LSD rebuild

[Chakka71]

I have an RSA with 75,000miles on it. My original plan was to re-gear 3-5th, but it's just not in the financial cards (4-5k). I'm assuming my LSD is pretty tired at this point in its life, so I'm wondering about having it rebuilt. I talked(email) with GT Gears and got the price to have them do it. All that have used them seem happy and for $500 it seems like a good option.

My question is would this give me some track improvement and what would be a good track rebuild %. Should I mess with a 4 plate vs. 2 plate set-up. From what I understand the 4 plate is a better track set-up, but shortens the life expectancy since the plates are thinner. Makes sense to me, but what does shorten mean......I do 5-7 DE's a year.

[jpachard]

What is the current lock ratio? If your engine is stock I would stick with 40/60 20/80 is too aggressive, at least for my driving style. Paul Guard did my last LSD for my '87 race car and that was awesome. I'm not sure how many plates were in it though.

Cheers, James

[Bill Verburg]

the only 964s that came w/ the asymmetric lsd's were the turbos and the RS, which have 20/100s, I agree that the 20/100 in't the best choice for most people.

a 40/60, 40/65 or for more serious track use 50/80 would be extremely desirable but if you are rebuilding a stock 964 so called 40% lsd then you are not going to be able to have the asymmetric properties.

so I'd rebuild w/ 4 plates(Gt has superior properties compared to oem) and modest preload, that's the way earlier 40% lsd's came from the factory, modest preload allows the diff to function more freely in normal street type use, but will still function effectively for track use. I wouldn't be overly concerned w/ wear

that said lsd plates are wear items and they do generate heat,

[MCIPSERIC]

Originally Posted by jpachard

What is the current lock ratio? If your engine is stock I would stick with 40/60 20/80 is too aggressive, at least for my driving style. Paul Guard did my last LSD for my '87 race car and that was awesome. I'm not sure how many plates were in it though.

Cheers, James

I just put a new Wavetrac LSD in my 993. This is my first LSD. I was told that it is set at 20/80. What do you mean by, "too aggressive." Thanks!

[Matt Monson]

Originally Posted by Bill Verburg

so I'd rebuild w/ 4 plates(Gt has superior properties compared to oem) and modest preload, that's the way earlier 40% lsd's came from the factory, modest preload allows the diff to function more freely in normal street type use, but will still function effectively for track use. I wouldn't be overly concerned w/ wear

that said lsd plates are wear items and they do generate heat,

The problem with these old ZF G50 LSDs is that Porsche designed them as a 2 plate diff to start with. For us to rebuild it to 4 plates requires that we use our thin little 1mm friction discs to make it fit, which is where the wear issues come into play. Our standard friction discs, are 2.1mm and have significantly more depth to the friction material, which is something that is added to the steel plates and then heat treated. So in this instance, it's really the thickness of the material, more than the number of friction surfaces that makes for accelerated wear.

As such, I only recommend converting these 2 plate LSDs over to 4 plates when the vehicle is being raced and is looking to maximize the performance out of the LSD. Otherwise for a street car that sees some occasional track duty, I really feel that just rebuilding it to the stock 2 plate configuration with the thicker standard plates is usually the best approach, both with respect to durability and day to day driveability.

Regards,

Matt Monson
Guard Transmission LLC

[Matt Monson]

Originally Posted by MCIPSERIC

I just put a new Wavetrac LSD in my 993. This is my first LSD. I was told that it is set at 20/80. What do you mean by, "too aggressive." Thanks!

The Wavetrac unit isn't an LSD in the traditional sense of the word. The locking percentages that we use to talk about a ZF style clutch type LSD are very different than the 20/80 that's being referred to in the Torsen Gleason style torque biasing differential you put in your car.

The 20/80 in your differential refers to the bias ratio of the differential. Here's a good article for you to read about how TBDs function:
http://www.torsen.com/files/Traction_Control_Article.pdf

When we talk about the locking factor of a clutch type LSD, it's talking about wheel slip and torque differentials from wheel to wheel before the LSD locks. It's not about how much power it transfers, but what is required to lock it. Whereas the TBD is talking about power transfer. It's two different things. On the LSD the first number is the lockup on acceleration and the second number is the lockup under braking.

