78 replies to this topic

[jpf]

In at least one printed article (which I can't recall) and in Chris Harris' video review of the new Toyota/Subaru GT-86/BRZ/etc., there were middling remarks about the car's differential. The diff is a Torsen unit, which I had understood to be a very good type of limited slip unit, whose purely mechanical design made it intrinsically responsive. I had thought that cost, the somewhat limited bias ratio, and tendency to wear were the main complaints against an otherwise admired design.

In the video, the engineer and Harris talk about a future optional "mechanical differential" for the drifting and race crowd. What would that be? Some other kind of LSD, like a cam-and-pawl or something? Some kind of locking diff?

I'm a little confused by this, because I had been thinking that speccing the car with a torsen was actually a pretty solid and welcome choice on Toyota/Subaru's part. What gives?

[Wolf]

My main 'beef' with Torsen is that it functions as poorly designed open diff, with internal losses providing the locking*. If a reputable book on design of gearboxes is to be believed- only ZF and Salisbury were used in F1 (the book was written in '90ies, I believe). But I don't think cam-and-pawl (as much as I'm enchanted by their design and simplicity) are viable options because of wear issues- can't recall any car other than certain variants of Ford Capri being factory equipped with them... But I'd like to see if any of advanced materials (like ceramics, &c) might alter that.

* as can be seen with switch from Torsen I to Torsen II - orienting the worm gears 'axially' increased efficiency, but accordingly locking ratio decreased...

[cheapracer]

This Toyota has gotten more hype and publicity than any other boring box in history me thinks although I do appreciate any return to RWD.

jpf, on Feb 10 2012, 10:29, said:

I'm a little confused by this, because I had been thinking that speccing the car with a torsen was actually a pretty solid and welcome choice on Toyota/Subaru's part. What gives?

It's not a limited slip diff (but they all fall under the heading), it offers no slip as such but offers the outside wheel to freewheel/go faster on it's longer radius. This however also allows the insde wheel to spin freely if it leaves the deck.

The Torsen is cheap, bulletproof, quiet, doesn't need adjustment/maintenance and will last the life of the car and that's why it's popular.

jpf, on Feb 10 2012, 10:29, said:

In the video, the engineer and Harris talk about a future optional "mechanical differential" for the drifting and race crowd. What would that be?

[Lee Nicolle]

cheapracer, on Feb 10 2012, 07:38, said:

This Toyota has gotten more hype and publicity than any other boring box in history me thinks although I do appreciate any return to RWD.



It's not a limited slip diff (but they all fall under the heading), it offers no slip as such but offers the outside wheel to freewheel/go faster on it's longer radius. This however also allows the insde wheel to spin freely if it leaves the deck.

The Torsen is cheap, bulletproof, quiet, doesn't need adjustment/maintenance and will last the life of the car and that's why it's popular.

Exactly, though done by slightly more civilised methods I would hope. Drifting is pure spool, nothing else is any good.

[kikiturbo2]

@cheapy's diff conversion LOL.... we ust did this on a for sierra banger for snow work..

Torsen, as was pointed out, is quite bulletproof and easy to drive on the road... Normal drivers will prefer it as it is quite progressive and makes no funny noises.. on the other hand is is of very little use in very slippery conditions and when one wheel is in the air..

The diff that will be used for drifting and more serious track work will be a "normal" 1.5 way plate diff... I suspect drifters will want it with some preload..
As for spool, local drifters I know acutally prefer the diff over spool..

[24gerrard]

Your gonna get a lotta wear on the plates if you use a plated LSD for drifting.
Cheaper to weld it up and put up with the problems of turn in and tyre wear.
After all drifting is 'slow' anyway.

[Catalina Park]

24gerrard, on Feb 10 2012, 20:37, said:

Your gonna get a lotta wear on the plates if you use a plated LSD for drifting.
Cheaper to weld it up and put up with the problems of turn in and tyre wear.
After all drifting is 'slow' anyway.

I imagine that tyre wear is a big concern when drifting.

[24gerrard]

Catalina Park, on Feb 10 2012, 09:47, said:

I imagine that tyre wear is a big concern when drifting.

Well you wont reduce it with a LSD.

[cheapracer]

Lee Nicolle, on Feb 10 2012, 15:34, said:

Exactly, though done by slightly more civilised methods I would hope. Drifting is pure spool, nothing else is any good.

