Differential Information Thread

[Opie]

Viscous Coupling (Center Differential) Info

The viscous coupling is found in most Subaru all-wheel-drive vehicles. It is used to link the back wheels to the front wheels so that when one set of wheels starts to slip, torque will be transferred to the other set, for example front set to the rear set.

The viscous coupling has two sets of plates inside a sealed housing that is filled with a thick fluid. One set of plates is connected to each output shaft. Under normal conditions, both sets of plates and the viscous fluid spin at the same speed. When one set of wheels tries to spin faster, perhaps because it is slipping, the set of plates corresponding to those wheels spins faster than the other. The viscous fluid, stuck between the plates, tries to catch up with the faster disks, dragging the slower disks along. This transfers more torque to the slower moving wheels -- the wheels that are not slipping. The plates are not directly connected, the "friction" of the thick fluid between the plates is responsible for the transfer of torque and therefore, one does not directly drive the other.

When a car is turning, the difference in speed between the wheels is not as large as when one wheel is slipping. The faster the plates are spinning relative to each other, the more torque the viscous coupling attempts to transfer. The coupling does not interfere with turns because the amount of torque transferred during a turn is so small. However, this also highlights a disadvantage of the viscous coupling: No torque transfer will occur until a wheel actually starts slipping. As stated earlier, because the plates are not directly connected to one another, it is possible to get into a situation where the viscous coupling can be defeated if the fluid between the plates becomes overheated and loses its ability to create enough friction to actually turn the plates.

A simple experiment with a raw egg will help explain the behavior of the viscous coupling. If you set the egg on the kitchen table, the shell and the yolk are both stationary. If you suddenly spin the egg, the shell will be moving at a faster speed than the yolk for a second, but the yolk will quickly catch up. To prove that the yolk is spinning, once you have the egg spinning quickly stop it and then let go -- the egg will start to spin again. In this experiment, we used the friction between the shell and the yolk to apply force to the yolk, speeding it up. When we stopped the shell, that friction -- between the still-moving yolk and the shell -- applied force to the shell, causing it to speed up. In a viscous coupling, the force is applied between the fluid and the sets of plates in the same way as between the yolk and the shell.


Open Differentials

Most Subaru's utilize open front & rear differentials. The open differential always applies the same amount of torque to each wheel. There are two factors that determine how much torque can be applied to the wheels: equipment and traction. In dry conditions, when there is plenty of traction, the amount of torque applied to the wheels is limited by the engine and gearing; in a low traction situation, such as when driving on ice, the amount of torque is limited to the greatest amount that will not cause a wheel to slip under those conditions. So, even though a car may be able to produce more torque, there needs to be enough traction to transmit that torque to the ground. If you give the car more gas after the wheels start to slip, the wheels will just spin faster.

If you've ever driven on ice, you may know of a trick that makes acceleration easier: If you start out in second gear, or even third gear, instead of first, because of the gearing in the transmission you will have less torque available to the wheels. This will make it easier to accelerate without spinning the wheels.

Now what happens if one of the drive wheels has good traction, and the other one is on ice? This is where the problem with open differentials comes in.

Remember that the open differential always applies the same torque to both wheels, and the maximum amount of torque is limited to the greatest amount that will not make the wheels slip. It doesn't take much torque to make a tire slip on ice. And when the wheel with good traction is only getting the very small amount of torque that can be applied to the wheel with less traction, your car isn't going to move very much.

Another time open differentials might get you into trouble is when you are driving off-road. In your AWD Subaru with an open differential on both the front and the back, you could get stuck. If one of the front tires and one of the back tires come off the ground (or loses traction on snow or ice), they will just spin helplessly, and you won't be able to move at all.



Limited Slip Differentials

Some Subaru vehicles feature a rear Limited Slip Differential, pre-Legacy Subaru's use a clutch-type and post-Legacy Subaru's use a viscous-type (See the Viscous Coupling section to understand it's operation).

The clutch- type of LSD has all of the same components as an open differential, but it adds a spring pack and a set of clutches. Some of these have a cone clutch that is just like the synchronizers in a manual transmission.

The spring pack pushes the side gears against the clutches, which are attached to the cage. Both side gears spin with the cage when both wheels are moving at the same speed, and the clutches aren't really needed -- the only time the clutches step in is when something happens to make one wheel spin faster than the other, as in a turn. The clutches fight this behavior, wanting both wheels to go the same speed. If one wheel wants to spin faster than the other, it must first overpower the clutch. The stiffness of the springs combined with the friction of the clutch determines how much torque it takes to overpower it.