[chris_seven]

I get really confused by ZF speak in LSDs.

I guess that the definition of the 964 diff at 20-100 means locking factors rather than ramp angles.

Does this means it will be quite gentle on the accel side and quite aggressive on the coast side?

The number of plates must have a significant effect on the set up as the total friction surface must affect the amount of torque that can be biased for a given ramp angle?

Dose anyone know the ramp angles of the 964 diff? it would be interesting to know.

The preload will also affect the amount of torque that can be biased and this needs to be measured.

As it looks as if there is only one Belville available the only way to adjust the preload will be to change the disc thicknesses.

With regad to the Wavetrac this is a Torsen Type Torque Biasing Diff and I really don't think it can be described as being set at 20/80.

Unlike other Torsen diff the Wavetrac does have the ability to have some preload added with a sophisticated cam design. The preload means it will still provide some torque bias with one wheel off the ground unlike other Torsen's but I don't think it has much capability on the decel side.

I have been speaking to Wavetrac's UK distributor to find out what accel side torque bias they can develop and if there is any decel lock up.

We want a diff for a 911 that will be used on the East African Safari Rally and we want the ability to drive out of mud on split friction surfaces but we don't want understeer on decel.

The Wavetrac may be just the job.

[Matt Monson]

Chris,
Here's a post that Paul Guard made back in 2007 that may help you a little bit:

Originally Posted by Geary

Porsche's use of "percentage of anti-slip effectiveness" (what we just call percentage of "lock-up") makes a lot more sense to the layman than ramp angles.

All of Porsche's early LSDs came with 30-degree ramps (in both directions, as they were symmetrical). When 2 friction discs were utilized, the LSD was called a 40% LSD; when 4 friction discs were utilized (on both sides), the LSD was called an 80% LSD.

Porsche Motorsports 40/60 LSD has a 50-degree ramp on the acceleration side and 37.5-degree ramp on the decel side, utilizing a total of 4 friction discs. Most pro race teams opted for somewhat more aggressive 50/80 ramps. Some teams even used dual Belleville washers per side, for a tremendous amount of pre-load.

In '03 Porsche came out with a version of their 40/60 LSD that had thinner discs, allowing a total of 8 friction discs and 8 plain plates. Most teams used 40/60 rather than 50/80 ramps. These LSDs had no place on the street .... strictly track, due to their aggressive nature.

The "zero pre-load" idea is nothing new. A number of shops like to set up their LSDs with very little pre-load, utilizing much thinner Belleville washers and/or greater clearances. This sort of set-up is less aggressive in slow speed turns, as the clamping pressure provided by ramp-up is at a minimum during light throttle. More recently, set-ups with more friction discs and no Belleville washers have been experimented with. But this can be done with any LSD .... you don't need to spend 4K on an LSD that comes set up this way.

About 4 years ago, an experimental LSD similar to the Hewland type was made for us by our LSD manufacturer. It's a pretty sweet unit .... adjustable pre-load (via a small window in the side cover), 6 friction discs, 3 large diameter spiders that eliminated the cross shafts. Nobody wanted to even try it. It wasn't from Germany. Everyone wanted to stay with the standard design. Big expense for development, now a shelved project.

So we don't get too involved in experimental LSD designs any more ..... but we have almost all the bits to now offer any of the following plate/Belleville combinations to any of our "traditional" GT LSDs (which covers all 911-based models from '64-on):

1.0, 1.5, 2.0, 2.2 Belleville washers... which can of course can be eliminated altogether

2, 4, 6, 8 friction discs

40/60, 50/80, 80/80 ramps (which can be reversed.... or figures halved by using only two friction discs or placing discs back-to-back).

So how do we rate LSD locking percentages today? With all these variables, it's pretty darn confusing. Is an LSD with 40/60 ramps and 6 or 8 friction discs still a 40/60 LSD? Of course not. So many factors now determine the true "degree of anti-slip effectiveness" (not least of which is the material and finish of the friction discs and plain plates, a story by itself). But likely 99% of GT LSDs sold will continue to have the standard 4 friction discs with 40/60//50/80 options.