C'mon Lee, you know all the speedway sedans, most budget rally guys and a few circuit racers CIG their diffs. Without doubt some Drifters do it too.

Catalina Park, on Feb 10 2012, 17:47, said:

I imagine that tyre wear is a big concern when drifting.

I seen what ya's done there ...

Edited by cheapracer, 10 February 2012 - 11:09.

[Catalina Park]

cheapracer, on Feb 10 2012, 22:04, said:

I seen what ya's done there ...

Yep, I fell for it and clicked the "View this post" button.
I won't make that mistake again.

The trouble with welding diffs is making sure that it all stays straight as the welds cool. Better off just filling it with molten lead and using it a a vicious* coupling.




*vi·cious/ˈviSHəs/
Adjective:
Deliberately cruel or violent.
(of an animal) Wild and dangerous to people.
Synonyms:
wicked - malicious - evil - bad - perverse

[NeilR]

given you have a major manufacturer making a high production numbers fun car with neutral handling I think the actions of the diff would be crucial. The torsen is gentle in application and effects under all conditions and would suit the vast majority of the market.

[kikiturbo2]

NeilR, on Feb 10 2012, 23:14, said:

given you have a major manufacturer making a high production numbers fun car with neutral handling I think the actions of the diff would be crucial. The torsen is gentle in application and effects under all conditions and would suit the vast majority of the market.

exactly.. and for those wishing for more edge, the aftermarket will oblige.... in great numbers..

for the record... FT-86's predecessor... AE86, had a "propper" plate diff..

[NeilR]

yes of course, but then the manufacturer is not open to legal issues, reputation damage and reliability issues with after market fit parts.

[Lee Nicolle]

cheapracer, on Feb 10 2012, 12:04, said:

C'mon Lee, you know all the speedway sedans, most budget rally guys and a few circuit racers CIG their diffs. Without doubt some Drifters do it too.




I seen what ya's done there ...

Cheapie, with the amount of mini spools advailable for most popular makes these days nobody bothers to weld em. Though some styles, especially Holden ones with minispools do not support as much axle and I will still weld those. I did one recently.
But Borg Warner salisburys live with mini sppols wheras welding them they never do.
I reckon I have welded or minispooled a 100 diffs now, Holdens,Commondores, Mazdas, Datsun, Galant, Sigma,Falcons, Valiants, Austin A90 and A40 and an early Landcruiser!

And Michael, leaded diffs always blow up. Welding breaks and lead is far softer!

[Lee Nicolle]

As an aside I have driven a couple of racecars with Detroit Lockers. They are the most god awfull pieces of crap ever. They clunk and bang and are locked when they should be unlocked and vice versa and they are very good at breaking axles. Including 31 spline Ford Axles.
Both were converted to spools and were faster afterwards one the driver learnt how to drive them. And both stopped their appetite for axles.
A clutch type LSD can be nice BUT will will be always on a regular rapid state of decline, and the friction material doesnt do diff bearings or gears any good either. That on a quick road car yet alone a track car!

[Wolf]

Lee Nicolle, on Feb 11 2012, 03:06, said:

As an aside I have driven a couple of racecars with Detroit Lockers. They are the most god awfull pieces of crap ever. They clunk and bang and are locked when they should be unlocked and vice versa and they are very good at breaking axles. Including 31 spline Ford Axles.
Both were converted to spools and were faster afterwards one the driver learnt how to drive them. And both stopped their appetite for axles.
A clutch type LSD can be nice BUT will will be always on a regular rapid state of decline, and the friction material doesnt do diff bearings or gears any good either. That on a quick road car yet alone a track car!

Lee, I had a discussion on another forum on merits of cam-and-pawl diffs which should be much smoother and driveable than Lockers- any expirience with them? BTW I've almost given up on trying to fathom how and why they work the way they do (but that discussion pointed out some similarities between the two, which helped me understand what they do, but not why)...

[Lee Nicolle]

Wolf, on Feb 11 2012, 03:29, said:

Lee, I had a discussion on another forum on merits of cam-and-pawl diffs which should be much smoother and driveable than Lockers- any expirience with them? BTW I've almost given up on trying to fathom how and why they work the way they do (but that discussion pointed out some similarities between the two, which helped me understand what they do, but not why)...