Getting back to the situation in which one drive wheel is on the ice and the other one has good traction: With this limited slip differential, even though the wheel on the ice is not able to transmit much torque to the ground, the other wheel will still get the torque it needs to move. The torque supplied to the wheel not on the ice is equal to the amount of torque it takes to overpower the clutches. The result is that you can move forward, although still not with the full power of your car.

On an AWD vehicle such as a Subaru, the amount of torque that gets transferred to the rear LSD first must pass through the viscous coupling and then through the LSD. In a situation where both front drive wheels and one rear drive wheel are on the ice and the other rear wheel has good traction: With this limited slip differential, the wheels on the ice will not be able to transmit much torque to the ground. The last wheel will still get some torque to try and move the vehicle. However, that torque will have been split between the free-spinning open front differential and then the remaining torque will be limited by the amount of friction the plates in the viscous coupling are able to produce before they overheat. This creates a delicate operation of not spinning the wheels so much that the fluid overheats, but still spinning them enough to build enough "friction" in the viscous coupling to get the car to move. This becomes even more complicated when you consider the open front differential is limiting (reducing) the torque availible to the viscous coupling and the rear LSD.

* Registered users of the site do not see these ads.

[Chi]

Well, I guess this just answered my questions regarding why I got stuck on that garage spiral.

This means that even if I install a front and rear Quaife LSD, my car will still be limited by the weak center diff... bummer.

Cheers!

[ciper]

nvm....

[sajohnson]

Good info Opie.

I would like to say that the following, from your last paragraph is rather disappointing:

"...the remaining torque will be limited by the amount of friction the plates in the viscous coupling are able to produce before they overheat. This creates a delicate operation of not spinning the wheels so much that the fluid overheats, but still spinning them enough to build enough "friction" in the viscous coupling to get the car to move. This becomes even more complicated when you consider the open front differential is limiting (reducing) the torque availible to the viscous coupling and the rear LSD."

While your description is acurate, 'delicate' and 'complicated' are not words most folks want to hear used to describe their vehicle's drivetrain! Most drivers have been trained to use light to moderate wheel spin to free a stuck vehicle -- 'rocking' it back and forth if necessary. What I have read here and other places indicates that we almost need special training to properly use Subaru's AWD system.

An AWD system should not only be able to provide ample torque to any wheel on demand, it should do so seamlessly and without drama. In addition, the torque supplied should be enough to move a fully loaded vehicle in any reasonable situation.

I am reminded of old farm tractors. They were RWD (of course) and many of them had TWO brake pedals -- one for the left rear wheel and one for the right. The differential was typically open, so if the driver noticed that one of the rear wheels was starting to spin, he would apply the brake on that wheel only, thus transferring torque to the opposite wheel (of course then the other wheel would have to overcome the drag of the brake). Some "4WD" SUVs use a traction control system that does essentially the same thing -- it automatically applies the brake on any spinning wheel (pretty lame actually).

Our Subarus are not tractors! I could understand if people were getting stuck after abusing their vehicles, but that doesn't seem to be the case.

Thanks again for the solid info. Hopefully it will help get everyone on the same page.

[SubaruImpreza_power]

Someone last year said if your Subie got suck was to apply a little bit of the brakes to slow the spinning wheel it should even out the Torque.

[Opie]

Thanks Ciper - I editted the post to reflect your information. If you can shed some light on how the rear viscous LSD works compared to the viscous coupling I'm sure it would be very useful.

[Orson]

It may be worth noting the difference between a viscous coupling and a clutch-based differential in practice.

In the example where you have one wheel on ice and one wheel on asphalt, with a viscous coupling, the wheel on ice would have to spin up before torque can be transferred to the asphalt wheel. With a spring-loaded clutch pack, no wheel spin would occur and your car would move forward, until you overload the clamping force, at which point the wheel on ice would spin.

There is a trade-off here, though. A spring-loaded clutch pack will naturally resist any wheel speed differences even when the wheel speed difference is small. A viscous coupling, as described, will not resist wheel speed differences until the wheel speed difference has become large. In practice, installing a spring-loaded clutch on a front axle, for example, will promote understeer since the tendency to equalize the wheel speeds is incompatible with vehicle cornering (where the outside wheel would naturally rotate faster than the inside wheel). A viscous coupling will promote understeer less since it exerts little wheel-equalizing force until the difference is large. An open differential, of course, would promote understeer the least.

