autocross.dsm.org getting updated

[DG-FNR]

I've been going through and doing some updates to http://autocross.dsm.org There's a new section on differential tech (which is almost complete) and over the next little while more will get updated and added.

It might be worth keeping an eye on.

DG

Learn: http://streetmodified.org/books.html

 

[Gaan]

Good reading...thanks for the updates.

rt

 

[HokieDSM]

That's great. Any plans to do some updating to the 1G tech section? It's not seen an update in a LONG time.

You need to change your sig though - there's no text allowed besides your name.

 

[cait sith]

See if you can get any input from Charles on the Cusco Tarmac 35-65 unit on the diff talk section.

 

[DG-FNR]

I see Charles all the time. It'll be in there.

DG

Learn: http://streetmodified.org/books.html

 

[cait sith]

Excellent, I look forward to hearing about it.

 

[DG-FNR]

More updates in.

DG

Learn: http://streetmodified.org/books.html

 

[jtmcinder]

DG -

I don't believe that your short description of how a Quaife works is correct. The key to these is that helical gears produce a force at a right angle to the force being transferred and this side force is what is used to lock the diff. It has nothing to do with the gears only "wanting" to turn in one direction; rather, the reason why these work in one direction is because, when the wheels are turning the engine, the side force is in the wrong direction.

Also, there are diffs that use input torque to lock the diff (in contrast to Quaifes, which use the difference in output torques to producing locking, and in contrast to VCs, which use the difference in output speeds). These really need to be included as an option, especially since I believe that this is what ACM runs in the rear of his car, which could easily be argued to be best 2G driveline in autocrossing.

Otherwise, very nice write-up.

- Jtoby

 

[DG-FNR]

Quaife disagrees.

If you can find a toy called Capsela (or something like that) which is a Tinkertoy-esque modular powertrain construction toy, there's a worm gear module in there that displays this property.

To the best of my knowledge, Charles is running a clutchpack in his rear slot. Clutchpacks produce locking force as a function of the mu of the friction surfaces, the amount of spring preload, and the differences in axle speeds across the differential.

As is stated in the writeup, I haven't driven Charles' car yet so I cannot report on its handling characteristics. But I would expect the clutchpack rear to increase turn-in understeer compared to a viscous rear. Given that Charles has the rear-biased centre, I expect that the rear diff has a larger effect on handling on Charles' car than on a DSM with a 50/50 centre, and the direction that it influences power-on handling is indeterminant, because it depends on how tightly it is set up.

The primary "flaw" in Charles' setup is that the centre slot locking mechanism is still the OEM viscous, which means it suffers from all the problems associated with the viscous. This is helped somewhat by the fact that Charles doesn't have a whole lot of power, and probably by the fact that the rearward bias on the centre diff should limit how bad the front end gets when the viscous is "underlocked" for the amount of power being sent to it.

If that centre diff had a Torsen locking mechanism, I'd have one in a week. As it is, I think the superior locking mechanism of the Quaife trumps the rearward torque bias of Charles' diff, especially on a car with a lot of power.

A car with Quaifes in the front and centre slot will understeer under power, but the amount of differential-induced understeer (as opposed to weight transfer distribution induced understeer, or camber understeer, etc etc) should be minimised, and it puts power down MUCH more effectively. Certainly going from a viscous/open setup to Quaife/Quaife reduced power-on understeer and dramatically increased corner exit speeds.

The understeer is there, but it's not bad. Turn earlier.

DG

Learn: http://streetmodified.org/books.html

 

[jtmcinder]

Quaife disagrees.

No, Quaife doesn't, since my description came from them. But if you couldn't manage that white paper that was circulated, maybe the following will help you. It also makes it clear that Quaifes capitalize on the side-force generated by a helical gear to generate the locking force.

Learn: http://www.sportcompactcarweb.com/projectcars/0205scc_focus/

- Jtoby

 

[jtmcinder]

To the best of my knowledge, Charles is running a clutchpack in his rear slot. Clutchpacks produce locking force as a function of the mu of the friction surfaces, the amount of spring preload, and the differences in axle speeds across the differential.

Actually, Charles has a Kaaz in the rear and a Kaaz does not get its locking force from a difference in output shaft speeds. Instead, what forces the sidegears against the clutchpack is the input torque wedging the pinion gears in between the sidegears, forcing them outwards.

Neither a Quaife nor a Kaaz uses a difference in output shaft speeds to generate locking force. They are not fancy VCs. They are totally different from VCs.

