The Lotus Evija requires enormous computing power just to handle like a normal car
You will have recently read about the incredible Lotus Evija hypercar: road test number 5763. That should be my new PIN number, because I will certainly never forget it.
Never before, not in 15 years in this job, have I felt high-speed acceleration like it. Never have I needed to have a word with myself in order to simply keep my right foot pinned – all the way to infinity and beyond. I’m not sure Dirk Benedict had that problem when launched out of his mothership in Battlestar Galactica. Either way, the experience must have felt fairly similar.
Here, though, I would like to talk about asymmetrical torque vectoring. I’ve lost days of my life dreaming about it, but what would it actually mean?
The Evija has four-wheel independent torque vectoring – and what is, I now appreciate, a pretty well-realised system – but it doesn’t do any of the trick ‘spin on the spot’ stuff. Not that I mind. This car has taught me the size of the notional gap between the dynamic potential of this technology and the reality faced by those who just have to make it work.
The problem is that unless you’re going to fit any car that happens to have four independent electric motors with as many independent accelerator pedals (go on, wiggle that fourth toe), controlling those motors at times when you would rather they weren’t so independent is always going to be a challenge. And that, by the way, is a lot of the time.
Take a conventional car, with normal axles front and rear, going around a corner. It’s moving at a constant 70mph, yet all four wheels need to spin at slightly different speeds, because each is ascribing a path of a slightly different radius. That wheel speed ‘delta’ is what a conventional mechanical differential allows for.
It also delivers something called axle drive equalisation. So now you’re cornering quickly enough to transfer weight onto the outside wheels. If they’re driving forwards, the inside ones may lose grip and start to spin. But as they do that, the open differential driving them will automatically divert torque to the unloaded side of the car, like a blow-off valve letting pressure out of a closed system. The car itself doesn’t suddenly yaw into a slide, as it otherwise might, but stays stable.
Now imagine a much more complex model in which that open diff is constantly correcting for something. Allowing for changes in road camber and steering angle, bump, rebound and surface grip. It’s just doing it; you barely know, because it’s one of the cleverest passive mechanical systems that a car has.
Now take all that passive equalisation and seamless compensation and consign it to history. Imagine driving a quad-motor EV that sought simply to put equal torque at each wheel. Powerful or not, it would be stubborn, unstable, hyperactive, unpredictable – a mess.
The way that cars like the Evija make up the difference is with a lightning-fast ‘chassis brain’ that’s constantly monitoring wheel speed and constantly reviewing steering angle, ‘ground speed’, yaw rate, throttle position and/or brake pressure, and then adjusting torque at each motor on the basis of strategies programmed into it, simply to send the car where you intend to go.
It’s all computing power. There’s no mechanical leg-up; no such thing as default, built-in chassis stability. Compared with a conventional road car running a normal stability control system, it’s almost dumb projectile versus guided missile.
It’s also clearly not some god-like stability control system that you would ever want to turn off. If the Evija is any guide, the tuning objective is simply to end up with a natural-handling car. One that doesn’t feel – on track, at least, and near the limit of grip – like it’s driving you.
One that is subtle and progressive with its wheel speed ‘regulations’; that knows not only what the car is doing and how to keep it on line but also your intentions – your mood, even.
Sounds impossible, doesn’t it? Might this, I’m now wondering, be why so many performance brands – Porsche, Polestar, BMW, Hyundai, Kia and others – have generally preferred to combine fewer motors with vectoring diffs in their ‘hot’ EVs, rather than trying to crack the full four-motor shebang?
Might really good, fully independent torque vectoring just be too damned hard? Or is there simply too much potential to know what the hell to do for the best?
Source: Autocar