I have ever only played with BW, GM, 9" clutch type LSDs, Detroits lockers once [I removed it and put in a full spool] and a lot of open wheelers either just fixing worn bearings, ring and pins, broken star gears or putting mini spools in them. Usually replacing the odd bearing or two in the process.
From what I understand they are just a form of locker, there is several styles and none are very smooth in my very limited experience with them.
I have a Landcruiser with diff locks and I have been tempted to pull one out and just have a look. But these days I lack energy and body to do unecesary stuff. The workshop manual is as clear as mud.

[cheapracer]

Wolf, on Feb 11 2012, 10:29, said:

Lee, I had a discussion on another forum on merits of cam-and-pawl diffs which should be much smoother and driveable than Lockers- any expirience with them? BTW I've almost given up on trying to fathom how and why they work the way they do

It's very simple (there is only 4 parts) but not as simple to explain, in fact I have written this several times already!

In theory it's a very simple double sprag clutch arrangement.

The center hub (the drive) holds the sprags inplace with equal amounts of the sprag poking outwards and inwards.

When the outer drum wants to go slower the the sprag with it's asymmetric end shape wedges it in place but if it wants to go faster the sprag allows it to pass - ie; sprag clutch.

Now, the same as the outer drum, when the inner drum wants to go slower the the sprag with it's asymmetric end shape wedges it in place but if it wants to go faster the sprag allows it to pass - again ie; sprag clutch.

Simple right - It's the relationship between the inner and outer that seems to confuse...

The double ended sprag has just the right shape and length to barely let the inner drum go by while the shape (ramp) helps to jam it into the outer drum as it goes by ie; ramps the sprag upwards and of course vice-versa when the outer drum goes faster it pushes/ramps the sprag into the inner drum and locks it as it goes by.

So you end up with similar as a Torsen, the outside wheel can freewheel/go faster and that action locks the inside wheel.

That would be a simple explanation for a 100% cam and pawl but they are not ..

By varying the number of ramps on the inner and outer drums as well as different number of sprags the outer can still freewheel faster but when the inner starts to spin/catch up it then actually starts to immediately apply torque to the outside wheel some by allowing the sprags to 'walk' rapidly in and out of the drums (effectively a gear ratio) - which is what they actually do in their real life applications and I really have trouble explaining.

A Hewland I saw once had 8 sprags, 11 ramps inner and 15 outer ramps I don't know what 'torque split' (gear ratio) that offers but I know they have choices of splits (ratios).

[Wolf]

I think I've seen in few sources that they claim the 'ideal' setup was 8 pawls, 11 lobes on inner cam and 13 on outer*. (BTW, my big mistake was assuming the diff would allow slower cam to drag rather than faster one to 'overrun' and was confused because it implied completely opposite behaviour than the one it actually has ) Actually, your mention of asymmetric shape of pawls/sprags is the first I've heard.

But what is still bothering me a little is that I (again) assume the cams must have radial 'play' in order to prevent the diff locking solid when two lobe 'peaks' try to pass a pawl (which is by the look of things inevitable)- and I've heard or seen no mention of it... (this is something that was reinforced when discussing it with a friend of mine over few cups of grog, and he being an electrician asked 'this may sound silly, but what if... one could unfold those cams to become like a rack?')

* I would think that the 'black art' of the pawl and lobe numbers is not so much related to split, but to ensure that whatever the positions of cams are, a reasonable number of pawls (I'd say three) is placed in such way to be able to transmit torque to both cams

[cheapracer]

The next time I have a bit of useless time I might try and draw it's action - more for me than anyone else!

What's amazing is they were used in F1 until the late 80's (possibly later) predictable, progressive action and unbreakable - but they wear quite fast apparently and that's why you don't see them in road cars.

[jpf]

I found this page with a 3D model of a cam-and-pawl diff, as well as photos of the finished cams. It doesn't fully explain (to my dumb eyes) the freewheeling action, but it is still instructive:

http://www.ashlar.co...e-differential/

[Tony Matthews]

Except that they've missed the pawls out - or is there another illustration/rendering?

[munks]

cheapracer, on Feb 10 2012, 06:38, said:

This Toyota has gotten more hype and publicity than any other boring box in history me thinks although I do appreciate any return to RWD.