Learn: http://www.kaazusa.com/technology.asp

- Jtoby

 

[DG-FNR]

From the link you just posted, it's clear that the Kaaz is a clutchpack, or "Salisbury" differential. As such, the locking mechanism is the friction in between the clutches. In fact, the Kaaz clutches have different friction surfaces on each side of each plate so that you can change the effective mu of the clutch stack by placing rough against rough or smooth against smooth in various patterns.

Because a clutchpack is an open diff with a slip limiter, the "gear part" of the diff tries to send all the power to the wheel with the highest speed - just like an open diff - and the clutches try and fight this behaviour by dragging the slow wheel along with it. The amount of power transferred depends on the amount of force required to first overcome the friction in the clutches (before this occurs, the diff is locked and both wheels must travel at the same speed) and once the clutches start slipping, the amount of dynamic friction between the plates, which is in turn a function of the relative speed of the plates, which is in turn a function of relative wheel speed.

Input torque only matters so far that it must exceed the amount of preload required to get the plates sliding and unlock the diff. If the amount of friction across the plates is constant with respect to the relative velocity of the plates as they slide against each other (which might be the case, but need not be) then any excess torque is going to be routed to the fastest wheel.

This is the CLASSIC case of a clutchpack wearing out - you get more and more outside wheelspin as less and less torque is sent to the inside wheel as the clutches produces less and less friction as they wear out. Conversely, a clutchpack that is VERY tight (lots of preload) will chirp and skip the inside wheel at low vehicle speeds and/or tight turns (high differential wheel speeds across the axle) because the input torque is not enough to break the clutches loose with respect to each other but IS high enough to break the (inside) tire loose with respect to the ground.

A viscous coupling is just another form of clutchpack, except that the clutch mechanism is highly nonlinear with regards to differential speed (where a clutchpack is more likely to be linear or nearly so) plus there is no preload to overcome, plus there is no temperature-sensitive "hump mode" in a clutchpack like there is in a viscous. But otherwise the power-transfer mechanism is exactly the same.

DG

Learn: http://streetmodified.org/books.html

 

[jtmcinder]

No. No. No. Your claim that all clutchpack diffs use relative output speeds to create additional locking force is simply not true. Your claim that a Kaaz is just a fancy VC is simply not correct.

One way that you can start to see the difference between a Kaaz and a stock VC is in how and what causes the locking force to change.

In a stock VC, slippage [by which I mean non-stationary internals to the diff, not actual tire slippage] causes heating of a special fluid and this alters the grip between the plates. The more it slips, the more it heats up, and the more it fights the slippage.

In a Kaaz, you start with X amount of grip between the plates (fighting slip) [preload on the clutchpack] but what causes the grip to increase (and, therefore, the diff to fight slip more and more) is not the actual slippage. Nor is it any kind of heating. What increases the grip between the plates and adds more locking is adding more normal force on the clutchpack (squeezing the plates together). And the source of that additional normal force is the pinions wedging between the sidegears and forcing the sidegears back (outwards) against the plates.

[edit: in case you never make it all the way through this thread, that last sentence is wrong. The additional normal force squeezing the clutchpack does come from the pinions forcing the sidegears against the clutchpack; rather, the additional normal force comes from a pressure ring that is force against the clutchpack when the pinions are forced outwards by the transmitted torque. (How or why I thought that transmitted force could make the pinions wedge inwards will be left to me and my therapist, since it's rather silly and embarrassing.) Please make the similar adjustment when reading the next paragraph.]

This is why and how a Kaaz can be made to lock more in one direction than the other (i.e., how you make a 1.5- or 1-way, instead of a 2-way), when this cannot be done with a VC. You set the interface between the pinions and sidegears to either only create outward force on the sidegears in one direction (1-way) or you set it such that more outward force is created in one direction than the other (1.5-way).

- Jtoby

 

[DG-FNR]

I took a closer look at the Kazz drawing, and I see the mechanism you are talking about. There's a cam or wedge or something there that adds extra preload to the clutchpack.

I also didn't see the text that goes along with the pictures, because it renders black in Netscape - you have to highlight it to see it.

So OK, whenever the differential is differentiating, it varies the amount of preload on the clutch stack. Clever.

BUT

This is STILL a function of the differential in WHEEL SPEED, NOT input torque!

Consider the case of driving in a straight line with great big sticky tires. Both wheels are turning at the same speed. The pinion gears are not turning relative to each other, so there is nothing increasing the preload on the stack - the clutches are not sliding either. We can vary input torque to our hearts content and the action of the differential will not change.

As we get a differential in wheel speed, we get more slip across the clutch pack and (in this design) more preload applied to the clutches. That is effectively the same thing as a viscous coupling. Different force curves,but the same principle.

