This article was first published in 2002.
An appropriate subtitle for this article could've been 'Non-Linearity of Automotive Ergonomic Interfaces'... but everyone would have stopped reading by now. Instead, let's start by looking at an example of the way in which this technology is very relevant to modified and high performance cars.
Performance? It's Easy!
I was talking to someone the other day who owns a turbocharged 2.5-litre six cylinder car. He mentioned in glowing terms a friend of his who has the same sort of car - but the other car's producing 250kW, about 35 per cent more power than standard. Wasn't that just fantastic how his mate had soooooo much power? That modified car, Gawd it kicked arse!
I didn't share his enthusiasm - to me it was more a case of, 'So what?'
The bloke I was talking to seemed somewhat crestfallen when I pointed out that getting extra power is quite easy. And after all, if he wanted lots more power for his own car, he could just buy a bigger turbo, programmable engine management, and a bigger exhaust. Along with the purchase of a few other bits and pieces, if he wanted to, he could probably use that combination to develop 300 or even 350kW.
In terms of driving the car on the road, it would almost certainly be quite horrible, with the very large turbo taking perhaps half of the engine's rev range to come on boost. The power output of the engine would look be very non-linear - like this graph. Driving it fast and well down a twisty road would take enormous skill. As with the 500hp 2-litre turbo Group A race cars of the Eighties, the driveability would be shocking - though the sheer performance unquestionable.
And the same story can be applied to almost any area of automotive engineering - improving performance is quite easy. Handling, brakes, steering - you want higher performance? Just spend the money and get the same mods made that have been done tens of thousands of times before.
But just try getting this happening within an envelope that doesn't degrade driveability.... Now that's hard!
The key attribute of driveability is deceptively simple - the car remains pleasant and easy to drive.
And, while we could go on in this preamble for a very long time, 'easy to drive' means - amongst other things - that the car behaves in a predictable fashion; that the required driver inputs are no more or less than expected; and that as part of the control process, the car clearly and unambiguously communicates with the driver.
For example, only an idiot thinks that a road car which handles well should require the skills of an expert driver before that excellent handling prowess can be displayed. To get the utmost out of the car, you'd certainly expect an expert to gain higher performance than an amateur - but even the amateur should be able to go faster around corners in car than handles well than in a car that handles badly.
And it's in the retention of excellent driveability at the same time as major performance improvements are being made that is one of the greatest challenges facing OE engineers. To reiterate: improving performance is simple. But to gain in performance while retaining the same level of driveability is one of the hardest challenges in automotive engineering.
How do you make the steering responsive and quick through urban roundabouts, while at the same time not making it nervous around centre at high speed on a narrow road? How do you provide an accelerator pedal that allows the easy control of a standard 300kW - through a total foot movement of perhaps just 50mm? What is the process in engineering a brake pedal that remains light and progressive and has the power to pull a car down from high speed at an acceleration approaching 1g - but can still provide sufficient leverage to stop the car should the brake booster fail?
To improve driveability at the same time as boosting driving performance, engineers are increasingly turning to non-linear control interfaces. (Another way of expressing the term 'non-linear' is to call it 'variable ratio'.)
Take steering as an example. In most cars of the past, the relationship between steering wheel driver input and the angle that the front wheels adopted was linear. Turn the steering wheel (say) 20 degrees and the wheels might turn 1 degree. Turn the steering wheel another 20 degrees and the wheels will turn a further single degree. In this case, the total reduction ratio of the steering is 20:1. It retains that ratio through the whole steering range - it is a linear system where a given input has a directly proportional output.
To a greater or lesser extent, all major car/driver interfaces have in the past been linear. The throttle, steering, brakes - you move the control further, and more of that outcome occurs in the proportion that you expect.
The resistance to the driver input - part of the feedback - also increases with the amount that you have moved the control. The throttle resists further deflection to a greater extent when it is near the floor, the self-centre'ing of the steering wheel becomes greater as you dial on more lock, and so on.
In fact we've become so used to this that if it fails to occur, we're unhappy with it. The first series of Mazda MX5/Miata, for example, has steering that won't self-centre once a major degree of lock has been input. I drove the car and didn't like it - the non-linear self-centre'ing discomfited me
But that's exactly what we're talking about in the new breed of control systems - non-linear controls with variable feedback. The difference to the lack of Mazda self-centre'ing is that if the control system is carefully developed as non-linear from its inception, the end result can very substantially improve driveability and performance.
