61 replies to this topic

[Fat Boy]

This is the video which encouraged me to start this thread.

 

https://www.chassiss...St_p4VZ8ncHJvUg

 

In an attempt to get us talking about _something_ related to race cars, I’m going to attempt to start a differential conversation. We had a version of this conversation about a year ago a year when talking about Detroit lockers. If you have a question about those particular contraptions, I think the information there is reasonable and don’t want to spend a lot of time on a differential that few, if any of us, actually use. So this vid is an interesting analysis of a car with a spool, or locked, rear end. It’s not a differential as much as the lack thereof.

 

When simulating this car (He calls it a ‘musclecar’, so something like an Aussie V8 Supercar?). He finds the locked rear is superior to running several different types of limited slip differential (LSD). He finds it unusual that a locked rear is faster and seems a bit confused by it. To me, it’s not a terribly unusual finding, but I think just looking at the simulation gives us a very narrow view of the issue. So I thought we could kick around this situation a little and perhaps not want to shoot each other after the exchange.

How I see it:

• It’s a tin-top, so a relatively high C.G. which leads to a lot of lateral load transfer. It’s likely the inside rear has little load mid-corner, so it can’t apply a meaningful U/S moment.
• I suspect the initial suspension settings, if not the entire suspension design, is meant to counteract the effects of the spool. Just plopping in an LSD without changing other components likely produces excessive O/S.
• High power cars with little aero and low-ish lateral capacities are very dependent on powerdown performance to produce lap time. Speed traces are very ‘V’ shaped as opposed to more of a ‘U’ shape that you see with single-seaters. Corner minimum speeds and throttle initiation point are much less important than the point at which full throttle is reached. (I cannot stress this last sentence enough.)
• Cars with high C.G. heights often have unstable braking (large longitudinal load transfer). Spools produce braking and corner entry stability. It also eliminates single-wheel locking.
• A ‘Sim’ driver has perfect throttle control. One big issue with a spool is it produces an unstable understeer condition. This may sound unusual, but it’s not. I define stability as the car acting in the manner which the driver expects. If the car is stable at the limit, then its balance is consistent, but the controls just become less and less effective. For instance, a car might have limit oversteer, but if that O/S is progressive w.r.t. the driver’s inputs, then it’s not always a bad thing, because it’s predictable. A small amount of throttle produces a little O/S & a large amount of throttle produces a lot of O/S. Balance, whether understeer or oversteer, is not a surrogate for stability. That’s a separate trait.
• In this scenario, as small amounts of throttle, the car will have significant U/S, but, at the moment the inside rear tire would begin to spin, both rear tires spin and subsequently dump a massive percentage of their lateral capacity producing a large understeer to oversteer shift. Push, push push, going to throttle……Snap loose! This will be a difficult car to drive, particularly at a 90-95% pace.
• The unloading of the inside rear which allows a type of differential-like sliding will be a function of the suspension geometry and chassis tuning, but also of the lateral load transfer at any given moment. Banked, flat & off-camber turns will all have a different balance, as will any corner with meaningfully different tarmac or grip level.
• The above condition is particularly influenced by the rain. Expect to drive the thing tail-out in the wet, because it’s not going to turn on its own. There’s not enough lateral load transfer to lighten the inside rear.
• This is a good example of why the best Sims include a driver in the loop. The spool may be faster in theory, but it remains to be seen as to the actual drivability. Every driver I’ve worked with hates the push-to-snap loose condition. Because of this, they stay on the U/S side of the curve and rarely optimize the potential powerdown advantage. The (time) penalty for overstepping the grip level is too high.
• The idea that a spool can be fast reinforces Fatty’s Theorem of Race Car Handling: Starting from a nominally ‘balanced’ racecar, a 10% increase in U/S will slow the driver 1% while a 1% increase in O/S will slow the driver 10%. It’s much, much better in terms of performance to have a car with U/S rather than O/S. The O/S car might have a slight advantage in a qualifying situation where tires are new and the driver only has to produce a single lap, but over the course of a stint or race, it’s not a comparison. The U/S car will nearly always be the best option.

 

OK, many of us have some free time. Shall we?

[desmo]

U/S has always been recognized as generally faster than O/S hasn't it? Do F/R tire footprints vs. weight distribution (both largely baked into a racecar on any given weekend) dictate to some degree the optimal diff setting? 

 

And how do driver selected electrohydraulic diff settings as in F1 fit into the set-up picture? Can a  starting grid set-up that's not working as hoped in a race be salvaged by changing driver selected diff settings, or is it more likely to just confuse the picture to no benefit?

[Fat Boy]

It's hard to know what others recognize as generally fastest. I think we are all in search of the mythical 'Neutral Balance' race car that does everything well. It's always a compromise and I often find, particularly with higher powered cars, that the closer you get to a 'neutral' car, the slower you go simply because you can't get a good drive off the corner. When I get a new driver I start with a conversation about car balance. I'll tell them that corner entry stability and corner exit traction are my priorities. Because of this, my cars often have mid-corner understeer. I only consider the mid U/S an issue if it's of the unstable variety, which means as they go to throttle it produces a snap-loose. If it's a push that continues toward the exit of the corner, I'll work on it, but I'll do so knowing that it's likely not going to produce big reductions in lap time.

 

I suspect a line from the movie "Days of Thunder", which we all laugh at, has contributed to more ill-handling race cars than an other single utterance on the planet. At the very least, it's had an influence on an entire generation of racers. Robert Duval's crew chief character sits Tom Cruise down and explains, "Loose is fast and on the edge of bein' out of control." I cannot tell you how many people in racing have told me, "Loose is fast." It's Bulls h i t. Loose just makes the driver feel like he's working hard. It's only fast in very unusual conditions.

 

I'm really not a good source for information on F1 differential settings. My guess is they at least have an adjustable preload setting, which is easy to do on-the-fly. Beyond that,there are many other possibilities, but I have no real idea on even the architecture of a Formula 1 diff. An adjustable preload setting would allow the driver to influence the stability of the car, particularly on corner entry. It's influence through the middle of the corner would effect U/S although, less so. At corner exit, it's influence exists, but is fairly muted unless you're at either extreme.

 

I don't use the differential as a common tuning tool. It's one of those things that I tend to find a reasonable setting for and then try to keep it constant. I know some guys who constantly have the diff in and out of the car. My mechanics like me better, and, when the time comes to change the diff, they know it's not just random mortar fire.

 

Can a driver adjustable diff allow the driver to balance a bad setup? I've never been able to. I can make a reasonable setup easier to drive. I can make a reasonable setup harder to drive. I can't make a bad setup a good one. Roll couple distribution, stiffness levels, ride heights & ride control, aero balance, damping, etc. are all better areas to spend one's time in terms of chassis development. I guess that's one reason why I found the simulated switch from spool to diff in the video a little naive.

 

If a chassis is set up for a spool, then you have to assume the other chassis settings are all shifted in the unstable/oversteer direction. You know the rear axle will induce stability, so you may run a relatively high rear roll center, a rearward bias in roll stiffness, a high amount of front aero, etc. in an effort to optimize the package. Just plopping in a LSD in such a situation is unlikely to be good. Logically, the car would have too little stability and too much oversteer. It's all a little bit of a "How long is a piece of string?" type of question as we have no information on the starting settings or initial balance reads.