It's all about trade-offs!

Some interesting info along these lines...

There is a GMC truck infomercial out now that discusses the "locking" rear diff design that is unique to their vehicles. Their rear diff is designed to lock when rear wheel slip is detected. I don't think it is electronic but they didn't say.

They did a test with their truck driving over an obstical course that caused one rear tire to come off the ground. At that point you could see the rear tire start to spin but then the rear diff locked and the truck kind of lurched forward and continued on. They ran a chevy truck with a VISCOUS rear diff over the same obstical and it just stayed there spinning the suspended tire at very high speeds and the truck didn't move forward. It was actually rocking back and forth as the spinning tire occasionally touched the surface and rocked the truck forward.

I found that test interesting because they said the Chevy had a VISCOUS rear diff. But even with very high relative rear wheel velocities it wouldn't move the truck forward even on level ground. There may have been some force holding the truck back but not much from what I could see. Our rear diff is a viscous coupling and it acts the same way. I wonder if some viscous couplings are just so weak they are practially useless. From that commercial and from my own testing it seems that is the case with the WRX rear diff and the Chevy truck. I may be missing something here but it seems that some vicous units aren't worth much at all except as a marketing feature.

From what I've seen, the torsen diff is the best choice for autocross and daily driving applications where you will rarely loose all traction at one wheel. A clutch type diff is probably best for drag racing and some off road applications. I don't know what works best for rally driving.

[sajohnson]

Tshulthise:

I haven't seen that ad yet, thanks for pointing it out.

As for what works well for rally racing, I think they touched on that in the MRT book, "Training WRX" that I quoted from in the other thread. IIRC, they said Subaru makes 3 VC units for the center diff (in addition to the one in the stock WRX): They are rated at 10, 12, and 20 kg/m. The 12 and 20 are popular for competion.

"Australian rally teams mostly use the STi viscous type...".

Of course, for most of us, the question is what works best on the street (with the possibility of some weekend racing thrown in). If money were no object, I would probably go for Torsen diffs -- at least center and rear. The front is more difficult because of the possibility of creating understeeer.

In any case, whatever method is used to limit wheelspin, it would be nice to have the ability to manually LOCK the differentials, or at least the center and rear. If they could lock automatically when required, that would be even better.

In other words, the current AWD system on the WRX would be marginally acceptable IF at least the center diff locked as claimed by SOA.

[ciper]

Orson got it right. I personally prefer the viscous unit. It works as a speed diference limiter. You are still able to spin one wheel at a greater speed than the other but there is always power being sent to the other. With a clutch based unit it acts like a locked differential until the amount of torque overcomes the clamping force of the plates. If the speed difference increases too much the amount of force sent to the other wheel decreases to very little.

I think my ideal setup would be a VTD center differential with a torsen rear and a visous front.

[sajohnson]

I had momentarily forgotten about the VTD diff.

Would it be possible to use them in all 3 differentials?

They seem like the best solution --

Proactive instead of reactive.

Diff is normally open (unlike simple clutch packs).

Can be programmed to allow any amount of wheel spin (output shaft speed differential).

When they do begin to lock up, the rate at which they progress towards full lock is variable.

They are actually capable of locking, unlike the other slip-limiting devices (VC, Torsen, fixed clutch packs).

Any reason this wouldn't be possible?

[ciper]

"Any reason this wouldn't be possible?"

Expensive and complicated are probably the only reasons. Heck, look at how cheap Subaru is now. Very few vehicles get 2 lsd let alone multiple units. Other than the STI did any other US subaru EVER get a non open front differential?

[SubaruImpreza_power]

So does the 4EAt have this problem?

[sajohnson]

SubaruImpreza_power:

My guess would be no -- at least not when the 2 front or rear tires are on a slick surface. The VTD center diff should lock and essentially the vehicle is then in '4WD High' -- at least one front and one rear tire will have torque applied.

However, when the 4EAT WRX is in a situation where either the left or right 2 tires are on ice, then it might still just sit and spin, since the rear VC unit is all but worthless.

[SubaruImpreza_power]

So if your car does not have a LSD your stuck?

[sajohnson]

It depends on the situation. A VC diff is a type of LSD. Apparently though, most of them do not really lock (as SOA claims).