[Bill Verburg]

Originally Posted by Matt Monson

The problem with these old ZF G50 LSDs is that Porsche designed them as a 2 plate diff to start with. For us to rebuild it to 4 plates requires that we use our thin little 1mm friction discs to make it fit, which is where the wear issues come into play. Our standard friction discs, are 2.1mm and have significantly more depth to the friction material, which is something that is added to the steel plates and then heat treated. So in this instance, it's really the thickness of the material, more than the number of friction surfaces that makes for accelerated wear.

As such, I only recommend converting these 2 plate LSDs over to 4 plates when the vehicle is being raced and is looking to maximize the performance out of the LSD. Otherwise for a street car that sees some occasional track duty, I really feel that just rebuilding it to the stock 2 plate configuration with the thicker standard plates is usually the best approach, both with respect to durability and day to day driveability.

Regards,

Matt Monson
Guard Transmission LLC

Thanks Matt, You are certainly in a better position than I to recognize the merits of 2 vs 4 rebuilds


In essence the early 4 plates were really effectively 2 plate anyway

here is something I posted a while ago

Early ZF clutch types were 4 plate 40% units that could be made into 4 plate 80% diffs by rearranging the assembly order of the driven(d) and drive plates(D) they would be a 40% if the assembly order was dDD, 80% if it was DdD(more friction faces this way) brake away torque can also be adjusted by varying the preload on lsd assembly, the various plates come in different thicknesses to do this.

here's a schematic of a 915, 4 plate set up for nominal 40%



There is something wrong with that schematic. It shows the outer-most friction disc on each side against the spring/Belleville washer. I don't know why they drew it that way, but that just doesn't work properly. The accepted way (at least today) is to start by stacking the first friction against the pressure ring, then plain plate, friction disc, two plain plates, and then the Belleville washer.

At least to my knowledge, comments Matt?

[Chakka71]

The interest I have leans towards improving the LSD on the track. I don't do much street driving with it. For the amount of events I do, 5-7 a season, I guess i'm looking to understand what is accelerated wear? Would a 2 plate set-up last 2x as long as a 4 plate. Example : 3-5 yr. rebuild vs. 5-7 yr. with 2 plates(obviously i'm just throwing out numbers).

Will 4 plates be more effective(noticeably) better on the track then 2? Just trying to maximize track performance if i'm going to have it rebuilt.

[Matt Monson]

Bill,
I think I can explain what you are seeing there. On the old 901/914 LSDs, and including the early 915 LSDs they used what we call a wave washer. It wasn't a belleville washer in the sense that we use today. Just like what we call the plain plates (outer discs in Porsche's terminology) the wave washer (outer disc {dished profile} in Porsche's terminology) was tabbed to rotate with the outer body of the differential. The plate just had a little bit of a bow in it to make it act similar to a belleville and apply some preload. So they did actually run a friction disc directly against it, though I do 100% agree with your logic on it.

If you look at that skematic, they do start with a friction disc against the cups (aka ramps or pressure rings), but then they double up the plain plates, double up the friction discs, and then run it against the wave washer. Doesn't make sense to me, when you could just as easily pull one of those plain plates over to the other side and put it against the friction disc. And it seems I am not alone in this because when I get sent early ZF LSDs to rebuild, 9 times out of ten they have been restacked that way. And usually the wave washers are cracked in half and doing nothing for preload anyway.

Starting with the mid-year 915 LSDs and the 930s they introduced the belleville washers to it. This was a significant improvement to the design and something we use to this day. Your skematic predates that change in design.

In my eyes, the ZF G50 LSD is the bastard stepchild of the product line. The problem is they made the ramps too thick in the middle to restack it properly as a full thickness 4 plate LSD. With the GKN 993 2 plate LSD Porsche made modifying them easy on us because in that one they stuck in these big thick 3mm spacers on each end and left the ramps in the center smaller. I can easily pull out those spacers and convert a GKN 2 plate LSD to 4 plates and give the customer something that will perform on par with a modern GT3 LSD. But with the ZF version unless you want to significantly cut down the ramps on a surface grinder, and potentially cut through and compromise the heat treat in the process, there's just not enough room inside that LSD with to make it into a fully functioning motorsports capable unit. But for a street car/weekend track toy, it's totally adequate.

