, it simply engages the second clutch at the same moment it disengages the first and the outer layshaft is now being driven from the engine. Because second gear was already engaged there is literally no delay in shifting so the gearchange is near instantaneous. Once in second gear, the inner layshaft is now freewheeling as the selector forks engage third gear on the first output shaft and so on and so forth.
The three images below show my typical manual gearbox example modified into a 5-speed DSG. In this first image, first gear is engaged and second gear is pre-selected. The transmission of power from the engine to the output shaft is shown with the green components. The dual clutch (shown in cutaway) has engaged the inner set of friction plates which are connected to the outer layshaft. The first dog-gear is engaged with the first helical gear.

In this second image, second gear is selected and third gear is pre-selected. Again, the transmission of power is shown with the green components. This time the dual clutch has engaged the outer set of friction plates that are connected to the inner layshaft. The second dog-gear was already engaged with the second helical gear and so is now driving the output shaft.

This final image shows a cutaway of a simplified dual-clutch, dual-layshaft so you can see how the friction plates, layshaft and gears all relate to each other. The green inner layshaft has the drive gears for second and fourth whilst the outer red layshaft has drive gears for first, third and fifth. The grey clutch housing contains all the springs and hydraulics used to engage the various clutch plates, although they're not rendered in this view.

CVT (continuously variable transmission) - what, why and how?

As they say in some circles, it's all downhill from here. Seriously. If you got your head around DSGs and automatic boxes, the rest of this page is going to be a veritable walk in the park, starting with the CVT - continuously variable transmission. CVTs are based on simplicity rather than complexity. Gone are the nightmare of spinning, whirling, intermeshing gears, cluches, clamps, bands, friction plates etc etc ad nauseum. Instead, the CVT essentially has three moving parts. No seriously. Read on.
If you live in the Netherlands, you're intimately familiar with the CVT - most brommers have a dry-belt CVT. For those unfortunate enough to have never lived there, a brommer is a small moped - typically less than 50cc in capacity. They're all two-stroke engines and they are uniquely identifiable from their sound - a constant pitched engine. No revving up and down, just a long, continous high-pitched drone, like bees on crack. It's a gorgeous sound. In fact the role that the Netherlands play in the CVT story is that dr. Hub van Doorne Invented the first CVT for automotive use in 1958. It was the Variomatic and it was used in DAF Cars Doornes Auto Factory.The first variomatic used rubber composite belts, but the durabilaty and strength of these belts just wasn't up to the job for serious car engines. This led to the development of a metal belt which allowed the normally slack side of the belt to actually push, and so was able to deal with higher torque values (up to 450Nm). The modern pushbelt is made up of hundreds of individual, specially designed steel elements, which are strung together along 2 high-alloy steel ring packs. Not so much of the rubber band any more. This product was sold by a new company by dr. Van Doorne - VDT - van Doornes Transmissie (transmission). In 1995 German multinational Robert Bosch Gmbh bought the VDT plant. In 2008, the stats for the factory were about 1100 employees producing about 2.4 million pushbelts. Nissan, Toyota, Mitsubishi, Hyundai, Jeep, Mercedes and Subaru are some of the manufacturers now employing CVTs although not all of them are buying direct from Bosch. Some buy belts produced under license by other transmission manufacturers such as JATCO, Fuji Heavy Industries (Subaru) and Punch. Japan is today's biggest market for pushbelt CVT's. Bosch have a CVT Pushbelt promo video available if you're interested.
But I digress. In 2005 CVTs really moved into the mainstream when Nissan introduced the "no shift shock" gearbox into their cars and SUVs. This followed a somewhat faltering start from Ford in 2004 who, frankly, botched the launch of their CVT so badly that barely anyone remembers it. So what the hell makes it so attractive to the automotive and motorcycling markets? Well apart from the simplicity, it has one extremely sound engineering principle : get the engine to peak torque and keep it there whilst infinitely varying the transmission. That way the engine is always performing at peak capacity. No changing gears, no revving up and down the rev range, and as Nissan so aptly put it - no shift shock.
Interesting factoid : CVT's were banned from Formula 1 in 1994 because they were making the cars too fast...

Two pulleys and a belt. It really is that simple.