[desmo]

Diffs fascinate and baffle me more than almost any mechanical system in a car. Looking at an animation of one in action still looks like magic to me even understanding the theory. Have racing diffs, designed under rule constraints that are illogical from a purely engineering point of view, completely diverged from road car diffs with none of the artificial constraints of racing regs? Or do road car engineers still fiddle with ramp angles and such? 

[gruntguru]

Maximum drive traction and latacc occur when slip % and slip angle are the same on each drive wheel. No passive differential can achieve that. In steady state cornering, equal slip % requires the inside wheel (turning slower) to receive less torque. (For equal slip % drive torque must be distributed in the same proportion as weight distribution). Passive LSDs can only bias torque towards the slower wheel. This means that once the mid-corner drive requirement is high enough for the LSD to start biasing torque towards the outside wheel, the inside wheel must be turning faster than the outside wheel, its slip % will be very high and therefore unable to provide useful drive or latacc. At this stage a locked diff will be outperforming it - the inside wheel turning at the SAME speed as the outside rather than FASTER than the outside.

Edited by gruntguru, 18 March 2020 - 22:38.

[Wuzak]

When simulating this car (He calls it a ‘musclecar’, so something like an Aussie V8 Supercar?).

 

The car pictured is a V8 Supercar - a BA Ford Falcon from what appears to be 2007.

 

In the V8Supercars I wonder if the choice of a spool was not due to any scientific reasons, but rather for other reasons such as economics.

 

Essentially I have always thought of V8Supercars as being designed around Mt Panorama, which is a big, fast and open race track, with few fiddly slow bits. The need for some sort of LSD on such a track I would think was not great. Hence the spool was mandated at the start, saving the teams money from what would be another area of development. The spool is mandated to this day.

 

At other, tighter circuits, the cars can struggle in the slow speed corners with understeer.

 

I haven't watched the video yet, but does the fact that the weight balance of the V8Supercar is forward (more so then than now) and the wheels and tyres are the same size all around play a role in advantages of the spool over LSD in that application?

[Kelpiecross]

  Beside the point a little -  but the "Ford"  and "Holden"   Supercars probably do not have even one item that came from the Ford or Holden factories.   nevertheless I am  a "Ford" supporter.  

Edited by Kelpiecross, 19 March 2020 - 12:08.

[Greg Locock]

20 years ago, roughly, the bodies were actually made on the production line. However that bit of nonsense stopped fairly quickly and they became silhouette cars. I can't think of any non trivial component that would be in common between the taxicab racer and the taxicabs. Both the Fords and Holdens used the Ford EA suspension (watts link 4 arm rear, SLA long spindle front), not Holden (Panhard and MacP). I think they used a reverse watts, centre to body rather than centre to axle which is what the production cars had.

[Fat Boy]

Maximum drive traction and latacc occur when slip % and slip angle are the same on each drive wheel. No passive differential can achieve that.

Well, no & yes.

 

You wouldn't actually have the same slip angle and slip % on both drive wheels because the normal loads are significantly different, so the optimum would be different side to side. In general, the lighter loaded tire would want less slip and & slip %.

 

Having said that, you're correct in that no passive diff can produce this. I can really only speak to mechanical differentials, but, in most situations, we operate somewhere in the gap between an open differential and a spool. I do know that there are some OEM's which apply individual brakes to produce a similar effect while using traditional mechanical components.

[Ross Stonefeld]

U/S has always been recognized as generally faster than O/S hasn't it? Do F/R tire footprints vs. weight distribution (both largely baked into a racecar on any given weekend) dictate to some degree the optimal diff setting? 
 
And how do driver selected electrohydraulic diff settings as in F1 fit into the set-up picture? Can a  starting grid set-up that's not working as hoped in a race be salvaged by changing driver selected diff settings, or is it more likely to just confuse the picture to no benefit?

 
I've often wondered why diff adjustments aren't talked about more. Isn't it somewhat 'free' balance adjustment? At least in terms of fine tuning the balance on the day/that run. Changing aero or suspension settings has knock-on effects that need to be chased(ask me about my iRacing spiral of doom on an Indy500 setup sometime...)?
 
 

It's hard to know what others recognize as generally fastest. I think we are all in search of the mythical 'Neutral Balance' race car that does everything well. It's always a compromise and I often find, particularly with higher powered cars, that the closer you get to a 'neutral' car, the slower you go simply because you can't get a good drive off the corner. When I get a new driver I start with a conversation about car balance. I'll tell them that corner entry stability and corner exit traction are my priorities. Because of this, my cars often have mid-corner understeer. I only consider the mid U/S an issue if it's of the unstable variety, which means as they go to throttle it produces a snap-loose. If it's a push that continues toward the exit of the corner, I'll work on it, but I'll do so knowing that it's likely not going to produce big reductions in lap time.
 
I suspect a line from the movie "Days of Thunder", which we all laugh at, has contributed to more ill-handling race cars than an other single utterance on the planet. At the very least, it's had an influence on an entire generation of racers. Robert Duval's crew chief character sits Tom Cruise down and explains, "Loose is fast and on the edge of bein' out of control." I cannot tell you how many people in racing have told me, "Loose is fast." It's Bulls h i t. Loose just makes the driver feel like he's working hard. It's only fast in very unusual conditions.
 

 

Anecdotally, and highly personally, and I'm not telling you anything you wouldn't have experienced with lots(most?) drivers; I reckon there's two things going on. 1. It feels fast/like you're doing something. The car feels fast. Both in responsiveness and on-the-edge-is-fast-ness. But you're generally 'under' the car's performance early on so as you go quicker it evolves into "**** calm this down yeah?" and you migrate back to where you should have been. If you're paying attention.

 

The continuation of 'it feels like something is happening' leads us to 2) how hard is is to learn to wait in a race car.

 

Whether that's U/S or delaying power on(different from the point where we reach full throttle) or backing up or otherwise softening the entry for a good exit, waiting for the tires to come in or for a long run to play out on an oval, etc etc. It's the easiest thing to do but the hardest thing to discipline yourself for. Also see: fiddling with the setup in general. The need to 'do something'. How many people have screwed up a car or their weekend when the pressure was on or things were going too well because they just couldn't sit on their hands? 

 

It was such an eye opener the first time someone explained "Up your entry speed and you'll have less understeer" that I still remember the conversation 19 years and a few months on. I was overslowing the corner in one of those brush-brakes-and-turn situations, which meant I was going to power too soon and too hard, lightening the front end as I got to an off camber exit, etc etc. 

 

Guess what happens if you brake a tiny bit later? Or at the same spot but squeeze lighter? It feels like you went into the corner an entire gear higher. You dare not touch the throttle so you have to wait but...oh, hmm, it's turning. You start feeding throttle rather than tapping it on. Suddenly you qualify on the third row, lead your first laps, set a fast lap, and get a trophy.

 

Crazy isn't it? 

 

 

[Fat Boy]

Diffs fascinate and baffle me more than almost any mechanical system in a car. Looking at an animation of one in action still looks like magic to me even understanding the theory. Have racing diffs, designed under rule constraints that are illogical from a purely engineering point of view, completely diverged from road car diffs with none of the artificial constraints of racing regs? Or do road car engineers still fiddle with ramp angles and such? 

The vast majority of road cars use a pretty simple open diff. The next step is an LSD like the GM Positraction or Ford Trac-Lok. Both are essentially an open differential with clutch packs on both sides. The meshing of the pinion and side gears in a differential produces a spreading force. In an open differential, this spreading force is taken on a thrust bearing. In one of these simple LSD's this side force presses the clutch plates together which produces a side-to-side locking force.