[Matt Monson]

Originally Posted by Chakka71

The interest I have leans towards improving the LSD on the track. I don't do much street driving with it. For the amount of events I do, 5-7 a season, I guess i'm looking to understand what is accelerated wear? Would a 2 plate set-up last 2x as long as a 4 plate. Example : 3-5 yr. rebuild vs. 5-7 yr. with 2 plates(obviously i'm just throwing out numbers).

Will 4 plates be more effective(noticeably) better on the track then 2? Just trying to maximize track performance if i'm going to have it rebuilt.

Our own 4 plate LSDs tend to last most PCA Club racers and DE drivers in the 4-6 year range, though obviously that's a gross generalization. If we rebuilt your LSD as 4 plates, I would figure your lifespan on it would be close to half that long.

4 plates will definitely be more effective and noticeable on track. The biggest thing you will notice is more rear end stability under braking.

[MCIPSERIC]

Originally Posted by Matt Monson

The Wavetrac unit isn't an LSD in the traditional sense of the word. The locking percentages that we use to talk about a ZF style clutch type LSD are very different than the 20/80 that's being referred to in the Torsen Gleason style torque biasing differential you put in your car.

The 20/80 in your differential refers to the bias ratio of the differential. Here's a good article for you to read about how TBDs function:
http://www.torsen.com/files/Traction_Control_Article.pdf

When we talk about the locking factor of a clutch type LSD, it's talking about wheel slip and torque differentials from wheel to wheel before the LSD locks. It's not about how much power it transfers, but what is required to lock it. Whereas the TBD is talking about power transfer. It's two different things. On the LSD the first number is the lockup on acceleration and the second number is the lockup under braking.

In a nutshell, what is the differences one would see between a LSD and TBD?

[Bill Verburg]

Originally Posted by Matt Monson

Bill,
I think I can explain what you are seeing there. On the old 901/914 LSDs, and including the early 915 LSDs they used what we call a wave washer. It wasn't a belleville washer in the sense that we use today. Just like what we call the plain plates (outer discs in Porsche's terminology) the wave washer (outer disc {dished profile} in Porsche's terminology) was tabbed to rotate with the outer body of the differential. The plate just had a little bit of a bow in it to make it act similar to a belleville and apply some preload. So they did actually run a friction disc directly against it, though I do 100% agree with your logic on it.

If you look at that skematic, they do start with a friction disc against the cups (aka ramps or pressure rings), but then they double up the plain plates, double up the friction discs, and then run it against the wave washer. Doesn't make sense to me, when you could just as easily pull one of those plain plates over to the other side and put it against the friction disc. And it seems I am not alone in this because when I get sent early ZF LSDs to rebuild, 9 times out of ten they have been restacked that way. And usually the wave washers are cracked in half and doing nothing for preload anyway.

Starting with the mid-year 915 LSDs and the 930s they introduced the belleville washers to it. This was a significant improvement to the design and something we use to this day. Your skematic predates that change in design.

In my eyes, the ZF G50 LSD is the bastard stepchild of the product line. The problem is they made the ramps too thick in the middle to restack it properly as a full thickness 4 plate LSD. With the GKN 993 2 plate LSD Porsche made modifying them easy on us because in that one they stuck in these big thick 3mm spacers on each end and left the ramps in the center smaller. I can easily pull out those spacers and convert a GKN 2 plate LSD to 4 plates and give the customer something that will perform on par with a modern GT3 LSD. But with the ZF version unless you want to significantly cut down the ramps on a surface grinder, and potentially cut through and compromise the heat treat in the process, there's just not enough room inside that LSD with to make it into a fully functioning motorsports capable unit. But for a street car/weekend track toy, it's totally adequate.

Matt, Thanks again, It's very interesting to me to see the changes over the years

[Bill Verburg]

Originally Posted by MCIPSERIC

In a nutshell, what is the differences one would see between a LSD and TBD?

clutch type has an effect during both acceleration and deceleration, TBD only in acceleration.

the net effect on handling is an increase in understeer for clutch type and better braking stability,

since tbd only has effect on acceleration, there will no change in other phases

clutch type also generates more heat

[Matt Monson]

Originally Posted by MCIPSERIC

In a nutshell, what is the differences one would see between a LSD and TBD?