So how does this magical CVT work? Simply. Very simply. The most basic CVT has two variable pulleys and either a steel-core rubber pull-belt or a steel alloy push-belt. One pulley is connected to the flywheel and the other to the gearbox output shaft. The belt loops around between the two. On simple scooter-type CVTs, the pulleys change geometry simply by rotational forces - the faster the engine pulley spins, the more it closes up and the faster the output pulley spins, the more it opens out. In automotive applications, the geometry of the pulley is governed by a hydraulic piston connected to the ECU. The pulley itself is basically a splined shaft with a pair of sliding conical wedges on it (called 'Sheaves'). The closer the wedges are together, the larger the radius 'loop' the belt has to make to get around them. The further they are apart, the smaller the radius 'loop' the belt has to make. Based on the principles established right at the top of the page when I was talking about intermeshing gears, if the flywheel pulley has a small radius and the output pulley has a large radius, then the transmission is essentially in low gear. As the car gets up to speed, the two pulleys are adjusted together so that they present an infintely changing series of radii to the belt which ends up with the flywheel pulley having the largest radius and the output pulley having the smallest. On then to the pictures. This first image shows the basic layout of a pulley-based CVT with the two sliding pulleys and the drive belt. This is the equivalent of 'low gear' - the drive pulley spins two or three times for each rotation of the output pulley. It's the equivalent of a small gear meshing with a large gear in a regular manual gearbox.

This image shows the same system in 'high gear'. The drive pulley has closed up forcing the drive belt to travel a larger radius. At the same time, the output pulley has pulled apart giving a smaller radius. The result is that for each turn of the drive pulley, the output pulley now spins two or three times. It's the equivalent of a large gear meshing with a small gear in a regular manual gearbox. The difference here is that to get from the low gear to the high gear, the infinite adjustment of the position of the pulleys basically means an infinite number of gears with no point where the drive is ever disconnected from the output. Sweet.

Toroidal CVT (Nissan Extroid)

As good as a belt-driven CVT is, the weak link is the belt. If it gets damaged in any way, the transmission becomes useless. Another solution then is the toroidal CVT which is equally as simple in operation but has parts which are less prone to wear than the belt-drive type. With a toroidal CVT, both the input and output shafts are sculpted metal discs that face each other. In between are two rollers that free-wheel on their x-axis, making contact with both discs. The position of the rollers is controlled hydraulically and they pivot in their z-axis around a common centre so that wherever they are in their rotation, the rollers always touch the discs. Because the position of contact changes on the discs, the relative rotation of each disc changes. The image below shows a toroidal CVT in low gear. The input shaft is on the left. As it spins, the rollers make contact on the surface of it in the area I've shaded red. This spins both rollers on their x-axes, and because they both touch the output disc, it is spun in turn. The contact area on the surface of the output disc scribes a much larger circle - again rendered in red. Going back to the most basic stuff you learned at the top of the page, this is the equivalent of a small gear driving a large gear - the gearbox is effectively in low gear.

For a toroidal CVT to increase the output shaft speed, both the rollers are pivotted slowly about their y-axes. As they do this, their point of contact on the input and output discs changes in an infinitely smooth, continous motion. Effectively, the radius of the path on the input disc gets larger and larger as the radius of the path on the output disc gets smaller and smaller. This creates and infinite number of gear ratios until 'top gear' is reached when the rollers are in the opposite position to where they started. Now you can see the equivalent of a large gear driving a small gear - the gearbox is effectively in high gear.

This type of infinitely adjustable toroidal CVT can deal with very high torque figures and can be stacked up end-to-end to provide other gearing options, and is essentially how the system in some of the JDM and Nissan home-market and CVT gearboxes works (think Skyline 350 GT-8). This last image shows a double-toroidal Nissan Extroid CVT configuration.

Viscous couplings

Viscous coupling aren't really a type of differential but they're worth mentioning because they're used a lot in all-wheel drive vehicles. Lower end AWD vehicles are actually mostly 2-wheel drive vehicles (see the article below for all the differences) until the front wheels begin to slip. When that happens, they become all-wheel-drive through the use of a viscous coupling. In it's most simple form, it's essentially identical to the torque converter found in an automatic gearbox. For a full description of how that works, see torque converters up above.

Hydraulic clutch couplings

Again, not really a differential, but another type of device used in AWD cars to engage the rear differential. With these types of coupling, the front and rear differentials drive hydraulic pumps - normally filled with oil. Any difference in the speed of the two pumps causes a pressure imbalance in the system that activates a clutch pack in-line to the rear differential to engage it. So again, when the front wheels spin faster than the rear (meaning slip), the clutch pack is engaged and the rear differential comes into play. These types of coupling typically also have braking and thermal overrides so that if the gearbox oil in the rear differential becomes too hot, or the car is braking, the clutch pack can be overridden and disengaged (without this, ABS-equipped vehicles would not be able to sense all four wheels correctly under braking).