 

To change the locking force the OEM's can change the cut on the pinion and side gears to increase or decrease the spreading force. My guess is that this is not a common change. They might change the stacking of the clutch packs or change the materials of the clutch plates, but that's a much easier change than producing all sorts of different pinions and side gears angles for each platform/model.

 

My guess is that the biggest tuning tool an OEM might play with is preload. In my experience, that's the coarse-threaded tuning knob in a differential. The GM diff has a plate with springs on it. The Ford preload spring is an 'S'-shaped bit that fits between the side gears. The preload setting produces a spreading force on the side gears that gives us some clutch locking under all drive conditions. The overall effect to the car is a yaw damper. It slows the car's turning, specifically when in lateral accel is low (i.e. in phase one of the corner or during transients), but the influence of preload once the gear spreading load increases becomes a smaller component of the overall locking from side-to-side.

 

 

The side gear differentials act like the ramps which are common in racing differentials. I don't know of any OEM differential, outside of some random exotic, that has ramps. The advantage of a ramp-style diff compared to a side-gear type is that you can independently change the locking characteristics from coast to drive. In a side-gear differential, the locking is the same forward and backwards. I think that any car which would have a ramp-style differential would make the step to some sort of electronically controlled option. After digging around a little, I see Ford is using a Torsen-style (locking action produced by worm gears, think Quaiffe or the original Gleason). These are less wear sensitive and probably (depending on how it's built) have a greater locking ability than a side-gear style differential, but it has the same coast/drive symmetry.

 

Ultimately, I can only guess what OEM's do based on what they tend to produce. My guess is, they are heavily influenced by price point and magazine articles. If they can make a new car handle well with a relatively inexpensive differential (side-gears + clutches), then I doubt they stray far from this 'easy' answer which produces 60% of the performance of a much more expensive, wear prone and difficult to develop component which only a tiny fraction of the customers will appreciate.

[Fat Boy]

 
I've often wondered why diff adjustments aren't talked about more. Isn't it somewhat 'free' balance adjustment? At least in terms of fine tuning the balance on the day/that run. Changing aero or suspension settings has knock-on effects that need to be chased(ask me about my iRacing spiral of doom on an Indy500 setup sometime...)?
 
 

 

Anecdotally, and highly personally, and I'm not telling you anything you wouldn't have experienced with lots(most?) drivers; I reckon there's two things going on. 1. It feels fast/like you're doing something. The car feels fast. Both in responsiveness and on-the-edge-is-fast-ness. But you're generally 'under' the car's performance early on so as you go quicker it evolves into "**** calm this down yeah?" and you migrate back to where you should have been. If you're paying attention.

 

The continuation of 'it feels like something is happening' leads us to 2) how hard is is to learn to wait in a race car.

 

Whether that's U/S or delaying power on(different from the point where we reach full throttle) or backing up or otherwise softening the entry for a good exit, waiting for the tires to come in or for a long run to play out on an oval, etc etc. It's the easiest thing to do but the hardest thing to discipline yourself for. Also see: fiddling with the setup in general. The need to 'do something'. How many people have screwed up a car or their weekend when the pressure was on or things were going too well because they just couldn't sit on their hands? 

 

It was such an eye opener the first time someone explained "Up your entry speed and you'll have less understeer" that I still remember the conversation 19 years and a few months on. I was overslowing the corner in one of those brush-brakes-and-turn situations, which meant I was going to power too soon and too hard, lightening the front end as I got to an off camber exit, etc etc. 

 

Guess what happens if you brake a tiny bit later? Or at the same spot but squeeze lighter? It feels like you went into the corner an entire gear higher. You dare not touch the throttle so you have to wait but...oh, hmm, it's turning. You start feeding throttle rather than tapping it on. Suddenly you qualify on the third row, lead your first laps, set a fast lap, and get a trophy.

 

Crazy isn't it? 

This is 100% true and I've got so many stories about inexperienced drivers chasing a car with 'understeer'. It's a really nasty feedback loop. It generally goes something like this:

 

The car has a little bit of entry instability, but the driver doesn't realize it. They underachieve on the entry and, naturally, are not at the appropriate mid-corner speed. They can recognize the car is not at it's cornering limit, so they apply a small amount of throttle artificially early to get the cornering speed higher. This ~20% throttle shifts longitudinal load rearward and locks the differential producing understeer. This understeer isn't 'real'; it's driver induced. It's also easy to recognize, so when the engineer asks about the balance, the driver will tell you, "Mid-corner understeer stops me from going to throttle."

 

The engineer wants to help the driver, so he'll do something to take rear grip off the car, let's say a softer FARB. Now the car is even less stable on entry. To avoid spinning on corner entry the driver is now turning in earlier and softer. That means our driver arrives at mid-corner while having done very little actual turning, and we're approaching corner entry even slower. The driver has to add _a lot_ of steering just to negotiate the corner and this is means *even more* understeer. What makes it even nastier is that while our driver goes to throttle the car will understeer up to the point where the rear tires start to spin. When we get wheelspin, now the car has a big snap loose. The driver come back and says, "Well, I've got understeer AND oversteer." They still don't understand how corner entry instability is the precursor to all of it.

 

In a slow-ish corner (1st or 2nd gear), you have to be able to carry a ton of entry speed and then get off both brake and throttle. Slow the car with the steering. If you've done it right, you can just wait for the car to do it's work. This is very difficult for some to accept, but often the best move is to do nothing. If you've carried the correct entry speed, there will be nothing in your head telling you to add throttle. Once the car has done it's rotation and you're able to begin straightening the wheel, then you can apply throttle. It's important to be able to apply throttle in a relatively quick manner. It will depend on the car, but somewhere in the range of 50-100 degrees of throttle per second.

 

Fatty's golden rule for low speed corners: The point at which the driver initiates throttle is inconsequential, but the point where the driver reaches full throttle is vital. If applying throttle earlier means you get to full throttle later, it's almost always slower.

[Fat Boy]

20 years ago, roughly, the bodies were actually made on the production line. However that bit of nonsense stopped fairly quickly and they became silhouette cars. I can't think of any non trivial component that would be in common between the taxicab racer and the taxicabs.

That makes Porsche SuperCup cars all that more impressive. They get a fair way through the production line before being pulled off at random and turned into a pretty damned nice race car. The GT4 cars make is significantly farther.

[Fat Boy]

The car pictured is a V8 Supercar - a BA Ford Falcon from what appears to be 2007.

 

In the V8Supercars I wonder if the choice of a spool was not due to any scientific reasons, but rather for other reasons such as economics.

 

Essentially I have always thought of V8Supercars as being designed around Mt Panorama, which is a big, fast and open race track, with few fiddly slow bits. The need for some sort of LSD on such a track I would think was not great. Hence the spool was mandated at the start, saving the teams money from what would be another area of development. The spool is mandated to this day.

 

At other, tighter circuits, the cars can struggle in the slow speed corners with understeer.

 

I haven't watched the video yet, but does the fact that the weight balance of the V8Supercar is forward (more so then than now) and the wheels and tyres are the same size all around play a role in advantages of the spool over LSD in that application?

I didn't really know if the picture was an accurate representation of the car being modeled, but it's close enough.