DG

Learn: http://streetmodified.org/books.html

 

[jtmcinder]

Well, I'm glad you're starting to get it (although I'd appreciate it if you'd admit that you had the Quaife all wrong, since I've gone to the trouble of explaining it).

To return the favor, I'll admit that the pinions don't actually push the sidegears against the clutchpack, but the shifts in what Kaaz calls the pressure rings alter the squeezing of the plates.

But the key point that I'm trying to correct remains the same. It is not the mere rotation of the pinions that causes the increase in locking force. These are nothing like VCs. Instead, what causes the increase in the locking force is the torque between transmitted between the pinions and the sidegears. This torque is equal to the sum of the torques between each of the output shafts and the input shaft. So, under power, it is input torque (coming from the engine) that locks a Kaaz, not the relative speeds of the output shafts.

- Jtoby

 

[jtmcinder]

Since I know that wasn't very clear, let me try again.

The key to understanding how a Kaaz operates starts by noting that the pinions can move inwards and outwards on their shafts. I think that we're now on the same page on this. The question, therefore, is what causes the pinions to move in and out.

There is nothing special about rotation that would cause a pinion to move in either direction. Spin a gear on a shaft and it doesn't magically start moving one way or the other on the shaft. That's why your claim that it's mere rotation that does it is wrong.

But mesh that gear (which is beveled <- that's key) with another and use one gear to transmit torque to the other and you get some movement.

OK, fine, so you got movement. How does that add more locking force? Well, the backs of the pinions are curved. So, when they push against the pressure rings (because they are receiving or transmitting torque), they force the left-side pressure ring to squeeze the left-side clutchpack and the right-side pressure ring to squeeze the right-side clutchpack. That's where the extra locking force comes from.

But it all starts with the pinions being forced outwards between the pressure rings by the torque that they are either giving to the sidegears or are getting from the sidegears.

(I'm sorry that I thought and said that the pinions forced the sidegears against the clutchpack. That was really, really silly, given how the gears are beveled. If my claim that the pinions moved in was what was causing your grief, that's my fault entirely.)

(With that said, if you could design a gear that caused the inverted tangential force that I thought was being used, then my design would kick butt, since it eliminates the middleman [i.e., the pressure rings].)

- Jtoby

 

[DG-FNR]

Drive in a straight line, well below the limit of the tires (just to eleiminate outside forces)

No differential wheel speed across the axle. Accordingly, no movement of the clutches relative to each other. No "slip" to limit. Right?

Keeping constant throttle - so NO variation in input torque. Assume tires sticky enough to not break loose relative to the ground. Enter a constant radius turn. Ouside wheel rotates faster than the inside - it MUST, as it is following a longer arc and there is no tire slip.

There now MUST be slip in the diff. The limited-slip fuction is now engaged. Friction is transmitting some amount of torque across the limited-slip mechanism. Input torque has not changed. Total output torque has not changed (save whatever energy is lost as heat in the clutches) Torque distribution HAS changed. What else changed? Differential wheel speed.

DG

Lern: http://streetmodified.org/books.html

 

[DG-FNR]

Because it appears we are (*sigh* once again) deep into "Jtoby automatic gainsay mode", I'll let some other people speak for me.

Racecar Engineering Volume 14, Number 12, article "Differntial Equations", page 44

It can be seen that the maximum force the differential transfers is dependant on the speed difference between the two wheels. Also, the differential can only transfer torque from the faster turning wheel to the slower turning wheel, which is the basic disadvantage to the limited slip differential.

Same article has a screenshot of the graph used to program differential force into Bosch Motorsports laptime simulatior. It has two dependant variables, and they are differential wheel speed and tire traction torque.

As far as the Quaife function, Carrol Smith Engineer to Win page 228

The Torsen diff is based on the simple fact that a worm-and-wheel gear setup can transmit torque in one direction only.

DG

Learn: http://streetmodified.org/books.html

 

[jtmcinder]

Dennis -

My posts are only half for you; they are also for everyone else. The only reason why you merit even half of my attention is that you have a tendency to post and author websites that use strong, confident words, but are actually wrong, as in this case.

Your description of how a Quaife works was flat out wrong. The increase in locking force comes from the side-forces generated when torque is transmitted across a pair of helical gears, as is very clearly explained on the page to which I linked. I really don't care what you have found in a Carroll Smith book. Any over-simplified text is immediately trumped by what the manufacturer says and what is patently obvious when you look at the actual differential. When the side-forces on the two helicals in the middle are equal, there's no locking force; as the side-forces become unequal, the helicals are forced to one side or the other and grind against the endwall and produce a locking force. Very straight-forward, but it requires that you aknowledge that a helical gearset produces a side-force, which is very clearly laid out on the page to which I linked if you don't already know this.