In answering this, I am going to completely disregard Wavetrac's proprietary center section design, which you can read about on their website. It's something that's not part of a typical TBD design and unique to their product only and is best explained by them.

A torque biasing differential is a gear driven differential. What it does is transfer power to the opposing wheel when it encounters limited traction conditions. I'll take the 20/80 limit of your TBD as an example. If you are going through a tight right hand corner in 2nd gear and start to spin your inside wheel, the TBD is going to transfer power. Under those particular circumstances, the inner wheel will be spinning so much and so traction limited that it will quite possibly reach the maximum transfer of the unit. So what you would see there would be 20% power to the spinning inside wheel and 80% of the power to the outer drive wheel that has good traction.

By direct comparison let's look at an LSD in the same conditions. The inside wheel starts to spin. When it spins up fast enough (and technically it's a question of difference in force, not actually wheel speed) what is going to happen is the LSD is going to lock. When it locks, it's locked. As in 50% of the power is being directed to the spinning inside wheel and 50% of the power is being directed to the outside wheel with good traction.

That's the simplest explanation of it. Does that make sense?

[Chakka71]

Originally Posted by Matt Monson

Our own 4 plate LSDs tend to last most PCA Club racers and DE drivers in the 4-6 year range, though obviously that's a gross generalization. If we rebuilt your LSD as 4 plates, I would figure your lifespan on it would be close to half that long.

4 plates will definitely be more effective and noticeable on track. The biggest thing you will notice is more rear end stability under braking.

Took a min. to understand what you meant. Beyond the plate change, is there anything that can be done to the OEM unit to make it better vs. buying a new unit? What makes the GT LSD last so much longer?

[chris_seven]

Bill,

I am not sure that plate diffs generate more heat than TBD types.

As the TBD's have gears in mesh all the time there must always be a loss. On industrial test rig gearboxes with helical gears we always used to estimate this as around 1% of transmitted power per mesh.

This could be checked using 'heat to oil' measurements in the gear box lube system and it wasn't too far away.

I think this means the TBD will be causing some heat build up all of the time while the plate diff will only generate heat in turns when the plates slip.

[Matt Monson]

Originally Posted by Chakka71

Took a min. to understand what you meant. Beyond the plate change, is there anything that can be done to the OEM unit to make it better vs. buying a new unit? What makes the GT LSD last so much longer?

Increasing the number of plates is the only thing that can be done to improve your OEM unit. To put it bluntly, Porsche never should have used that LSD in something carrying the RS moniker. It's a disservice to the marque.

As mentioned above, it's the thickness of the friction material as it is applied to the plates. There's significantly more friction material depth there. It's the same as if you had a brake pad that started with 15mm of material versus 10mm, the former is going to last longer between changes.

Chris_Seven,

An LSD generates way more heat than a TBD. When people add our LSDs to a vehicle that was previously an open differential it can raise the gearbox temps by 20 degrees or more. Adding a TBD only increases the temps by a few degrees. The friction discs in an LSD are always sliding against each other generating heat. In fact the only time they aren't sliding is when the LSD is locked.

[Jon B]

A very interesting discussion!

TBD's are not used in any upper level racing venues I'm aware of. As already mentioned, they have minimal lock on coast, and minimal if any adjustment feature. They also don't hold up well in high-torque applications.

I spent 3 years building transmissions in CART/Indycar, using multi-disc LSD's with variable ramps for road courses. Formula cars run very little lock on coast, as opposed to Porsches. Our diffs (Swift) ran negative preload, adjustable through an access plug on the transmission. We only defined the set-ups in ramp angle degrees and friction faces, never percentage of lock. Keep in mind, a 30 degree ramp develops less thrust/force on coast than power, so I'm curious to know what "80/80" means in terms of ramp angles- the same? Yes, multi-disc LSD's generate heat in use- that heat is power loss. They also wear, requiring frequent adjustment or rebuilding, depending how serious your racing is.

A Guard LSD, in my opinion, is far preferable to a ZF for an older Porsche on the racetrack. New diffs, later technology, more adjustment options, parts and service.