To those who run or tried a Torsen diff: how is it?

[Pavlo]

Curious about lap times and general difference on track versus a clutch type. Currently using a 4-clutch custom ramp ZF unit (30*/75* ramps) in my E36M, but always been curious about the performance of a Torsen. On the street I absolutely love them and have had multiple cars with them, but never driven with one on track. I am looking to change gear ratios and debating building an entirely new one or just swap the gear set in my current unit. The current clutch diff does work extremely well, quick and smooth to react with absolutely no issues when lifting thanks to the steep deccel ramp.

Judging by the vast majority of threads about this topic, everyone claims it's inferiority yet no one has actual data or experience to back their opinion. So my questions to those that have tried one: what did you like or dislike about it? Did it have intuitive behavior on track? Any notable lap time difference?

Thanks for the input!

[CoMZ302]

Many years ago I ran one. If the driven tire ever gets unloaded, it takes forever to stop spinning. Maybe great for a perfectly smooth surface, with no camber changes and no curbing.

They're not that hard to try; didn't one E36 come with a Torsen? I know the 2002 3.0L Z3 did. You could always grab one of those and install in an E36 diff housing.

[Pavlo]

The Z3 only did get a Torsen unit which I have considered, but it has no internal resistance to it which would make the one wheel off problem severe. I do run the rear fairly soft with a stock sway bar and have not had too much wheel lift unless hopping curbs.
Did you track with it? Was it the OE version or aftermarket?

[CoMZ302]

Some years of miata's came with Torsens - that's what I tracked (not a BMW).

I tried that stock torsen first and liked it just fine - but then I put in an OSGiken LSD and went a good bit faster. I honestly didn't notice much wheelspin with the torsen. A friend ran the torsen in his miata with great success...

...my current road racer has the DiffsOnline 70/30 or whatever they recommend. I hear it chattering on tight corners in the pits, but the rest of the time, it's magic.

[bionicbelly]

I have one in my Z3. Seems to work just fine, as in it keeps wheelspin down. I have never had a problem with that on track or at autocross. Stock rear bar, not a lot of rebound dialed in on the rear shocks. I don't really have a refence to compare it to though.

[Pavlo]

Awesome, thank you. I think it may be a worthwhile test to conduct for learning purposes. I will be going from a 3.46 to a 3.64 also, so lap times may differ based on the ratio change but should be able to overlay corner speed and lateral accel to see if the road holding abilities have changed.

- - - Updated - - -

...my current road racer has the DiffsOnline 70/30 or whatever they recommend. I hear it chattering on tight corners in the pits, but the rest of the time, it's magic.

I had some chatter with mine and ended up adding some extra friction modifier to Redline gear oil, feels smoother overall and the chatter went away. May be a worthwhile thing to try and is quick and cheap.

[CoMZ302]

I had some chatter with mine and ended up adding some extra friction modifier to Redline gear oil, feels smoother overall and the chatter went away. May be a worthwhile thing to try and is quick and cheap.

Awesome thanks for the tip. I use the Amsoil diff fluid and never had any problems. I think it's just when you're, you know all the way to lock getting around someone or trying to load or whatever.

[Def]

Went from an S15 helical (a torsen type) to a Cusco RS (20 clutch plates, coil springs for preload, completely customizable) and went about 1 s a lap faster on a 2 min lap. The car definitely felt more stable in corners once on the throttle. It inspired more confidence. Trail braking was also really hit or miss on the helical since there is no biasing in that direction. Any small bump would get you inside lockup, and if you're cornering at the limit as you should when trail braking, boom, around you go. With the clutch type diff (1.5 way, so some biasing on braking), it felt really good all the way into the apex on the brakes).

But I did just pick up an AP1 S2000 that has a torsen/helical stock as well, so it's definitely not a BAD diff, I just think clutch types are generally faster on track while sometimes being a bit less smooth/unobtrusive on the street.

[joebaehr]

Running Quaife torsen type for autocross and it works great, but there is that smooth surface and no curb hopping aspect.

[loftygoals]

I've been a big fan of Torsen diffs in BMWs at the track. I prefer the way they feel under trail braking, entry, and mid corner throttle. I find that they don't carry as much angle on corner exit, which could be good or bad. I agree that they don't work as well when using curbs and or driving on rough pavement, though. If unloaded wheel spin is a concern, there are options like the Wavetrac that provide a clutch device to provide pre-load and transfer torque in those situations.

-bj

[Def]

I've been a big fan of Torsen diffs in BMWs at the track. I prefer the way they feel under trail braking, entry, and mid corner throttle. I find that they don't carry as much angle on corner exit, which could be good or bad. I agree that they don't work as well when using curbs and or driving on rough pavement, though. If unloaded wheel spin is a concern, there are options like the Wavetrac that provide a clutch device to provide pre-load and transfer torque in those situations.