Same goes for a Kaaz. You keep babbling on and on about what happens when a car switches from going straight ahead to turning. That is entirely irrelevant to the question of what increases the amount of locking force. According to your description, the plates inside a Kaaz are never squeezed by anything more than the preload, when that is well-known to be false. The reason why it is called "preload" and not simply "load" is that the preload is increased under certain conditions. The question is what the conditions are. And this is where you have it totally wrong.

You have repeatedly said (with rather little explanation or evidence) that you get increases in locking force when the differential is operating as a differential. That is not true. Yes, you get increases in the amount of torque the differential's locking mechanism produces (because the amount of torque is proportional to the rotation speed of the spider-gears (aka internal pinions) times the current clamping force), but, again, that misses the whole point of a Kaaz, making it some silly little posi-traction device with a constant clamping force.

A Kaaz increases the clamping force on the plates when torque is being transmitted between the input shaft and the output shafts. This is done by using some of this torque to force the pressure rings down on each of the two clutchpacks. Mere rotation of the spider-gears will not cause the cups to press down harder (which is why a Kaaz does not increase the locking force above the preload when you coast around a corner, which is why they are so much better than a VC). There has to be torque being transferred to raise the clamping force above the preload.

As yesterday, my understanding of this is evolving, but that has only made it more and more clear that my basic point is correct - that it is transmitted torque that creates the extra clamping force - and that your claim that a Kaaz is just a fancy VC is wrong. While I yesterday believed that the bevel between the spider-gears and the side-gears is what produced the extra clamping force, ACM has just explained to me that it is actually the shafts of the spider wedging against the V-shaped grooves in the pressure rings that does it. This makes perfect sense (in retrospect), because it suddenly makes it clear how a Kaaz can be designed to be 2-way, 1.5-way, or 1-way. On a 2-way Kaaz, the two sides of the V-shaped cuts are symmetrical. On a 1.5-way, one side is steeper. On a 1-way, one side is flat, so that no wedging force is produced.

You can get personal all you want, referring to past arguments that we have had. That I am one of the few people who doesn't think that it is worth staying on your "good side" for the little tidbits that you might throw us is something that I thought about for a while before deciding that it was better than just being another "DG toady." If you listened to me and, thereby, made it clear to everyone that people could argue with you without fear of being banned from your private discussion groups, then you'd probably learn a lot more, a lot more quickly. Maybe then you could get within sniffing range of a trophy at Nationals in a real class (instead of a first-year supplemental class), as did ACM this year. But that's your business. It becomes my business when you author websites that are flatly wrong and then come on here and advertise such. Get it right before you start acting like an expert by writing a webpage on differentials. And one of the best ways to get it right before you write is to float your ideas on lists like this one, taking the feedback as helpful criticism, and admitting what you had wrong (when that happens) so that people who have followed the entire thread can end up with a real understanding, instead of just soundbytes from another worthless website.

In summary and keeping this thread on topic: learn, discuss, and then type the HTML code.

- Jtoby

ps. after you have updated your understanding of how the various diffs work, you really need to go back and rework the section on why certain diffs cause understeer and/or oversteer, since a better understanding of the diffs - most of all: what causes them to increase their lock - plays a huge role

 

[jtmcinder]

As far as the Quaife function, Carrol Smith Engineer to Win page 228

Quote:
The Torsen diff is based on the simple fact that a worm-and-wheel gear setup can transmit torque in one direction only.

I'm starting to understand why you get yourself into so much trouble. You see a quote like this one from Carroll Smith and automatically assume that he must have meant that the transmitted torque's one and only one direction must be the direction that you, DG, think it is. But it's not. The one direction is the tangent to the crest of the gear teeth. And since helical gears have their teeth angled, the transmitted torque is not all exactly across the interface between the two gears. Some of it is side-force, instead, which is exactly how a Quaife gets the side-forces it needs to work.

In other words, your quote is correct (of course ... after all, this is Carroll Smith!), but you failed to understand what it meant. It is exactly the same as what Quaife says and it is exactly the same as on the site I linked to.

Let go. Open your mind. Most of all: consider for a moment - as a goofy experiment - that you might be wrong and I might be right. Explore where that takes you. Learn.

- Jtoby

 

[DG-FNR]

Ladies and gentlemen, if you've ever wondered why it is I don't contribute anywhere near as much as I did back in the day of the Talon Digest, here is the example. My own, personal, hater troll, who follows me around from place to place gainsaying everything I say, who pretends to be reasonable, but is either malignly and diabolically motivated or just dense to mythic proportions.

