One potential problem I see is your idea to weld a gear and housing. If the gear is case carburized steel, then it is not weldable unless you remove the case e.g. By hard machining). If the housing is a casting, then it is very difficult to weld, requiring special geometry, materials and process equipment.

I understand the idea of welding hardened gears and differential cases is less than ideal but limitations impose working under less than ideal conditions sometimes. I have personally tig welded one of these differentials, both input and output spider gears to make a full center spool. It welded quite nicely and it has not had any issues so far. I am also not the first to do this and many people using much cruder equipment than I used have gotten away with it as well. There are no commercial spools for either the manual or auto transmission for the center differential, there is not even any aftermarket center for the auto at all. The stock spider gear bearings fail when the viscous coupling wears out and allows excessive wheel spin, this destroys the cross shaft and gears. It's not ideal to weld these things but it has proven much stronger than any other alternative. There are at least two awd automatic cars running in the 7 sec quarter mile range using welded center differentials, a few more in the 8's and quite a few more in the 9s and 10's. Not that I am saying your points are not completely valid and that all differentials can safely be welded this way. I am simply saying that part itself I am not pioneering, it is a fairly well proven method.

There was allot of problems with even the expensive quaife torsen center differentials breaking and they had all sorts of warranty issues because of it. The physical size of the differential housing limits how strong the components can be even using very strong materials and the power these cars can make and the traction of 4 racing slicks puts the diff between a rock and a hard place. I believe this is why most aftermarket manufacturers avoid this particular diff but still sell alternatives for the front and rear. I believe welding it and putting the "differential" action elsewhere as I have suggested could solve that problem.

I've been researching the details of this again and have come across some interesting information to add.

Two other forum discussions in particular have caught my interest. The first is an FSAE forum discussion about the wiesmann locker,
http://www.fsae.com/forums/showthread.php?6339-Weismann-Differential

The second is a machinist forum regarding a bidirectional roller clutch, http://www.practicalmachinist.com/vb/general/bidirectional-roller-clutch-138620/

Firstly I would like to say I have mischaracterized these lockers, kiaser and wiesmann by describing them as sprags. The sprag is a very specific one way device, these use cam activated two way roller clutches.

Quote (Chris Wiesmann)

Back to the Locker. It is not a spague device for Pete's sake. (pun intended, my dad's name was Pete!) It is a roller clutch that locks in both directions and freewheels in both directions.

This is from the first referenced FSAE discussion, post 36 on page 4 http://www.fsae.com/forums/showthread.php?6339-Weismann-Differential/page4

The practical machinist forum references a warn automatic locking hub as just such a bidirectional roller clutch device as it seems I am looking for.



If I could get my hands on a set of heavier duty auto lockers this could be a very real possibility to look at.

Quote (Chris Weismann)

We used the Weismann Locker concept but made it only one-way drive for two main reasons. One to allow the front wheels to roll when landing on the front, thereby eliminating the back drive thru the ring gear and pinion and to the transfer case and gearbox, thus saving parts from breakage; and two, to allow the front end to disconnect and act like a Pro-2 truck and rotate quickly.

Quote (murpia)

As an aside, the most successful 4WD Group B rally car, the Peugeot 205 T16 employed a one-way coupling between the centre and front diffs at some point in it's development history.

Both from the FSAE forum. Maybe there is a good argument in a center diff application NOT to have a bidirectional function. I need to see if I can find out more about the peugoet but if they already had a similar idea as what I am proposing and implemented it using a one way clutch that is pretty good evidence it works well and that I'm probably on the right path.

The auto hub lockers do see full wheel torque after the front differential gear reduction as well so it may well be far more robust in an application before the front diff. Thought? Opinions? Lenco one way or auto locking hub 2 way?

Wow, that's the first time I ever got cross-quoted from another forum...

The idea of using a one-way sprag instead of a centre diff has a merit in motorsport I think. I'm not so sure about daily driving and the original post does not clarify.

I understand 4WD off road vehicles used to be built without centre differentials and possibly front axle differentials. When 4WD was selected at the transfer case the front axle was driven at the same speed as the rear axle. To allow for (brief) on-road driving and to prevent excess wind-up torques in the transmission either one-way or locking / unlocking front hubs were used. I'm sure there are knowledgeable people here who can correct or clarify those statements.

