Let's talk LSD/TBDs

[Driven97]

My 84 / 915 box car is primarily an autocross car. It currently is open diff. In certain situations I'm getting a fair amount of wheelspin, and am considering upgrading the differential to find a little time out there.

Now if I understand correctly, there are two major types of diffs.

An LSD such as a KAAZ, OS Giken, or factory ZF uses clutch packs and ramps. The ramps progressively squeeze the clutches tighter together locking the diff when there is torque difference between the input and output. Depending on the ramp angles put in by the designer, the differential can have different lockup characteristics.

A Torque Biasing / Torsen type differential such as a Quiafe instead uses gears that squeeze together when there is a torque difference between the left and right outputs.

^1st question - is that correct?

Now I've only driven torque biasing types in other cars before. I like them because of their progressive nature, and I also like that they don't wear out. But apparently they do nothing in deceleration, of if a single wheel gets heavily unloaded.

Clutch types can work in both directions, and don't give an eff if a wheel is totally in the air. But they also have preload, which can create understeer in tight turns. Plus they are sensitive to setup, and also wear over time.

Looks like there are a LOT of choices for a 915 car - at least 6 with some casual searching. How do I know which one is best for my needs?

[Mehoff]

Nothing to talk about; if you want to add competitiveness to the car on the track you need to go to a clutch based diff like the ones GT makes (talk to Matt Monson on this board). Torque biasing on a track car is just added weight to the bottom of the car....no significant benefit.

If you want to go the less expensive/inexpensive route just leave the diff open. Paying for the part and installation of the cheaper TBD is a complete waste of money (IMHO).

[Driven97]

You've used a torsen type? Why do they work on other cars and not a 911?

Btw, the car rarely sees a track. It does see about 20 autocross courses a year though, which is where I would like the improvement.

[Bill Verburg]

Originally Posted by Driven97

My 84 / 915 box car is primarily an autocross car. It currently is open diff. In certain situations I'm getting a fair amount of wheelspin, and am considering upgrading the differential to find a little time out there.

Now if I understand correctly, there are two major types of diffs.

An LSD such as a KAAZ, OS Giken, or factory ZF uses clutch packs and ramps. The ramps progressively squeeze the clutches tighter together locking the diff when there is torque difference between the input and output. Depending on the ramp angles put in by the designer, the differential can have different lockup characteristics.

A Torque Biasing / Torsen type differential such as a Quiafe instead uses gears that squeeze together when there is a torque difference between the left and right outputs.

^1st question - is that correct?

Now I've only driven torque biasing types in other cars before. I like them because of their progressive nature, and I also like that they don't wear out. But apparently they do nothing in deceleration, of if a single wheel gets heavily unloaded.

Clutch types can work in both directions, and don't give an eff if a wheel is totally in the air. But they also have preload, which can create understeer in tight turns. Plus they are sensitive to setup, and also wear over time.

Looks like there are a LOT of choices for a 915 car - at least 6 with some casual searching. How do I know which one is best for my needs?

Yes, to the forest part

for A/X the usual choice is the torque biasing type, you don't want the diff to do anything on decel.

The clutch types on decel tend to add a bit of understeer to the car the higher the lockup the more understeer, the last thing needed on a tight A/X course

GT carries both types, an asymmetric clutch type would be the choice for track use and the gear type for A/X.

[rfloz]

Originally Posted by Driven97

Now I've only driven torque biasing types in other cars before.

I like them because of their progressive nature YES,

and I also like that they don't wear out BIG YES - replacing clutches is a STINKY PITA.

But apparently they do nothing in deceleration, of if a single wheel gets heavily unloaded. [I]NOT AN ISSUE ON PORSCHES /I] - with rear weight bias, sticky tires and weight shift to the rear on acceleration, this is not a problem unless you have mega horsepower.

Have fun.

[Bleedsblue]

Matt, have you considered just swapping in an OEM LSD 915? I know STR RWD allows any aftermarket diff, but you could (conceivably) save some $ by just trading for a 915 w/ factory LSD. I'm sure a custom job could out-perform it, but...

I've only a couple events under my belt, but I can tell you the LSD in mine induces no understeer (especially when I lift, ha). I do have some neg. camber up front (like you) which helps with the initial turn-in, car is otherwise stock.

I've never autox-ed an open diff Carrera, so I have no base of comparison.

[sjf911]

I chose a TBD for AX as I don't otherwise track the car. It is night and day over an open diff at least at my HP level.

[Driven97]

Matt, have you considered just swapping in an OEM LSD 915?