There is no win for me here. "The problem with wrestling with a pig is that you both get dirty - and the pig likes it"

Luckily, I am stubborn enough to want to continue to contribute. Lord knows there was so little good information when I first started out, and every scrap of decent data to fall my way was like manna from heaven. Now that I am able to do racing engineering full time, and have access to resources well beyond what I ever dreamed possible, plus the accumulated knowledge from actively racing in the top level autocross racing series (mostly in a DSM) for the last 6 years, I like to pass that knowledge along. It shouldn't stay locked up inside my head; it wants to be free.

"Knowledge shared is knowledge squared" I firmly believe that. I wish that attitude was more prevelent amongst my peers, and I like to lead by example.

I also do a TON of research, and I know a lot of people in the racing industry. I will be talking to the fine folks at Quaife face-to-face in a couple of days, for example. It's very clear from the literature and from writups in other technical publications that the helical "Torsen" diff relies on the principle of a worm gear being a one-way torque transmitter, and by varying the helix angle they can vary the amount of how "one way" the interface between the helix and its meshing gear is. By having a couple of these interoperate, they can vary the amount of lockup provided. I don't understand the exact power path through all the various gears and the exact mechanism by which it progressively locks/unlocks - but I don't need to either, as I have the basic jist of it and I understand its operation very well from an OPERATIONAL standpoint.

As far as clutchpacks go, these I understand much more intimately, as I have played with a number of different models (in other cars) over the years and so have hands-on experience - plus I can find references in racing literature.

The basic nature of a differential is that it routes power to the wheel that is turning the fastest. It has no choice; that's how the mechanism works. In order to oppose this, in order to send more power to the slower-turning wheel, something must slow down the faster wheel. If you can hook this braking device to the other axle and so get extra effect, hey, that's great - but don't need to. It could be done with the brakes, assuming that the brakes could be biased side to side on the same axle.

As it happens, I have real-world, hands-on experience with exactly such a setup. I spent part of my military career as a military mechanic. One of the vehicles I worked with on a daily basis was the M113 APC. This has a motor that drives an open differential, with each axle driving a drive sprocket on each track. On each axle is a drum brake, and the actuator for the brake is attached to a lever in the cockpit - two brakes, two levers. Pull back on the lever, and that axle slows down. When the axle slows down, the differential sends more power to the opposite side track, and the vehicle turns.

This is not some theoretical thing I read in a book somewhere; I held this in my hands.

Now I did some more research on clutchpack diffs last night, and it turns out that a mechanism for increasing the preload on the clutchpack is more common than I knew of. The Kaaz mechanism is one of many - there's more than one way to do it. And I will agree that these mechanisms push against the tractive torque of each wheel in order to increase preload on the clutch stack.

That's clever. That means - assuming that the ramp angle is "correct" and the friction characteristics of the plates are properly matched - that a properly-tuned clutchpack should work every bit as well as a Quaife - perhaps even better, given that the clutchpack doesn't turn into an open diff when one wheel leaves the ground. It is, however, going to take a lot of testing to proprly configure that diff and once you start running it its performance is going to start deteriorating as the plates wear.

But the amount of pressure on the clutch stack DOES NOT determine the amount of torque distribution provided by the differential mechanism. That is totally, wholely, and solely determined by the amount of differential wheel speed and the tractive force of the tires. The tires are either hooked up, or they are not. If they are not (by which I mean the fastest turning tire is not hooked up, as the nature of the differential is to send more power to the fastest wheel) then either the slip-limiting mechanism is either capable of slowing the fastest wheel down to the point at which it regrips (which MUST send power to the slower-turning wheel because of the essential nature of the mechanism) or it is not, and the outside wheel lights up - at which point either the limited-slip mechanism either limits how fast the gripless wheel is turning (which sends some portion of power to the slower wheel) or it does not (and 100% of the power goes up in smoke)

If the slip-limiting force is dependant on input torque, bulley for it - it probably gives the slip-limiting mechanism a better chance to slow the spinning wheel down enough to the point where it regrips. BUT THAT DOES NOT CHANGE THE FACT THAT THE ULTIMATE POWER DELIVERY OF THE DIFFERENTIAL IS DEPENDANT ON THE RELATIVE SPEEDS OF THE WHEELS.

I cannot possibly state this any clearer. If JToby continues to claim otherwise... well, that's his loss. Hopefully the rest of you understand.