 

I really don't know the genesis of Supercar rules. It's likely the spool was something that a lead tech official liked and it reduced tech/cost/tuning variables. That's a complete guess, though.

 

I think the package as a whole is what makes this particular sim want a spool. The tires are for sure a component, but it's going to be all the spices in the soup.

[Wuzak]

  Beside the point a little -  but the "Ford"  and "Holden"   Supercars probably do not have even one item that came from the Ford or Holden factories.   nevertheless I am  a "Ford" supporter.  

 

 

20 years ago, roughly, the bodies were actually made on the production line. However that bit of nonsense stopped fairly quickly and they became silhouette cars. I can't think of any non trivial component that would be in common between the taxicab racer and the taxicabs. Both the Fords and Holdens used the Ford EA suspension (watts link 4 arm rear, SLA long spindle front), not Holden (Panhard and MacP). I think they used a reverse watts, centre to body rather than centre to axle which is what the production cars had.

 

The cars have only been truly "silhouette" since 2013.

 

From the start (1993) the cars were stock body shells with the 4 link and solid axle. And an extensive roll cage, which soon became like a space frame chassis inside and connected to the shell.

 

In the late 1990s/early 2000s the equalization between the two manufacturers became a problem. Ford dominated a season or two and had massive aero penalties imposed, which led to Holden dominating for a few years, leading to them having their aero changed. However, this was less drastic than had been for Ford, so there was some suggestions by Ford that they may quit.

(About this time Ford was developing an aero kit for the AU Falcon in the USA. I believe this was disallowed, and the AU ended up with a kit that was essentially carried over from the EL racer.)

 

The first move to balance the performance was mandating the front splitter and diffuser. 

 

The second part was Project Blueprint, which standardised the front (double wishbones, were previously McPherson Struts on the Holden) and rear suspension (4 link as before, but with mandated pick-up point locations). Wheelbase was standardised, which wasn't an issue when the rules were introduced in 2003, but required the VE Commodore to be shortened by more than 100mm (4 inches) to fit the rules. But the shells were production body shells taken off the line and modified by the race teams.

 

The standard space frame chassis was introduced in 2013. It had standard suspension all around, though the teams would use their own springs and shocks. The cars featured IRS for the first time, coupled with a rear mounted gearbox. Still kept the spool diff.

 

Most of the body panels were replaced by plastic, CF or GF parts. But, oddly, they still use some OEM panels.

 

Kelly Racing has been kind enough to put the build of their Mustangs on YouTube

 

The standard chassis is why the Supercars Mustang is so hideously disfigured!

[desmo]

I was looking at the last Piola F1 tech book last night and noticed there are *three* driver-variable diff settings on the Ferrari steering wheel shown; one for corner entry, one for corner exit, and one for "hi-speed". Do they *really* let the driver muck about with those in-race without direction from the pits? And even these electrohydraulic diffs, per Peter Wright, are simple and tightly controlled compared to the anything goes stuff allowed in WRC cars, which obviously have a diff on each axle.

 

And the impulse of having to "do something" even if you aren't sure what it is brings to mind the (rather timely, I guess) Blaise Pascal quote, ""All of humanity's problems stem from man's inability to sit quietly in a room alone". Amen.

[Ross Stonefeld]

What would it do to the car if you went too far other than make you uncomfortable? I wouldn't fiddle with engine braking settings or fuel or etc without clearance. 

 

I'm guessing at high speed with all the DF and little traction worries you wouldn't mind it being really pointy. But how does the diff know what kind of corner you're in? Or maybe it's effectively a full throttle cornering setting? 

 

I remember in Zanardi's biography he talked about being Benetton test driver and Schumacher's diff preferences made the car turn 'like a bullet'. But that would have been during the active suspension era, though I think the Benetton was somewhat crude. 

[gruntguru]

Well, no & yes.

 

You wouldn't actually have the same slip angle and slip % on both drive wheels because the normal loads are significantly different, so the optimum would be different side to side. In general, the lighter loaded tire would want less slip and & slip %.

 

Having said that, you're correct in that no passive diff can produce this. I can really only speak to mechanical differentials, but, in most situations, we operate somewhere in the gap between an open differential and a spool. I do know that there are some OEM's which apply individual brakes to produce a similar effect while using traditional mechanical components.

Slip angle and slip % are both dimensionless and to a reasonable approximation, the optimum value of each is independent of normal load.

 

The "perfect" utilisation of the driven tyres during steady-state cornering is when drive and latacc are both shared in proportion to normal force. "To a reasonable approximation" that means slip % same on each tyre and slip angle same on each tyre.

Edited by gruntguru, 21 March 2020 - 06:37.

[gruntguru]

I remember in Zanardi's biography he talked about being Benetton test driver and Schumacher's diff preferences made the car turn 'like a bullet'. 

Interesting turn of phrase. "Bullets" are not renowned for their ability to turn (except perhaps on the X axis).

[Ross Stonefeld]

Probably the sharpness? I know what you mean about it being a weird description but at the same time I understood what he meant.

[Fat Boy]

I was looking at the last Piola F1 tech book last night and noticed there are *three* driver-variable diff settings on the Ferrari steering wheel shown; one for corner entry, one for corner exit, and one for "hi-speed". Do they *really* let the driver muck about with those in-race without direction from the pits? And even these electrohydraulic diffs, per Peter Wright, are simple and tightly controlled compared to the anything goes stuff allowed in WRC cars, which obviously have a diff on each axle.

 

And the impulse of having to "do something" even if you aren't sure what it is brings to mind the (rather timely, I guess) Blaise Pascal quote, ""All of humanity's problems stem from man's inability to sit quietly in a room alone". Amen.

I've worked with 1 diff that had 3 external adjustments. I'm going to struggle to explain it correctly, because I always have a pen in my hand when I'm explaining this type of thing. Let me preface this by saying I have _NO_ real knowledge of what F1 teams run. (Although there are former members on this site that do. Maybe they'll be free for the next couple weeks?)

 

Anyway, my guess is that the 'high speed' setting would be the preload adjustment. The more preload on the diff, the more stable the car. Like I said earlier, a good way to think of preload is as of a total car yaw damper. It slows the response, which is almost always a good thing at high speed.

 

The corner entry and corner exit adjustments may be adjusting something that is commonly referred to as 'negative preload', which is a misnomer that I really don't like. I prefer the terms 'drive ramp delay' and 'coast ramp delay'. These are preloaded springs (in this particular case, probably hydraulics) which resist the spreading force of the ramps (one drive/one coast). As we increase 'delay' we let the diff act open (with the exception of whatever preload is present) until the spreading force from the ramp overcomes the force of the delay spring where it then acts as a normal ramp-based differential.

 

The preload setting is one of those I'd let the driver adjust as he sees fit. It would be like a T/C setting or anti-roll adjustment and, unless I had information that the driver was not aware, I'd let them do what they want to do. The delay settings I might be a little more protective, but when you see them labeled 'Corner Entry' and 'Corner Exit', it certainly seems as if the driver has free reign. It goes back to my statement in a different thread. There are no stupid racing drivers at the higher level of the sport. They might be impulsive and risk-takers, but they're also intelligent.

 

As far as the good Mr. Pascal...indeed.

[Fat Boy]

What would it do to the car if you went too far other than make you uncomfortable? I wouldn't fiddle with engine braking settings or fuel or etc without clearance. 