Well I think you nailed the reason for the hype. At least in the States, our current choices for RWD are either overpriced luxury/super cars or overweight pigs that can't turn (oops I mean muscle cars). Well, okay, you can also get a Miata.

[MatsNorway]

And miatas har hairdressers cars.

Decpite it being a great car.

They should make a coupe of it.

[kikiturbo2]

munks, on Feb 13 2012, 20:12, said:

Well I think you nailed the reason for the hype. At least in the States, our current choices for RWD are either overpriced luxury/super cars or overweight pigs that can't turn (oops I mean muscle cars). Well, okay, you can also get a Miata.

au contraire, you have genesis coupe, which is cheap and tuner friendly.. We do not have it in Europe..

[kikiturbo2]

MatsNorway, on Feb 13 2012, 21:45, said:

And miatas har hairdressers cars.

Decpite it being a great car.

They should make a coupe of it.

they did...

[cheapracer]

kikiturbo2, on Feb 14 2012, 11:25, said:

au contraire, you have genesis coupe, which is cheap and tuner friendly.. We do not have it in Europe..

390Z (getting in ahead of time!), BMW, Benz etc, there's a few if you look.

If I was asked by an up and coming Chinese car company today I would tell them to make a lightweight sporty 2 litre RWD sedan to get a foot into the market (cross MK2 Escort/Toyota AE86 replacement). They certainly have the off the shelf parts ready to make them because of the proliferation of light commercials vehicles here.

[MatsNorway]

kikiturbo2, on Feb 14 2012, 04:25, said:

they did...

Show me.

[kikiturbo2]

MatsNorway, on Feb 14 2012, 07:55, said:

Show me.

http://image.motortr...zda-m-coupe.jpg

[MatsNorway]

kikiturbo2, on Feb 14 2012, 09:01, said:

http://image.motortr...zda-m-coupe.jpg

Someone made a custom car..

[kikiturbo2]

actually, mazda did... as a one off...

here is an old test I read in the us..

http://forums.vwvort...ead.php?4639151

[MatsNorway]

Could you buy it over the counter?

Im guessing no.. and i rest my case.

[jpf]

Tony -- yes, they left out the pawls, which I would love to see, but it's better than nothing if you are starting from scratch in your understanding of these mechanisms, like I am...

As for the car itself, I agree that it's getting a lot of hype, but so far people seem happy with it. And I'm glad to see someone embracing the idea of building and selling a lightweight car that is designed to be fun to drive at merely misdemeanor speeds.

I agree that the Miata and Genesis are good options in this area already. But I still think this segment needs more attention

[cheapracer]

jpf, on Feb 15 2012, 00:09, said:

And I'm glad to see someone embracing the idea of building and selling a lightweight car that is designed to be fun to drive at merely misdemeanor speeds.

Lightweight is 1250kgs? Let me assure you, we are from different eras.

[Wolf]

I'll upload few pics later, but here's my attempt of 'reinventing' it (had to do a dreaded step back in CAD technology)...

After a bit of fiddling (I'm not quite there yet), I get pawls 3, 6 & 8 transmitting the torque (assuming the whole thing rotates clockwise). But my point about cams having 'play' is because of pawls positioned at 12 and 6 o'clock. If there was no way for cams to move relative to each other those pawls will lock the diff. In case the outer diff should try to rotate clockwise, the pawl at 12 o'clock will not alow it, and the pawl at 6 o'clock will prevent inner cam from rotating clockwise...

[Tony Matthews]

I thought the pawls were free to move on their axes, but you have shown them all radial to the centre line of the diff. Is that right?

[Wolf]

Tony, each of them can move radially- e.g. top and bottom ones can move up or down, and left and right ones can move left or right... And the whole thing rotates- if there's no differential action both cams and the cage holding pawls rotate at the same speed, so we can look at it as stationary. Should differential action occur, we then merely have to rotate either of the cams clockwise while the cage with pawls and other cam are stationary.

Edit, here's the pic from 'Rennwagentechnik':

Edited by Wolf, 14 February 2012 - 18:25.

[Tony Matthews]

Ah, I needed to see the cage! So they can't turn, only move in and out. I had visions of them all ending up at the bottom of the diff in a heap!