As far as being a "DG Toady" - there is no such thing. I have peers. I occasionally have students, and it is great when they graduate to also become peers. I am very proud of Charles Moss, who has gone his own way on a few things, and is starting to have some success. It's great to see hard work pay off for somebody, especially when it takes the intestinal fortitude to take your lumps running against the best. It is not at all easy to do one's time as a backmarker while you learn the ropes - many so-called "experts" never leave their nice comfortable sandbox.

As far as I am concerned, there is nothing sadder than someone who feels it necessary to fling mud at somebody who has had some success, especially when he is totally lacking in such success himself.

I am told Mr Markoff has some ability dealing with rats in Skinner boxes or some such. Good for him; my study of such subjects was limited to a couple of courses in college, and he is undoubtably much more of an expert on such things than I am. I would not dream of attempting to take him on in such subjects. What a pity I cannot get the same courtesy.

DG

Learn: http://streetmodified.org/books.html

 

[jtmcinder]

If the slip-limiting force is dependant on input torque, bulley for it - it probably gives the slip-limiting mechanism a better chance to slow the spinning wheel down enough to the point where it regrips.

The first sentence is all that really matters. You finally agree with what I've been saying, contradicting what you've said before, so, as long as someone else doesn't come along and disagree with both of us, the discussion is over. You're welcome for the lesson.

.

BUT THAT DOES NOT CHANGE THE FACT THAT THE ULTIMATE POWER DELIVERY OF THE DIFFERENTIAL IS DEPENDANT ON THE RELATIVE SPEEDS OF THE WHEELS.

I cannot possibly state this any clearer. If JToby continues to claim otherwise... well, that's his loss. Hopefully the rest of you understand.

All I can say about the first (shouted) part is Duh! No-one ever argued against this idea, because it's patently obvious. If I post that "one plus one is two" does that mean that I automatically win whatever argument I'm involved in?

As to the idea that I ever said anything that contradicts it, I challenge you to find it. I have always been very careful to talk about the clamping force and how that changes with changes in input torque; I never said that the amount of bias doesn't depend on relative wheel speed. Mis-representation is a sad way to say thanks for the lesson.

I could also reply to all the personal crap, but why bother? Anyone can read the thread and make up their own mind. What matters is that you have stopped mis-describing how a Quaife works and how a Kaaz works. I never expect (nor want) you to like me.

- Jtoby

Learn: http://dictionary.com

 

[ACM]

I hate seeing you two fight. Either of you has probably forgotten more than I know about these subjects. I wish we could disagree rationally and resolve the differences logically.



On the subject of Quaifes vs submarine valves - I didn't realise there was a difference, I always thought they both operated on the same principle. Just goes to show. Never thought about it after that initial comparison and reading * To Win.

I run a Quaife front, Kaaz 1.5 way rear, and a Cusco 35:65 centre diff - which is actually an open diff, much to the surprise of everyone, not the least myself ! The Cusco require a VC to be present to operate as an LSD. Despite being an uneven torque split, the output shafts do indeed rotate at the same speed. How this is achieved I don't know, but I'm reliably informed it can be done, and the behaviour of the VC bears this out.

My choice of a Kaaz rear was to 'speed up' the locking effect, to control some wheelspin and loss of acceleration when the inside rear is unloaded. The stock VC rear takes time to react to the unloaded inside wheel, wheelspin ensues, loss of acceleration and also unnecessary heating of the tyre that's spinning - a very bad thing when running Hoosiers. That the Kaaz hasn't quite behaved itself this season is something of a mystery, I have not isolated the problem(s) as yet. When it works it's an all-around Good Thing, and a definite improvement over the VC rear. It turns that car much better under power, and helps unstick the rear off power. It require a more committed driving style to achieve this however - and timidity is punished with bucketfuls of understeer. The effect on handling changes distinctly with the locking factor. The more active friction surfaces, the more aggressively the dif locks, and the more understeer is generated - to the point that 100% locking makes the car understeer impossibly. I have settled this season on about 80%, whereas last season (pre-Cusco) I ran ~60% and that was perfect, there were no downsides at all, only positive, excellent results. With 80% I currently experience too much understeer and not enough wheelspin control - I don't have my cake and I can't eat any of it either !