So, if you find yourself in a situation where only one or two tires have traction, you would be better off with VTD-type diffs -- or at least differentials that can be manually locked.

Keep in mind that even with all open diffs, AWD is better than FWD or RWD. For most people, in most situations, it would be adequate. It is only in those (typically rare) situations where one or more tires have significantly less traction than the others that the 'limited slip' systems come into play.

It is kind of like brakes or tires (although admittedly not a perfect analogy). Most of the time, most of us could run tires that are worn down to the 'wear bars' or below without any trouble. In fact, they might actually perform better than new tires -- as long as it doesn't rain. If you drive mostly on Interstates you very rarely need brakes either!

[Chi]

I used to have a Jeep Grand Cherokee with Quadradrive. Looks to be a front/center/rear VTD system.

Here's the link:
Jeep Grand Cherokee

Click on the Quadradrive link on the right menu.

Wonder if that system would work on the scooby.

[Orson]

Originally posted by Tshulthise
...
There is a GMC truck infomercial out now that discusses the "locking" rear diff design that is unique to their vehicles. Their rear diff is designed to lock when rear wheel slip is detected. I don't think it is electronic but they didn't say.

They did a test with their truck driving over an obstical course that caused one rear tire to come off the ground. At that point you could see the rear tire start to spin but then the rear diff locked and the truck kind of lurched forward and continued on. ....

From what I've seen, the torsen diff is the best choice for autocross and daily driving applications where you will rarely loose all traction at one wheel. A clutch type diff is probably best for drag racing and some off road applications. I don't know what works best for rally driving.

Hard to say what was in the GMC. There are so many LSD designs that haven't even been touched upon here. (One type that is increasing in popularity is the pump-based clutch LSD.)

A lot of racers do not like Torsen's. The torque-bias curve (the point at which you get full LSD versus the point at which the LSD can not transfer any more torque) can be somewhat abrupt. It's one reason why you do not see Torsen's on cost-is-no-object cars like Ferraris or even merely-expensive cars like Porsches.

[Orson]

Originally posted by Chi
I used to have a Jeep Grand Cherokee with Quadradrive. Looks to be a front/center/rear VTD system.

Here's the link:
Jeep Grand Cherokee

Click on the Quadradrive link on the right menu.

Wonder if that system would work on the scooby.

The Quadradrive uses a pump-based clutch system. The trademark name in this case is "Gerotor"

From http://www.trucktrend.com/features/t...x4/index1.html

In the Jeep Quadra-Drive system, a gerotor limited-slip is used in both front and rear axles.

A pump-based LSD like the Gerotor uses the wheel speed difference to drive a pump. The pump is in turn used to pressurize fluid that clamps together two clutch plates. Gerotors can be tuned to be extremely sensitive, but have their share of problems, such as abruptness of engagement. But the pump is regulated by a computer and it has been campaigned in one rally car (I believe group A).

[Orson]

Originally posted by sajohnson
In any case, whatever method is used to limit wheelspin, it would be nice to have the ability to manually LOCK the differentials, or at least the center and rear. If they could lock automatically when required, that would be even better.

You don't want a Subaru - you want a Ford F-150.

Orson... what do Porshe and Ferrari use for diffs? There's a Porsche 966 AWD in our region and I was watching him start the other day and he seemed to get more rear wheel spin than front, which is really perfect. He claims that the AWD works great and the car handles nearly perfectly. I usually beat him in my WRX so I'm not sure he has that good of a feel for his car to know how well everything is working but just watching him launch made me wonder what kind of diffs they use. I would imagine Porshe and Ferrari would use the best diffs for road racing. Any info on that?

BTW, thanks for the other new information. I think I agree with you that I wouldn't want a locking diff. In the commercial I referenced the GMC rear diff did engage pretty abruptly. It definately wasn't linear.

[Orson]

Originally posted by Tshulthise
Orson... what do Porshe and Ferrari use for diffs?

I'm not sure. Information is tough to find. I only know they don't use Torsen's because I have sat in presentations from Torsen people where they admitted that they had no "sports car" applications. (They talked about the torque bias requirements being different for sports car and family cars.) I also know they don't use Gerotors for the same reason.