 

I'm guessing at high speed with all the DF and little traction worries you wouldn't mind it being really pointy. But how does the diff know what kind of corner you're in? Or maybe it's effectively a full throttle cornering setting? 

 

I remember in Zanardi's biography he talked about being Benetton test driver and Schumacher's diff preferences made the car turn 'like a bullet'. But that would have been during the active suspension era, though I think the Benetton was somewhat crude. 

1. I assume you'd change it back? A lot of the settings you just can't know to change without looking at telemetry.

 

2. I've never found *any* driver that likes a pointy car at high speed. There are all sorts of things that you do on ovals specifically to make the car more stable. The most obvious is running a spool. Indycars all run spools with a certain amount of stagger. The more stagger you run, the more the car turns on its own, but having the rear tires locked together gives stability.

 

My goal is to give my driver a car which can be driven aggressively. I *never* want my driver to have to drive on fingertips. I want them to be able to grab the wheel firmly and turn the wheel quickly. On an oval, we often run a slower steering rack pinion because this allows the driver to physically turn the wheel at a faster rate, even if it's not a particularly fast rate at the tire. I did know one guy who said he drives the car with steering wheel load, not position. That guy liked having his 'road course' steering ratio because it put more load in the wheel, which he liked (& this guy was _The Man_).

 

3. For whatever reason, it seems Indycar guys learn to prefer stability and understeer. F1 guys, particularly the top level, seem to require less overall stability and which allows certain car developments which allows them to go faster. I've worked with two F1 level drivers. Both were very, very good, but, ultimately, mortal. When I look at the performance of drivers like Zanardi and Bourdais, I see similar traits. These are just observations; I have no real proof to back this up.

 

I've worked with a bunch of really good drivers, but they've all been 'normal'. It's not necessarily about what makes the fastest race car. It's about what makes the fastest driver/car combo and that means enough stability and understeer to allow the driver to be aggressive with relatively small consequences for minor missteps.

[Fat Boy]

Slip angle and slip % are both dimensionless and to a reasonable approximation, the optimum value of each is independent of normal load.

 

The "perfect" utilisation of the driven tyres during steady-state cornering is when drive and latacc are both shared in proportion to normal force. "To a reasonable approximation" that means slip % same on each tyre and slip angle same on each tyre.

 

Without getting too deep in the weeds in terms of tire dynamics, I’m going to leave it with they’re all different and it depends on the particular tire.

 

Regardless, let’s assume you have a differential which keeps each tire at the optimum slip % and chassis tuning which keeps both tires at their optimum slip angle. What happens when you apply too much throttle? Both rear tires simultaneously break longitudinal traction which causes a dump of lateral capacity at the same instant. When this happens, the driver reports, “Snap oversteer at corner exit.” The vast majority of the time, the drivers aren’t worried about crashing, they’re just trying to go as fast as possible without overstepping the grip level of the tire. When they do, there’s a slide or correction that slows them down. It’s not scary as much as it is annoying.

 

If I get things correct with my diff and chassis tuning, I’ll let the inside rear tire spin up maybe 5-10% faster than the outside rear before enough throttle is applied to lock the diff (& potentially side-step the rear). Is it theoretically slower? Yes. Is it much easier to drive? Yes. Do I actually find it to be slower in practice? No.

 

Further, intentionally letting the inside tire spin up is more benign to tune and drive. It allows the drivers to have a little warning prior to the exit oversteer and overall gives enough predictability to encourage aggressiveness without significant lap time consequences if they get it wrong. I’ll also say that having the differential with a more ‘user friendly’ setting is a real help when it rains. That’s the worst case scenario for a spool.

 

I have had the situation where the diff was too open and the inside rear wheelspin was excessive. The driver reports that the engine buzzes up as if the clutch were slipping. It’s a relatively trivial task to engage another plate on either side of the diff or give it a little ramp angle to get to a good operating window. If it’s early in the weekend, the best plan is to (as the good Mr. Pascal reminds us) do nothing. Track grip will come up after a couple sessions and the problem goes away.

[gruntguru]

If I get things correct with my diff and chassis tuning, I’ll let the inside rear tire spin up maybe 5-10% faster than the outside rear before enough throttle is applied to lock the diff (& potentially side-step the rear). Is it theoretically slower? Yes. Is it much easier to drive? Yes. Do I actually find it to be slower in practice? No.

I assume your experience and comments don't extend to a race car with tuneable torque vectoring? Not sure if you saw this in another thread?

 

http://robotics.ee.u...ntrol-Brown.pdf  See Page 31 onwards. The "non-torque-vectored" comparison is a simple electric drive to each rear wheel ie similar to an open diff up to the point of traction loss (equal torque to each wheel) but different in that it maintains drive to the outside wheel in proportion to throttle setting after loss of traction to the inside.

 

"CONCLUSIONS As predicted, improvements were hard to observe during the test itself but from the data it can be seen that the torque vectoring system is effective. Understeer is still evident but the system was able to vector torque away from the inside wheel while cornering and increase outside wheel torque where it could be utilised. More test situations would show additional effectiveness, such as a chicane test or double-lane-change test. Even with the 2WD simplification of the algorithm, it is clear that a 17% increase in cornering speed for a given 6m radius is a valuable improvement in performance. The results of this test could be extrapolated to suggest improvement in handling would be observable when the system is tested on the 2013 AWD car."

 

Of course 17% isn't going to translate to every cornering situation and every car but there is no doubt equal slip % will increase steady state cornering speed in every case. Another example we have is the McLaren F1 "fiddle brake" saga where the car was significantly faster at every track in spite of the following limitations of the system:

1. It only operated on one side so only assisted LH or RH cornering at any given track.

2. It relied on driver skill to regulate slip %.

 

To this day McLaren use brake actuated torque vectoring on their road cars.

Edited by gruntguru, 22 March 2020 - 00:19.

[Ross Stonefeld]

1. I assume you'd change it back? A lot of the settings you just can't know to change without looking at telemetry.

 

2. I've never found *any* driver that likes a pointy car at high speed. There are all sorts of things that you do on ovals specifically to make the car more stable. The most obvious is running a spool. Indycars all run spools with a certain amount of stagger. The more stagger you run, the more the car turns on its own, but having the rear tires locked together gives stability.

 

My goal is to give my driver a car which can be driven aggressively. I *never* want my driver to have to drive on fingertips. I want them to be able to grab the wheel firmly and turn the wheel quickly. On an oval, we often run a slower steering rack pinion because this allows the driver to physically turn the wheel at a faster rate, even if it's not a particularly fast rate at the tire. I did know one guy who said he drives the car with steering wheel load, not position. That guy liked having his 'road course' steering ratio because it put more load in the wheel, which he liked (& this guy was _The Man_).

 

3. For whatever reason, it seems Indycar guys learn to prefer stability and understeer. F1 guys, particularly the top level, seem to require less overall stability and which allows certain car developments which allows them to go faster. I've worked with two F1 level drivers. Both were very, very good, but, ultimately, mortal. When I look at the performance of drivers like Zanardi and Bourdais, I see similar traits. These are just observations; I have no real proof to back this up.

 

I've worked with a bunch of really good drivers, but they've all been 'normal'. It's not necessarily about what makes the fastest race car. It's about what makes the fastest driver/car combo and that means enough stability and understeer to allow the driver to be aggressive with relatively small consequences for minor missteps.