The Quaife front is a mixed blessing. Being a 2 way, it added a lot of understeer off-throttle. On the other hand, when throttle is applied it quite literally yanks the car to the inside, helping the car hold a tighter line under throttle - the more throttle the tighter the line up until too much throttle of course. It's quite disconcerting initially. I would prefer a 1-way in the front, but I don't believe that's possible (so I'm told). Again, just like the Kaaz, the Quaife will punish the driver ruthlessly if driven apprehensively. The combination of these two LSDs mean that the car has to be driven aggressively, with authority; it cannot be driven timidly, it has to driven with confidence and determination, so it's not a setup for the faint-hearted. Some of this will be generated by other aspects of the setup, but a lot of it comes from the interaction of these two differentials. My choice of Quaife for the front is because I don't believe a mechanical LSD would be beneficial in the front. The typical mechanical diff tries to lock the axles together, resisting the axle rpm differential necessary to turn tightly. Whilst the locking factor could be reduced to minimise the effect, it will never go away - I believe the very function of a mechanical diff is at odds with the needs of the front wheels. It may be possible to get the Cusco 1-way mechanical front to work, but I don't see how that could be equal to the effects of a Quaife under throttle. A mechanical front LSD may well be the right choice for a gravel car though (but still a 1-way). Other than a VC front LSD, there are no other front LSDs for DSMs that I'm aware of.

And now the Cusco...This has had a profound effect on the handling. I have probably half the power DG runs with - I can't simply mash the throttle and the car rotates instantly on command, by virtue of gobs of power unsticking the rear. 250bhp doesn't do that I personally do not think a Quaife is the best possible centre diff - although it could well be the best available choice. A stock centre diff with VC routes power to the end with the least traction, just like any open dif does; once the shaft rpm differential passes a certain point, the VC acts as a brake to pull the shaft speeds closer together, but it doesn't pull the speeds back to equal, it allows a certain percentage of rpm differential between the two shafts. In the real world what that means is the as the rear starts walking around it keeps getting fed lots of power, and stays unstuck, and thus the car rotates very well. With a Quaife, the effect you feel is that when one end starts to lose grip, the torque is channeled to the other end, and the wheels that were losing grip now regain it. I don't want that to happen - I want the power to continue being funneled to the rear, I actively _want_ to overpower the rear grip. With the Cusco, in conjunction with a well-worn VC (just as with the stock diff), I continue with this effect, but now I get more of the same - I have been able to add more rear grip whilst still maintaining balance, the reduction of power transmission to the front has resulted in better front grip. I don't want the LSD'ing to happen until it's almost too late, and with this setup it (mostly) does what I want.

With one problem.

Under certain conditions, the VC heats up to the point that it turns the 35:65 split to 50:50, and the car turns into an understeering pig - the Cusco has allowed me to add a lot of grip to the rear, the 35:65 split allows me to dislodge the rear much more easily than before. When the split goes to 50:50, the front gets pushy because of the extra power, and the back won't break loose. I have not been able to identify specifically what causes this, other than certain courses, but I don't know what it is about the course design that affects this. For example, Day 1 of the SCCA Nationals (South course ?), the car was wonderfully loose, I paid no attention to the car as such, all I did was drive. Day 2 however (North I think), the car got tight half way through the first run, and just got worse and worse as the day wore on. Something about that second day's course had the VC heat up dramatically and effectively lock. Since the Nationals I've been testing, up until I bent a valve, and this does seem to be the case. So the Cusco is a mixed blessing - when it works, it's really great, but when it doesn't, it's like I'm running positive camber in the front and it's utterly miserable.

Unfortunately I do not have datalogging facilities, so some of the above is conjecture, however I believe I'm correct. A similar issue exists with the brakes, where I now believe the LSDs are negatively affecting the ABS system, but I don't have hard data to support this.

I would like to add that an awful lot of the setup on my car has come from working with Fedja Jeleskovic, and he - not me - is the primary reason the car is as good as it is. Without his efforts the car and I would have finished about 45th this year, if we'd even bothered going at all.

Dennis, feel free to use any of the above if it is useful.

Charles

 

[DG-FNR]

Charles, you sell yourself way too short. By the simple method of building a car and competing at the National level, you have already accomplished (and have more experience) a thousand times more than Professor RatBox has.

I run a Quaife front, Kaaz 1.5 way rear, and a Cusco 35:65 centre diff - which is actually an open diff, much to the surprise of everyone, not the least myself ! The Cusco require a VC to be present to operate as an LSD. Despite being an uneven torque split, the output shafts do indeed rotate at the same speed. How this is achieved I don't know, but I'm reliably informed it can be done, and the behaviour of the VC bears this out.

You're not alone in not understanding the mechanism. One of my PRI priorities is tracking down some transmission people and getting them to explain the concepts to me. I think Xtrac's booth is near ours....

My choice of a Kaaz rear was to 'speed up' the locking effect, to control some wheelspin and loss of acceleration when the inside rear is unloaded.

When do you figure you'll have the combination of power-on and an unloaded inside rear?