For a long time, Porsche used viscous units, including their first AWD models. I looked up Edmunds and it says VC for a 2003 model: http://www.edmunds.com/new/2003/pors...av..7.Porsche*

But I don't know how reliable that information is. I've read some sources on the internet that say recent Porsche's also use ABD (a form of limited traction control using ABS). I doubt that they use ABD exclusively since ABD can be very unpredictable. Perhaps a combination of the two or perhaps with computer controlled clutches.

The legendary 959 had computer controlled clutches everywhere.

The only thing I have read about Ferrari LSD's (at least on recent models) is that they vary the torque bias based on whether the car is accelerating (25%) or braking (40%). This kind of behavior would suggest a computer controlled clutch since I don't know of any mechanical unit that can accomplish that.

A cost-no-object set-up would definitely be a computer controlled clutch - almost any engagement characteristic could be created. Nearly every WRC car uses computer controlled clutches, to varying degrees of sophistication. A few viscous units remain in a select few applications and there are supposed to be plenty of viscous still in use in group A. I have yet to hear of someone using a Torsen (or similar) in rallying, even though the durability of Torsen type units would appear to be an advantage.

[sajohnson]

A couple questions:

Doesn't Subaru's VTD AWD system also use comuter-controlled clutches?

While I am not interested in driving around in a F-150, wouldn't it be possible to have one or more locking mechanisms on just about any type of AWD system?

In other words, if Torsen and/or Gerotors don't work well for street/autocross/roadracing, and the current VC-based AWD system in the WRX doesn't work particularly well in slick/mild off-road conditions -- why not have the best AWD setup for 'normal' conditions, but provide some way for the driver to lock one or more diffs temporarily?

I've used this example before, but we have a '97 Toyota RAV4 with AWD. I believe the rear diff is a Torsen (or similar), and the front is open. The center diff is normally open too (IIRC) -- but there is a switch on the dash that enables the driver to lock the center diff when conditions call for it (or if you're stuck). This isn't some monster off-road truck or SUV. It doesn't even have a low range. Like Subarus, it is meant primarily for use on paved roads (or graded dirt/gravel).

SA... I think your statement that Torsen diffs don't work well for autocross or road racing is incorrect. Gary Sheehan very successfully road races a WRX that has Torsen diffs front and rear and an open Cusco center diff. Quaife Torsen diffs are very popular with the autocross crowd also. I've heard enough good things about them that I bought one. The STi also uses a Torsen diff up front.

Concerning a electronic locking diff.... my guess is that its purely a cost issue. Reliability/longevity may play into it also. SUV's are more likely to be used in situations that require a diff lock. I don't know about other people but I wouldn't be willing to pay more than an extra $150 or so for that feature and I'm sure it would cost much more than that. The stiffer STi center diff and Torsen front diff is the ultimate purely mechanical setup I can think of. Having 3 computer controlled diffs would be great but I wouldn't pay more than $1500 or so extra for that over the STi setup either.

Another thought is that if you have a Torsen diff and one wheel is in the air or slipping you will be stuck. But, if you apply the brakes lightly then you will get the other side turning. Since a Torsen diff sends some multiple of the slipping wheel's torque to the other side (lets say 5x, I got this from www.howstuffworks.com) then applying the brake would cause torque to be sent to both sides. So, in essence you would have your locking diff without the cost or reliability issues associated with a electro-mechanical system. Something to think about.

[Orson]

Originally posted by Tshulthise
SA... I think your statement that Torsen diffs don't work well for autocross or road racing is incorrect. Gary Sheehan very successfully road races a WRX that has Torsen diffs front and rear and an open Cusco center diff. Quaife Torsen diffs are very popular with the autocross crowd also. I've heard enough good things about them that I bought one. The STi also uses a Torsen diff up front.

A few things to straighten out. "Torsen" is a trade-mark, not an engineering term. So there is no such thing as a Quaife Torsen because Quaife is also a trademark. Quaife is also purely mechanical like Torsen, but works on a different principle. I do not understand it much, so maybe someone else can help.

The STi does not use a Torsen. It uses a "SureTrac" which also operates on a different priniple from Torsen or Quaife. (I'm not sure of the spelling, but there are two different ones out there with similar spellings and they are also different from each other!)

Whew! All the different designs just goes to show - every design has a trade-off that someone else is trying to get around.

To say that Torsen diff's do not work well for road racing is merely a generalization. Many people in race application appear to not like them. Doesn't mean that ALL people don't like them! But it is true that in huge budget racing (F1, WRC), you tend not to see Torsen's.