I thought about keeping an eye out in the classifieds for a used 915 with some sort of lsd in there, but I'm a little nervous about buying a possible problem box. I know my box is decent with the exception of a tired first gear synchro, plus it's original to the car which is nice.

I don't really have the budget to buy a whole box and get it rebuilt.

I chose a TBD for AX as I don't otherwise track the car. It is night and day over an open diff at least at my HP level.

Well hell, I only have 200hp, not a fire breathing turbo.

FWIW, this is what the car is like with the open diff.
YouTube skip to about 1:35, the inside rear just blows up. It was doing that much worse on day 2 on both courses, in more places. I'm running pretty tame rear camber, and the OE Carrera rear sway. I have a larger rear sway I'd like to use, but I know it'll just make the one wheel peel worse.

[Trackrash]

A video or pictures from the rear would better show what is going on with your car.

Looks like a stiff suspension. What tires are you running?

[Driven97]

Dunlop z2, 255/40-17

[kjchristopher]

OS Giken can be tuned to vary the amount of lock it has in decel and in accell. I suspect other clutch pack designs have this capability as well. I ran an OS Giken in the RX8. Stupendous improvement over the stock torsen.

The A Mod has a torsen and I hate it.

I know the guy who owns/runs OS Giken in the US. I don't think they have anything for the 915 - but that is based on discussions with him last year. Shoot me an email if you want his contact info or introduction.

It is hard to say for sure based on what we can see in the video, but a torsen would do much for what appears to be unloaded wheel spin.

Also, torsens suck.

[Plavan]

Call Matt at GT. I have an Guard/ GT LSD in my 911 racer- Love it. I have TBD going in my 818R (they dont make an "Good" LSD for it)-

I have driven a TBD- Still does what you want without the decel component. Still very fast guys on the tracks with them.

OS GIKEN...... Nothing but horror stories from my buddies that used them.

[Driven97]

Kj - what goes on with the amod car that you don't like?

I've got friends with an stx rx8 who went from the os giken back to the oe torsen. Both drivers love it, and they've gone from reasonably competitive to serious contenders with the change.

More confusing is a pair of drivers in an STR NC. The car owner loved the os giken so much he bought a spare. The co-driver hates it, thinks it locks unpredictably on occasion. They're both inconsistent in the car, though.

Other people run clutch diffs with success, and yet others blow then up - Davies had to bring a backup car to Toledo due to a broken diff.

So I'm seeing both success and horror stories with diffs. If I'm going to do this, in only going to have the budget for a one and done, and I'll be stuck with it.

[kjchristopher]

Originally Posted by Driven97

Kj - what goes on with the amod car that you don't like?

I've got friends with an stx rx8 who went from the os giken back to the oe torsen. Both drivers love it, and they've gone from reasonably competitive to serious contenders with the change.

More confusing is a pair of drivers in an STR NC. The car owner loved the os giken so much he bought a spare. The co-driver hates it, thinks it locks unpredictably on occasion. They're both inconsistent in the car, though.

Other people run clutch diffs with success, and yet others blow then up - Davies had to bring a backup car to Toledo due to a broken diff.

So I'm seeing both success and horror stories with diffs. If I'm going to do this, in only going to have the budget for a one and done, and I'll be stuck with it.

I'd say your friends didn't take the time to actually tune everything to what they individually needed. The easy button is to go back to the torsen. (Granted - it is a PITA to disassemble, change, reassemble and retest. But you should be able to find a setting that is 99.9% there, even when you change tires with differences in longitudinal grip characteristics.)

I'm not necessarily pushing the OS Giken for you (I'm pushing away from the Torsen). Like I said earlier, I don't think they have a model that fits your 915, plus very few Porsche drivers use them and even fewer would be willing to take them in/out and get the best settings. But there are a bunch that have done this with other diffs, so you won't be charting completely new ground.

What do I not like about the torsen in the Amod? I don't like inside rear wheel spin on an unloaded tire. There are compromises in every set-up - as a result, I do on occasion unload my rear tire enough in turns that I have to wait on the gas.

[Geary]

OSG .. can be tuned to vary the amount of lock it has in decel and in accell. I suspect other clutch pack designs have this capability as well.
.. Also, torsens suck.

To me, this suggests a general lack of knowledge about LSDs. ZF and Porsche LSDs are known for their tunability, and have been for many decades. Quantity/sequence of friction discs, preload settings, Belleville washer profile/thickness, ramp angles .. all of these elements offer possible combinations that experienced tuners can utilize.