Ahhhhh.... you're running a lot of rear bar and rear spring, aren't you? You're transferring load off the inside rear to the outside rear when cornering because you're running a lot of rear weight transfer bias.

If you shift that bias forward, you load both the outside front and the inside rear.

Lemme put it this way - I don't see inside rear wheelspin under power, and I have a lot more power than you do. For that matter, I don't see wheelspin _at all_.

Yet.

That the Kaaz hasn't quite behaved itself this season is something of a mystery, I have not isolated the problem(s) as yet.

For sheer morbid curiosity, try pulling the rear sway bar and see what happens.

It require a more committed driving style to achieve this however

This is a characteristic of DSMs in general. Even back when I was running in Stock class, the car really responded to being taken by the scruff of the neck and shaken HARD.

The beautiful thing about my current setup is that you can get away with it. The car has definite limits, but it doesn't punish mild transgressions very much. It's a car that can really be leaned on and driven very aggressively.

It's really obvious when you jump out of my car and drive something like the company RWD supercharged Focus, that has to be snuck up on and tickled to go fast. Any transgression of the limit is immediately and painfully punished.

Another car that can really be leaned on is Marcus Merideth's Mustang. That car is really very forgiving. In fact, my generic rule of thumb is that any time you make a change that makes the car both faster and easier to drive is moving in the right direction. Joe Cheng (aka The Phantom) will tell you the same thing.

The effect on handling changes distinctly with the locking factor. The more active friction surfaces, the more aggressively the dif locks, and the more understeer is generated - to the point that 100% locking makes the car understeer impossibly.

This is as I'd expect.

And now the Cusco...This has had a profound effect on the handling. I have probably half the power DG runs with - I can't simply mash the throttle and the car rotates instantly on command, by virtue of gobs of power unsticking the rear.

This never happens on my car.

Well... almost never. If I do something to _really_ unstick the rear tires on corner entry, the amount of rearward weight transfer isn't _quite_ enough to completely restick them, and I can get power-on oversteer. It takes MONSTER trail braking to do this. The rears have to be _sliding_ before power is applied to get it to happen - and it's only become possible at all since I moved weight transfer distribution forward. With the old super-rear-biased setup, crossing this particular line was an instant and unpredictable snap-spin.

My car rotates on entry (it still has a slight rear bias to weight transfer distribution) and then when you hit the throttle the rear plants and it roars out of the turn with just a hint of understeer. It drives like the Mother of All FWDs. A FWD that actually puts down power and accelerates.

DG

Learn: http://autocross.dsm.org/books.html

 

[jtmcinder]

Dennis -

Grow up.

Here's the quote from your webpage that I argued with:

Clutchpack The clutchpack diff uses a series of friction plates and springs to provide a locking mechanism. By varying the spring force and/or the friction characteristics of the friction plates, one can change the amount of lockup provided at various differential axle speeds.

This is a tried and true solution with decades of history behind it. It works. There are, however, some characteristics of this design that limit how effective it can be at approximating the ideal lockup forces we'd like it to provide. Firstly, it tends to be very linear, in that the amount of lockup provided is a linear function of the friction/spring preload characteristics of the clutchs and springs used. As the "ideal" lockup curve is not necessarily linear, that means this diff design will, at some parts of its operating range, induce more understeer than is ideal. Secondly, the clutches and friction surfaces wear out with time, changing the slope of the power transfer curve. In order to get best performance out of the diff over its rebuild shift, it may have to be installed "tighter" than otherwise optimal with an eye towards it "loosening up" as it wears. Thirdly, one can expect a good deal of "trial and error" tuning, it which the diff is reconfigured and reinstalled with different values of locking agressiveness in the search for the best compromise setting. This is likely to be labour-intensive.

And here is what I said about it:

Also, there are diffs that use input torque to lock the diff (in contrast to Quaifes, which use the difference in output torques to producing locking, and in contrast to VCs, which use the difference in output speeds). These really need to be included as an option, especially since I believe that this is what ACM runs in the rear of his car, which could easily be argued to be best 2G driveline in autocrossing.

Otherwise, very nice write-up.

As you know (now), a Kaaz does increase the clamping force on the clutchpack as a function of input torque and this is critical to why they are so much better than a VC.

Stay on topic, avoid getting personal, and avoid bringing up things that have nothing to do with the topic, and we'll get along fine. Keep being defensive and lashing out in anger and we'll have problems.

If my title means anything to the moderators, then really think about this before calling me anything other than Toby or Dr. Mordkoff. (I didn't go to evil graduate school for six years in order to be called Mister.)

- Jtoby