 

F1 street circuits have less severe bumps, and better paving jobs, than America's best road courses? And I say that fondly. Anyone racing in Indycar goes testing at Sebring pretty early in their learning curve.

 

F1 - "We don't like that kerb" "We'll reprofile it, tonight"
Indy - "You saw the traintracks on the front straight?" "Yeah, but, we're not in charge of the layout of the city"

 

I have this theory that racing in Indycar ruins you in two ways for F1, even if you came out of F3000/F2/whatever. First is you learn to drive a more practical car. Indycars have gotten more docile in the past...I dunno, 20 years? Today's car is way more refined than what people were hanging on to for dear life in 2000. But they're still somewhat unsuitable for a lot of the tracks they race on. Long Beach is *kind of* ridiculous. Also the nature of the budgets and the amount of fiddle, and testing and track time means you can't **** around too much. You do have to get something that works and get on with it.

 

Second is the team treats you like a human being for the most part and an organic piece of the suspension. Plus, ovals. They know the driver's head is as important as any other setting on the car. Or can at least undo your settings.

 

So you go to F1 and they don't care(why would they, it's worked so far), or about you(why would they, you can be replace) and etc. I don't think Montoya and Villeneuve had any fewer problems than Zanardi or Bourdais(or da Matta) but they were different personalities entirely. 

 

F1 teams spend entire seasons scratching their heads in dimly lit dead-ends. Sometimes an Indycar team goes on a tear of a few awkward weekends but no one that is any good or well run stays lost for long. They're not above reversing out. I acknowledge that it's a lot easier to Ctrl-Z in Indycar than F1. For complexity, cost, political, etc reasons.

Edited by Ross Stonefeld, 22 March 2020 - 02:41.

[Kelpiecross]

How does a locked diff affect the handling of a front wheel drive car?   I remember being told that the  Mini Lightweight   drivers (from the 60s and 70s)  said that it was very good for the handling etc.  - but made the steering much heavier.   

[Fat Boy]

I assume your experience and comments don't extend to a race car with tuneable torque vectoring? Not sure if you saw this in another thread?

 

http://robotics.ee.u...ntrol-Brown.pdf  See Page 31 onwards. The "non-torque-vectored" comparison is a simple electric drive to each rear wheel ie similar to an open diff up to the point of traction loss (equal torque to each wheel) but different in that it maintains drive to the outside wheel in proportion to throttle setting after loss of traction to the inside.

 

"CONCLUSIONS As predicted, improvements were hard to observe during the test itself but from the data it can be seen that the torque vectoring system is effective. Understeer is still evident but the system was able to vector torque away from the inside wheel while cornering and increase outside wheel torque where it could be utilised. More test situations would show additional effectiveness, such as a chicane test or double-lane-change test. Even with the 2WD simplification of the algorithm, it is clear that a 17% increase in cornering speed for a given 6m radius is a valuable improvement in performance.

 

I suppose it would be good to preface (too late!) this entire thread with the statement that I’ve never played with any ELSD. So, as far as using torque vectoring to influence handling characteristic, no, I have zero experience on that end of things (as a side note, I have damned little experience with corners of 6m radius, either). The only person I do know that has done that type of thing worked for an OEM and did the work on several different platforms, the last one was a front engine car & is now a mid-engine car (hint, hint). Apparently, you can play all sorts of fun games, but honestly, even though we’ve been friends for a long time, he’s never really spoken about it in specifics to me, only generalities. Having said this, we can do a thought experiment as to how we might use torque vectoring and how it would affect our chassis tuning.

 

Starting with low-speed corner entry, we know that we want to keep the car relatively stable on corner entry to allow the driver to turn the car in relatively late and relatively abrupt. We would want to put our driver in the position of inducing instability as opposed to the car being unstable and him having to deal with it. So in the low-G/high-Yaw rate portion of corner entry, you’d tune the diff to keep the car stable. Having said this, I would also probably have the chassis tuned in a manner which takes advantage of this somewhat artificial increase in stability, which means more front aero %, less front ride/roll stiffness, quicker responding rear, etc. Whatever the driver limit for stability is, we’ll eventually get back to the point where we’re approaching that threshold.

 

When we get to the mid-corner, rolling phase of the corner (high G/moderate yaw rate with no throttle or brake), we’d probably have the chassis tuned to induce just a small bit of understeer and then use our differential to drag the inside rear (or something of that nature) enough to induce turning and bring the car back to neutral steer.

 

It’s always a challenge to get a high-horsepower car to turn well at corner exit (high, but falling G/moderate yaw rate + increasing throttle). Yes, when you can spin the tires to let the car slide, but that’s only going to really work on loose surfaces. On tarmac, you’re going to sacrifice too much lateral capability if you take that approach. The forward drive transfers normal load to the rear of the car and induces understeer a number of different ways. I think I’d have the rear of the car set to produce maximum traction (more understeer than a passive differential) and then challenge the torque-vectoring to turn the car. If we *just* drive the outside rear, we can put a hell of a turning moment in the car, but we’re limiting forward drive to a single contact patch. As we increase the drive on our inside rear, our forward acceleration increases, but we have a lesser ability to apply a turning moment to the car. It’ll be a compromise. You’d drive the inside rear as much as you could to increase inline accel., but as little as possible so you can still keep neutral-ish steering. Presumably, if your electronics are controlling the diff, then you’ve also got a well-tuned T/C so you don’t over-saturate the rear tires.

 

In short, having an electronically controlled differential is not going to appreciably change my setup philosophy (at least not from a purely theoretical standpoint), but I do think it would allow me to go to some extremes that are just not practical with a passive arrangement.

 

[Fat Boy]

 

Second is the team treats you like a human being for the most part and an organic piece of the suspension. Plus, ovals. They know the driver's head is as important as any other setting on the car. Or can at least undo your settings.

 

So you go to F1 and they don't care(why would they, it's worked so far), or about you(why would they, you can be replace) and etc. I don't think Montoya and Villeneuve had any fewer problems than Zanardi or Bourdais(or da Matta) but they were different personalities entirely. 

 

From the outside, it definitely appears that F-1 teams treat their drivers with about as much concern as a used bump rubber. So, it does seem to be an advantage for the driver to be a bit of a twa-t in F-1 (which explains the relative success of Villenueve and Montoya). Like I said, I have limited experience with F-1 level drivers.The ones I have worked with have all been decent human beings, but none of them were world champs. My guess is that F-1 is such a pressure cooker that if you have any self-awareness at all, then it's too much.

[Fat Boy]

 

I have this theory that racing in Indycar ruins you in two ways for F1, even if you came out of F3000/F2/whatever. First is you learn to drive a more practical car. Indycars have gotten more docile in the past...I dunno, 20 years? Today's car is way more refined than what people were hanging on to for dear life in 2000. But they're still somewhat unsuitable for a lot of the tracks they race on. Long Beach is *kind of* ridiculous. Also the nature of the budgets and the amount of fiddle, and testing and track time means you can't **** around too much. You do have to get something that works and get on with it.

 

Couldn't agree more. Having said that, when watching F-1 on-boards, I'm always impressed by their driving. The (~top 1/2 of the grid) F-1 driver's ability to keep the car on edge is pretty incredible. An Indycar is a lot less sophisticated, which is understandable when you consider they have budgets which are 1-2 orders of magnitude smaller. I do think F-1 drivers have a lot less influence on car setup. From what I understand, most have to just drive what they're given and have little say on the matter one way or another. A large portion of this is the cultural differences between the American and Euro racers I think as well. Driver friends of mine who go to Euro-based sports car teams have often commented that this is just how they operate.