-bj

I've had a few of them and really hated them for trail braking. There's no biasing, so you're always going to lock the inside rear WAY before you get any lockup with a clutch diff that's transferring torque across the axle.

Or are all these cars you've driven have ABS? That'll hide some of the trail braking nastiness, but it's still there and a wheel in ABS is providing less grip than one freely rolling.

[Pavlo]

That's a very solid point that I haven't considered, clutch lockup preventing the rear wheels from locking individually. Mine is a 95 so a 3 channel abs, if one wheel sees lock the entire rear axle loses brakes. The Torsen in my Volvo has so much preload that you can't turn the wheels opposite ways without pulling a sting in your butthole, maybe that's how Eaton approached that problem. I do love the natural rotation under power that the car currently has with the clutch type, it holds a steady slip angle that makes the rear leave two thin black lines through certain turns, it's awesome. The car will get a bit angry over abrupt inputs though, which may be because of sudden changes in diff lock. Quaife does sell preload shims for theirs, maybe that would be a way to go in the BMWs. Thank you all for the input, great points to consider.

[loftygoals]

I've had a few of them and really hated them for trail braking. There's no biasing, so you're always going to lock the inside rear WAY before you get any lockup with a clutch diff that's transferring torque across the axle.

Or are all these cars you've driven have ABS? That'll hide some of the trail braking nastiness, but it's still there and a wheel in ABS is providing less grip than one freely rolling.

I've driven cars with and without ABS with a Torsen. We ran Torsens in non-ABS E30's for years. Maybe I prefer it because my very first race car was a Spec 944 with an open diff? I just feel like the salisbury (especially with high lock up on decel) tries to force the car to straighten out when trail braking into a corner. This creates entry understeer. We were fighting this in our E30's. The switch to the Torsen made it go away.

On my S54 M Coupe, the Torsen nearly eliminated the tendency to oversteer on corner exit. I found I could apply full throttle earlier with the Torsen than I could the factory LSD.

One interesting data point is that most modern factory track oriented cars tend to use one of two systems: 1. Torsen 2. Torque Vectoring

The torque vectoring is what the BMW Active M differential is. It is basically an electronically controlled torque biasing differential, where a Torsen is purely mechanical. Torque vectoring includes things like steering angle, wheel speed, and lateral acceleration to vary the amount of torque bias. They are incredibly effective systems!

-bj

[CoMZ302]

On my S54 M Coupe, the Torsen nearly eliminated the tendency to oversteer on corner exit. I found I could apply full throttle earlier with the Torsen than I could the factory LSD.

Wow, I just love the stock LSD on my S54 M Roadster. Did pretty well with it autox'ing nationally too.

[Def]

I've driven cars with and without ABS with a Torsen. We ran Torsens in non-ABS E30's for years. Maybe I prefer it because my very first race car was a Spec 944 with an open diff? I just feel like the salisbury (especially with high lock up on decel) tries to force the car to straighten out when trail braking into a corner. This creates entry understeer. We were fighting this in our E30's. The switch to the Torsen made it go away.

On my S54 M Coupe, the Torsen nearly eliminated the tendency to oversteer on corner exit. I found I could apply full throttle earlier with the Torsen than I could the factory LSD.

One interesting data point is that most modern factory track oriented cars tend to use one of two systems: 1. Torsen 2. Torque Vectoring

The torque vectoring is what the BMW Active M differential is. It is basically an electronically controlled torque biasing differential, where a Torsen is purely mechanical. Torque vectoring includes things like steering angle, wheel speed, and lateral acceleration to vary the amount of torque bias. They are incredibly effective systems!

-bj

I agree the Torsen gives less rear end slip angle on corner exit, but if the car is setup to be really stable on corner exit, I've found the clutch type diff can apply power sooner and harder than the Torsen. If the car has funky camber and/or rear bumpsteer, the torsen might be better on exit by not putting as much slip angle on the rear when on throttle.

2 way clutch diffs give bad trailbraking performance like you described, which BMW diffs are (I don't know why they set them up that way personally), but a 1.5 or 1 way diff is really nice in trail braking. My Cusco RS has about a 55 degree ramp on the accel side and a ~10 degree ramp on the decel side. It's just enough to bias across the axle, but still allow the car to rotate into corners and be stable the entire time. I do think it could use less biasing when turning in off the throttle, so I've thought of bumping down the spring preload, but she's going off to a new owner, so no more development.

I think the OEM BMW diffs don't provide enough lockup on accel, and too much on decel by default. More clutch plates is just going to make the decel problem worse.