[Wolf]

Yep, the cage transmits torque from the gearbox... To me, from engineering point of view- it's a beauty (simplicity and ingenious design). It's a shame that so little is available on them, seeing how long they've lasted in grand prix cars (from '30ies to '90ies), and they according to Carroll Smith worked great (but for constant wear which deteriorated their performance).

[MatsNorway]

All this makes me want to draw diff idea i have. For RC cars. balldiff/geardiff under braking and spool during acc.

If anyone knows about a existing design let me know. I can scale it down.

And now that i can get Inventor installed at my home PC that kinda likely in the future.

Edited by MatsNorway, 14 February 2012 - 20:04.

[Wolf]

Mats, how about having a proper spool with one of the halfshafts having a clutch (I'd say with dogs) which would disengage when not under power?

[GreenMachine]

MatsNorway, on Feb 15 2012, 02:26, said:

Could you buy it over the counter?

Im guessing no.. and i rest my case.

Yes. If you were lucky/quick - it was made in low numbers (few hundred?), iirc by a subcontractor. There were two models styled slightly differently.

... returning you now to normal programming ...

[jpf]

cheapracer, on Feb 14 2012, 11:32, said:

Lightweight is 1250kgs? Let me assure you, we are from different eras.

I firmly agree, and I mourn the coming of Dany Bahar's Lotus with the rest of you, but it's 2012, and a 2600lb series production sports car counts as lightweight in this market. I don't know where it all goes.

[cheapracer]

You 3D'ing it Wolf?

Kewl .... the internet demands it!

[MatsNorway]

At least make a sketch om paper or something.

GreenMachine, on Feb 14 2012, 22:20, said:

Yes. If you were lucky/quick - it was made in low numbers (few hundred?), iirc by a subcontractor. There were two models styled slightly differently.

All i can find is that it was a concept.

Wiki got nothing.

And no..linking to some dude on a forum sayin stuff does not count for much.

Edited by MatsNorway, 15 February 2012 - 07:33.

[Lee Nicolle]

Wolf, on Feb 14 2012, 21:35, said:

Mats, how about having a proper spool with one of the halfshafts having a clutch (I'd say with dogs) which would disengage when not under power?

And what is a locker?

[cheapracer]

Wolf, on Feb 15 2012, 04:35, said:

Mats, how about having a proper spool with one of the halfshafts having a clutch (I'd say with dogs) which would disengage when not under power?

I have offered my twin sprag clutch LSD idea many a time. Spool drives left and right sprag clutches, clutches drive axles, turn corner outside wheel is free to go faster.

Lee Nicolle, on Feb 15 2012, 18:39, said:

And what is a locker?

A cabinet that you keep your schoolbooks in?

[GreenMachine]

MatsNorway, on Feb 15 2012, 18:15, said:

All i can find is that it was a concept.

Wiki got nothing.

And no..linking to some dude on a forum sayin stuff does not count for much.

"During the 2004 model year, a division of Mazda in Japan produced the Roadster Coupé, with an integral hardtop roof. The body structure was reworked to incorporate the roof for a substantial increase in chassis rigidity and a weight increase of 10 kg (22 lb). Production was limited to 350 units for Japan only." from Wiki (just above "Mazdaspeed MX-5 (2004–2005)")

Photos can be found here MX5 forum

[Wolf]

cheapracer, on Feb 15 2012, 06:13, said:

You 3D'ing it Wolf?

Kewl .... the internet demands it!

Yep, I was always put off by proE's user unfriendly interface, so decided to teach myself by making fully parametric assembly with this diff... Didn't want to put any finishing touches incl. pawls) before I was done understanding the concept properly. But e.g. one can with merely one click change number of lobes on each cam, number of pawls or their length, or spline lengths and the whole thing will automatically change accordingly.

P.S. My apologies to Tony for showcasing a program that lets any fool, like myself, willing to spend half an hour learning it to do a decent looking technical illustration from a 3D model.

[24gerrard]

You could always connect each wheels drive shaft through an oil pump.
Then block off the oil outlet from each pump so the pump becomes rigid. rotates with the outlet and full torque is transmitted.
Simple matter then to bleed off oil under pressure to achieve a full range LSD.
Of course you can then use the oil under pressure to recover energy as well.
Damn I did that back in 1979 and the regulations banned it.
Hmm, back to the drawing board, mind you wolf's program looks interesting, fed up with keeping on top of solid works in between being retired and doing the allotment.