 

I understand an engineers inclination to say, "I don't care if your eyes are bouncing out of your head in that bumpy brake zone, the Sim says stiffer is faster and it's what you'll drive." I've just never found it to be an advantage at the actual racetrack.

[Fat Boy]

How does a locked diff affect the handling of a front wheel drive car?   I remember being told that the  Mini Lightweight   drivers (from the 60s and 70s)  said that it was very good for the handling etc.  - but made the steering much heavier.   

Again, you're going to places which are completely uncharted for me, but here's my guess based on a VW Corrado that I had 20 years ago (pretty much the benchmark by which all front drivers are measured, right, LOL?)

 

Inside front wheelspin is a big limiter as far as corner exit is concerned for a front drive car. If you run a spool on the front, you stop that entirely, but I bet the steering does get a lot heavier, particularly a low-G's. By driving the outside front harder, you induce a turning moment to the car which reduces understeer in a reasonably predicable manner. If you do completely blow away the front tires with the throttle I suppose you'd have 'snap-push'? I'm not exactly sure WTF that is, but it sounds like something which wouldn't be that tough to control with the throttle pedal.

 

I bet if we talk to someone with TCR experience we could get a better idea of what works here.

[gruntguru]

 . .  Having said this, we can do a thought experiment as to how we might use torque vectoring and how it would affect our chassis tuning.

 

Starting with low-speed corner entry, we know that we want to keep the car relatively stable on corner entry to allow the driver to turn the car in relatively late and relatively abrupt. We would want to put our driver in the position of inducing instability as opposed to the car being unstable and him having to deal with it. So in the low-G/high-Yaw rate portion of corner entry, you’d tune the diff to keep the car stable. Having said this, I would also probably have the chassis tuned in a manner which takes advantage of this somewhat artificial increase in stability, which means more front aero %, less front ride/roll stiffness, quicker responding rear, etc. Whatever the driver limit for stability is, we’ll eventually get back to the point where we’re approaching that threshold.

 

When we get to the mid-corner, rolling phase of the corner (high G/moderate yaw rate with no throttle or brake), we’d probably have the chassis tuned to induce just a small bit of understeer and then use our differential to drag the inside rear (or something of that nature) enough to induce turning and bring the car back to neutral steer.

 

It’s always a challenge to get a high-horsepower car to turn well at corner exit (high, but falling G/moderate yaw rate + increasing throttle). Yes, when you can spin the tires to let the car slide, but that’s only going to really work on loose surfaces. On tarmac, you’re going to sacrifice too much lateral capability if you take that approach. The forward drive transfers normal load to the rear of the car and induces understeer a number of different ways. I think I’d have the rear of the car set to produce maximum traction (more understeer than a passive differential) and then challenge the torque-vectoring to turn the car. If we *just* drive the outside rear, we can put a hell of a turning moment in the car, but we’re limiting forward drive to a single contact patch. As we increase the drive on our inside rear, our forward acceleration increases, but we have a lesser ability to apply a turning moment to the car. It’ll be a compromise. You’d drive the inside rear as much as you could to increase inline accel., but as little as possible so you can still keep neutral-ish steering. Presumably, if your electronics are controlling the diff, then you’ve also got a well-tuned T/C so you don’t over-saturate the rear tires.

 

In short, having an electronically controlled differential is not going to appreciably change my setup philosophy (at least not from a purely theoretical standpoint), but I do think it would allow me to go to some extremes that are just not practical with a passive arrangement.

Good insights there. I have spent some time thinking about T.V. algorithms for RWD racecars and have come to the conclusion that the ESC approach adopted on road cars is totally unsuitable. A race car needs a system that is entirely controlled by the driver (with no secondary effects such as trying to change the attitude of the car based on an accelerometer input.)

 

At the simplest level a system that sends torque to the drive wheels in a proportion determined by the steering angle. This would be a (very tune-able) 2-D map or lookup table. Such a system could greatly enhance agility and help with issues you mention around turn-in and corner-exit drive and understeer. It would also enable a better approximation of ideal slip % in steady state cornering. With experience the driver would be able to "find" the best slip % by adjusting steering angle - "I want to tighten the turn - I just turn the wheel a bit more - and vice versa."

 

A more sophisticated version would control the ratio of drive-wheel speeds in a proportion determined by the steering angle. If you think about it, the system is matching yaw-rate to steering angle. If the yaw-rate is less than that dictated by the steering angle (understeer) the outside tyre will be operating with a higher slip % (and hence more drive) than the inside, which will produce a turning moment tending to increase the yaw rate and reduce understeer. The same would apply for oversteer. In steady state cornering under power, the system (correctly tuned) would maintain the same slip % at each drive wheel.

 

Either algorithm would probably benefit from throttle as an input. I haven't thought through all the permutations of trailing throttle, off throttle etc - reverting to open-diff is an obvious option.

[gruntguru]

How does a locked diff affect the handling of a front wheel drive car?   I remember being told that the  Mini Lightweight   drivers (from the 60s and 70s)  said that it was very good for the handling etc.  - but made the steering much heavier.   

If you've ever experienced a FWD with torque steer - multiply that by ten. Awful till you approach the cornering limit then the steering lightens up and works well. I spent a little bit of time at the wheel of a dirt oval Mini many years ago. Turning tight in a car park is near impossible - unless you are going backwards - then its near impossible to stop the steering wheel from spinning to full lock.

[Greg Locock]

A friend of mine races off road buggies and wanted to go electric 4wd. In the absence of any better control algorithm I suggested three knobs on the dash, which would control three electronic 'diffs'. These would run from 100% open to 100% locked. The driver plays with them as he drives (it is a short loop) and once we know what he wants we begin to figure out some sort of automation. that car is a year away from completion. 

 

Hayabusa engines, way overpowered, undertired

 

[gruntguru]

If it has four electric motors I would put a steering angle sensor on it and two knobs on the dash - one for L/R bias/steering sensitivity and one for F/R bias.

 

Edit. On second thoughts - you say "100% locked". That would imply closed loop wheel speed control. If that is available I would make the F/R bias knob a PID (velocity,displacement,acceleration) locker (as you suggest?) (with I (displacement) set to zero to avoid "wind up" on tarmac) and control L/R velocity ratio/steering sensitivity with the other knob.

Edited by gruntguru, 24 March 2020 - 01:33.

[Greg Locock]

The motor controllers typically have two inputs, demand velocity and demand torque (like a dyno controller). In mechanical terminology for a pair of motors, that's a spool diff and an open diff. 

[Fat Boy]

 

 

At the simplest level a system that sends torque to the drive wheels in a proportion determined by the steering angle. This would be a (very tune-able) 2-D map or lookup table. Such a system could greatly enhance agility and help with issues you mention around turn-in and corner-exit drive and understeer. It would also enable a better approximation of ideal slip % in steady state cornering. With experience the driver would be able to "find" the best slip % by adjusting steering angle - "I want to tighten the turn - I just turn the wheel a bit more - and vice versa."

...

 

Either algorithm would probably benefit from throttle as an input. I haven't thought through all the permutations of trailing throttle, off throttle etc - reverting to open-diff is an obvious option.