I also understand that the Peugeot 205 T16 at some point used a one-way sprag to drive the front axle, with no centre diff. I would have read this in a magazine in the 80s, probably Car and Car Conversions. The reason I remember is because 4WD RC cars of the time had a similar transmission arrangement. The advantages are supposed to be:

No front axle wheelspin which would be allowed by a free centre diff.
No need for heavy / complex / unreliable / slow responding centre diff torque biasing hardware.
Up to 100% torque bias to either front or rear axles as dictated by weight distribution / grip distribution.
No understeer component induced by driveline wind-up if the front axle track outside the rear*
Similar oversteer component on a trailing throttle to a RWD car, due to engine braking only being applied to the rear axle*

I have 'back to back' comparison experience from 4WD RC cars in the 80s. One way front hubs or driveshafts were available as factory options and the difference in turn-in performance was notable. Much less US with the one-way feature fitted. Which of the 2 '*' marked effects causes this, I cannot answer, nor can I answer how the effect scales to full size.

One interesting possibility also exploited by 4WD RC cars is to under-drive the one-way front axle relative to the rear. Also achievable by fitted larger diameter rear tyres (or smaller fronts). If the front rotates 5% slower than the rear then when 5% slip is achieved at the rear axle the front starts to transmit torque. Could be quite tunable for an on-throttle OS driving style. Kind of the reverse effect to the VW synchro FWD to 4WD transition.

One drawback of the one-way front axle configuration is of course reverse is RWD only.

Regards, Ian

There's an error above, I wrote: "If the front rotates 5% slower than the rear", should be "If the front rotates 5% faster than the rear..."

Don't worry, I heard what you meant not what you said. It'a a confusing concept really, from the drivetrains point of view the front is rotating 5% slower than it thinks it's driving it at, but from the wheels the gearing is trying to rotate them 5% faster than it can without any slip in the rear. Are relativistic anthropomorphisms tolerated on an engineering forum?

Thank you for seeing this and responding, I really appreciate it. I was actually looking to try and register to contact you on the fsae forum to ask if you had any more specific information about the Peugeot. I have not been able to find much more about it. I wonder if they stuck with that design or changed it and why if so.

My application is mainly a street car/ drag car. I would like to try some rally stuff eventually and maybe play with autocross, most of my cars in the past have been far to drag oriented to be much good at technical stuff. I think a properly thought out awd system could be quite good at it though. The center diff is really key here though. You can't just go plowing into corners and power on drifting out of them with much speed or precision. I will take a welded center over an open or slowly responding lsd though. You can do lots of tricks to get awd cars to try and power on oversteer but much of it comes down to the center diff characteristics and most oem systems require some amount of front wheelspin to achieve much lock so you are already generating more and more understeer before you can even start getting a better power balance to the back. Corner entry understeer to corner exit understeer. I will take a welded diff and entry under, exit over. granted most of the corner entry understeer for an oem system is not generally related to the center diff but weight distribution and front/rear roll stiffness ect. Which brings up another point about the flip side of that, aftermarket solutions to awd exit understeer. This generally means a more heavily preloaded center lsd type diff, most systems of this type will react to slip on power out corner exit proportionally to the amount of drag and requisite entry understeer caused by not allowing the front wheels to freely turn in at a different rate than the rear, the most extreme example being a spool or welded center but most effective lsd systems are shades of gray in this regard. Other than a one way clutch, I don't know if there is a way to do both well. Is there another reason why more oem or race teams don't use this system that I might be missing? I know most oems won't favor any kind of oversteer for safety reasons and most high end rally teams use electronically controlled differenetials but certainly there must still be much outside of those 2 extremes. If you have not seen this before, http://farnorthracing.com/autocross/diffs.html I believe it to be the most comprehensive explanation of the center differentials interaction with suspension tuning I've ever come across and generated most of my thinking on this subject, then Carroll Smith's thoughts on limited slips really got me to thinking even more. Ultimately I would like to try a weismann style locker in both the center and rear diff.

I had thought about the possibility of running different ratios or tire sizes front to back as well. It would not really suit drag racing because when running matched ratios the clutch system would be driving all the wheels when traveling straight, I have read of offroad racers doing the opposite of what you suggest running the front and rear locked and rotating the front slower to achieve continuous slip to match the speed of the rear, in mud or sand when you are sliding around this tends to straighten the truck out under power. I do agree the oversteer setup sounds like more fun though and would be very interesting specifically for drifting, I've been to an event with my friend who has a purpose built 240sx and you can say what you will about the sport but it certainly is allot of fun. Maybe not enough for ring and pinion swaping but maybe experimenting with front to rear tire stagger.

I feel like at least some part of the specific advantages you listed could apply to just about any motorsport where people run awd systems though, the idea just on it's own I think deserves thought and discussion from all angles.