On the other hand, Giken burst onto the Porsche scene five years ago, claiming some previously unexplored technology (that other manufacturers were “incapable” of), perfect out of the box for any application, and with zero failures of any type, EVER. Yet, when tested by Pro Porsche drivers, they were notable for their lack of smooth progression, exhibited full and immediate lock/unlock causing unpredictable performance on the race track .. and the list of catastrophic failures continues to grow in higher horsepower installations.

Torque-biasing differentials by nature are transparent in operation, with smooth, progressive locking. They are never on-off jerky like the spring-unloaded OSG plate differentials. Despite this apparent perfect fit of TBDs for autocross, some old-timers prefer to fine tune their ZFs even for this application.

[Matt Monson]

Only thing I would add to what others have said is that the performance of a particular differential in a chassis other than a 911 pretty much never crossses over to how it will perform in a 911. We need to remember that this is pretty much the only car in the world with a rear engine, RWD configuration. That makes for very unique chassis behaviours and requires differential set up specific to that configuration. Put a LSD that is tuned for a Front Engine RWD car into a 911 and it's not going to drive correctly. FE RWD vehicles historically tune their LSDs exactly opposite of how we tune 911 LSDs.

BMWs, Mustangs, Camaros, Mazda RX7's,8's and MX5's all go with more lock on throttle than on braking. They might run a 60/40 configuration. 911s prefer it the other way around, 40/60, with less lock on throttle because the rear engine design does offer such phenominal on throttle traction. The weakness of the 911 chassis is all that weight out behind the rear axle makes for a certain pendulum effect and a rear end that feels like it wants to walk and step out on you. A lot of braking lock up remedies that and settles it down.

However, when you do that, you make the car want to push on corner entry. This is why the auto-x 911 racers prefer the TBD. It has no lock under braking. In fact, good autoxers use that rear engine weight to toss the car coming in and get the car rotated way early, like a rally car, so that when they put the loud pedal down, the car is already straight and going right where they want it to go. I prefer a TBD for auto-x, though I can, and will tune an LSD for people that would prefer to have one. That's usually the guy who does both auto-x and DEs. The dedicated auto-x guys almost exclusively run TBDs.

I also want to dispell two myths:
1. LSD rebuild cycles, on a good LSD, like the Guard Transmission unit, are not subject to frequent rebuilds. Just ask member CStreit. I just rebuilt his LSD. He's a regional NASA championship contender every year in the Great Lakes. He ran our LSD in his long nose with engine swap and G50 for 3 years. Then he bought a 996 Cup. Moved the LSD over to that. Ran two more seasons on it. 5 years of winning racing between rebuilds. Not something you do very often, at least if you use a GT product.

2.TBDs do NOT last forever. Just like the rest of your gearbox, they are driven by gears. Gears do have a service life. It may be greater than the service life of a clutch, but it's not forever. I have seen guys wear out every single brand of TBD on the market, including ours. It may be a decade or more, but do not assume that because you bought a TBD that you will never need another differential for your car.

[Geary]

To further what Matt said about a TBD's lack of durability .. I've seen a top ALMS team practically destroy a TBD in a single weekend of testing. Loose, lifeless .. almost unusable, even for a street car. On the other hand, for autocross/street they can last for a very long time.

[kjchristopher]

Originally Posted by Geary

To me, this suggests a general lack of knowledge about LSDs. ZF and Porsche LSDs are known for their tunability, and have been for many decades. Quantity/sequence of friction discs, preload settings, Belleville washer profile/thickness, ramp angles .. all of these elements offer possible combinations that experienced tuners can utilize.

On the other hand, Giken burst onto the Porsche scene five years ago, claiming some previously unexplored technology (that other manufacturers were “incapable” of), perfect out of the box for any application, and with zero failures of any type, EVER. Yet, when tested by Pro Porsche drivers, they were notable for their lack of smooth progression, exhibited full and immediate lock/unlock causing unpredictable performance on the race track .. and the list of catastrophic failures continues to grow in higher horsepower installations.

Torque-biasing differentials by nature are transparent in operation, with smooth, progressive locking. They are never on-off jerky like the spring-unloaded OSG plate differentials. Despite this apparent perfect fit of TBDs for autocross, some old-timers prefer to fine tune their ZFs even for this application.

Yeah. I don't know ****.

[aschen]

Please excuse the following highly technical and long winded advice:

I'd trust Matt M on this one

[Matt Monson]

Aschen,
I appreciate the endorsement. I would also like to point out the Geary is Mr. Guard, my mentor. He is retired and has zero financial interest in my company. You can take his input as that of an expert with nothing to gain. I am not so unbiased.