Maybe you're right, like I said, I have no experience with an active diff. If I were going to guess, though, I would guess it's significantly more involved that this, at least if you're going to beat a good mechanical unit tuned well.

 

I have worked with a traction control system that was essentially a throttle-based stability control. It was tough to get working correctly and we had to have constant track support from Bosch by the guy who led the project. We had (at least) 4 wheelspeeds, 3-axis accel, yaw rate, engine RPM, steering & throttle channels to make it work well. The driver also had to drive it in a little bit of unusual manner. He would intentionally give it a little pitch on entry to unsettle the rear then immediately go 100% throttle. From that point on, the car would pretty much hold the line he was requesting based on steering angle. It was a really cool system when it worked.

 

There were plenty of shortcomings. For one, the car had to be reasonably neutral handling prior to us even turning it on. If the car had excessive U/S, then the TC/ESC couldn't 'kick' the rear out to turn the car. It would just pop & fart while grinding the fronts off. The second was we had to find the appropriate yaw angle for the surface/grip level. It required a fair bit of track-specific tuning to make it fast.

 

In terms of lap time, I don't know if I can tell you it was worth X% or anything like that. We definitely felt it was an advantage, but they would often (~1/2 the time) turn it off for qualifying. The drivers felt that if it wasn't working at a very high level, then they could beat it over a single lap. Where it was a particular advantage was over the course of a long run, especially on a high tire wear track or over a double stint.

 

I can also tell you that a lot of 'canned' T/C for present day race cars is pretty mediocre. The driver will always have 1 adjustment (a generic more/less) and  usually has control of 'Slip %' and 'Gain', but even with those adjustments, it can be significantly lacking. Sure, it'll stop you from doing something stupid, but it will also slow you down in the process because the timing of when it steps in and when it releases is rarely what the driver wants.

[Fat Boy]

A friend of mine races off road buggies and wanted to go electric 4wd. In the absence of any better control algorithm I suggested three knobs on the dash, which would control three electronic 'diffs'. These would run from 100% open to 100% locked. The driver plays with them as he drives (it is a short loop) and once we know what he wants we begin to figure out some sort of automation. that car is a year away from completion. 

 

Hayabusa engines, way overpowered, undertired

That looks like a hell of a lot of fun.

 

So are you approaching these as essentially a 'preload only' diffs or with some sort of torque dependence in the 'lock' characteristic? Preload only can be a good, and relatively easy, way to find a what you need for locking characteristics. Like I said earlier, I've always found preload to be the most powerful adjustment. Having said that, I've talked with others who I respect that disagree. It probably has to do with how you go about tuning the rest of the car. In this particular application, I bet you'll end up with a lot of lock on the rear, a rearward bias on the middle and somewhat free on the front. I'd be curious to know how it ends up.

[Fat Boy]

If you've ever experienced a FWD with torque steer - multiply that by ten. Awful till you approach the cornering limit then the steering lightens up and works well. I spent a little bit of time at the wheel of a dirt oval Mini many years ago. Turning tight in a car park is near impossible - unless you are going backwards - then its near impossible to stop the steering wheel from spinning to full lock.

That's a really good point. I bet with any reasonable power level, a locked front would just about rip the steering out of your hands. It made me think of an old road car I had, a 1988 Mazda 323 GTX. It was a cool little econo-box with a 1.6L turbo and AWD. The AWD had a planetary middle diff which gave a 35F/65R torque split. For ~130 HP, it could scoot around pretty good. There was a button on the console that locked the center diff. If you ever hit that in a parking lot, the steering got _very_ heavy. I only ever used when playing around a sand dune area in the desert. I figured if I locked it on tarmac, it'd break.

[Fat Boy]

The motor controllers typically have two inputs, demand velocity and demand torque (like a dyno controller). In mechanical terminology for a pair of motors, that's a spool diff and an open diff. 

That's good to mention. If you have equal drive torque, that's an open. If you have equal drive speed, that's a spool.

[Kelpiecross]

  My interest in the locked diff for front wheel drive is because I have been building (for quite a few years)  a small  "special"  (as we used to call them)  inspired by the little Berkeley  sports car.  Overall  - somewhat smaller than  a B  -  64ins wheelbase (B is 70), weight about 400 lbs (B is about 600.  Engine - 4 cylinder bike engine about 600cc.  Overall appearance inspired by CanAm  cars of the 1970s.  About 90% done.  Suitable diffs are expensive, heavy and get in the way etc.  - so I decided right from the start to have no diff.  

  Things to try:  steering arms parallel  (no point in Akerman if both wheels are turning at the same speed),  Equal  length drive shafts and possibly electric power steering if steering effort too high.     

[Allan Lupton]

  I decided right from the start to have no diff.  

  Things to try:  steering arms parallel  (no point in Akerman if both wheels are turning at the same speed),  Equal  length drive shafts and possibly electric power steering if steering effort too high.     

Diffless will make the inherent power-on understeer of a front-driven car far worse of course. If you can generate a decent cornering force, the inside wheel can slip but, as you say, not much point in Ackermann.

Drive shaft length has no effect, with or without a diff.
 

Edited by Allan Lupton, 25 March 2020 - 10:02.

[Kelpiecross]

I thought I read somewhere that drive shaft length does have an effect on torque-steer - don't know why it would though. 

[Greg Locock]

Unequal halfshaft length is on a  laundry list of things not to do if you have a torque steer problem. One of the causes is that if the halfshaft angles vary side to side there is a torque exerted at the outboard CV that is different side to side. That is to make the drive torque go round the corner from halfshaft to stub axle there must be a torque  applied to make that happen. However that certainly isn't the whole story, and I have several powerpoints to prove it!

[Allan Lupton]

Unequal halfshaft length is on a  laundry list of things not to do if you have a torque steer problem. One of the causes is that if the halfshaft angles vary side to side there is a torque exerted at the outboard CV that is different side to side. That is to make the drive torque go round the corner from halfshaft to stub axle there must be a torque  applied to make that happen.

I cannot see why output torque should not equal input torque for any shaft.

I also cannot see why, apart from second order effects of a non-CV joint, angularity would affect input torque.
 

[Greg Locock]

How do you get a torque to turn a corner?

[desmo]

Vectoring!  

[Greg Locock]

Give that man a goat. Indeed. Torques add in 3d just like forces.  So if my halfshaft is at 6 degrees to the stub axle, there is an external torque on the CV joint of approximately 1/10 of the shaft torque. 

[Kelpiecross]

Give that man a goat. Indeed. Torques add in 3d just like forces.  So if my halfshaft is at 6 degrees to the stub axle, there is an external torque on the CV joint of approximately 1/10 of the shaft torque. 

 

 Now I remember how it works.  But equal lengths/angles etc. don't solve all problems, as body roll, one wheel dropping into hole cause torque steer still.   Car makers go to a lot of trouble and expense to get equal length  shafts - so it must be important.  

[Fat Boy]

Give that man a goat. Indeed. Torques add in 3d just like forces.  So if my halfshaft is at 6 degrees to the stub axle, there is an external torque on the CV joint of approximately 1/10 of the shaft torque. 

This is analogous to the issue which gives rear suspension an anti/pro squat quality when there is a reasonable fore/aft offset between the gearbox output flange and the stub axle on the upright.

[desmo]

1/10 of shaft torque could be rather a lot.  Might that be enough to affect the aero via pitch changes? I picture a car's rear squatting under power and trying to lift the rear under engine braking adding/subtracting DF.