But enough of the background - let's get into some real systems. This week, we'll expand on steering.
Variable ratio steering has been around for many years, with most of the technology developed by an Australian company, Bishop Steering Technology Pty Ltd. However, as a recent engineering paper by Andrew Heathershaw of that company makes clear, the variable ratio that is chosen has in the past not necessarily been picked with any eye to giving best control over the whole steering range.
"While steering systems that vary the angular ratio between the steering wheel and road wheels have been in use for many years, development of a variable ratio steering characteristic has normally focused [only] on selection of a steering ratio for on-centre control and the specification of turns lock-to-lock," he says. Furthermore, "The change in steering ratio from on-centre to full lock has often been arbitrary."
The variable ratio that is used in most non-linear steering systems is chosen to give slower steering around straightahead and quicker steering as more lock is applied. The 2000-model Mercedes E240, for example, uses a steering ratio of about 18:1 at straightahead, retains this ratio for as much as 90 degrees of steering lock either side of centre, then reduces down to 12:1 at full lock. Its steering ratio variation is shown in pink on this graph.
Bishop decided to investigate the driveability of steering ratios which varied quite dramatically in a non-linear manner - both in the relationship between ratio at straightahead and at full lock, and also in how quickly the ratio changed. For example, the Mercedes approach shown above changes ratio at its steepest gradient by 11 per cent for each 90 degrees of lock - some of the trialled non-linear steering racks changed their ratio by as much as 28 per cent for each 90 degrees of lock!
But with such a radical variable ratio, wouldn't people really lose their grasp on where the front wheels were pointing? Guinea pig Mercedes E240 cars were fitted with five different steering racks - including the standard variable ratio system whose characteristics are shown above. One of the other racks gave very similar performance to the standard rack (being just a little different at steering inputs of over 180 degrees), while the other three racks varied quite substantially from standard.
The most radical rack ratio is shown here in blue. In addition to altering its steering ratio very quickly as a 90-degree steering input was reached, it also reduced the numbers of turns lock-to-lock from the standard 3.25 down to 2.69.
Drivers of the vehicles were put through a number of tests, including a low-speed drive around a tight kart track, a high-speed double lane change obstacle avoidance test, a cornering test, and straight-line driving. The drivers evaluated the steering on the basis of whether it was better or worse than the standard steering, with weightings reflecting how strongly they held their belief.
The steering ratio which gave the best subjective performance overall was the one shown here by the blue line...a variable ratio very different to that usually employed in cars!
Speed as Well?
But it's worth thinking for a moment about different driving conditions, and the type of steering that would be preferential in each case. For example, wouldn't it be better to change the ratio not on the basis of the amount of lock being used, but on the input of road speed?
(Note that we're not talking about changing steering weight with speed - an approach almost universally adopted in current power steering systems - but actually changing the steering ratio as the car goes faster.)
Honda is one company that has experimented with just an approach - a steering system that alters in ratio with vehicle speed. And, to make it even more interesting, their system additionally varies in ratio with the amount of steering lock!
Let's first take a look at Honda's approach to varying the ratio according to vehicle speed. The company decided to use a steering approach that with 180 degrees of steering wheel input, gives 25 degrees of tyre turn at 20 km/h - and just 10 degrees of tyre turn at 100 km/h. That is, the ratio at 90 degrees of steering lock varies from 7.2:1 to 18:1, depending on road speed. So the slower you go, the quicker becomes the steering.
This graph shows the relationship between steering wheel angle and tyre angle at 20, 40, 60, 80 and 100 km/h.
However, Honda engineers were also concerned that they should compensate for the increasing understeer that occurs as steering lock increases, especially at higher speeds. So they added another non-linearity where as speed rises, more steering wheel angle is applied for a given input in steering lock. At 100 km/h and with 180-degrees(!) of steering input, under the above system about 10 degrees of tyre angle would be developed. But with the 'understeer correction' element added, this doubles to 20 degrees.
An electro-mechanical steering system was produced that allowed the steering ratio to vary on the basis of both steering lock and road speed inputs. The system was fitted to a 1.8-litre front-wheel drive Honda and then tested.
One of the goals of the system was to reduce the amount of steering input needed during low-speed parking manoeuvres. The test sequence involved negotiating a narrow right-angle bend and then reversing around another right-angle bend.
The steering wheel angular inputs needed to complete the manoeuvre with conventional steering are shown here...
... while with the variable ratio system, the steering inputs are reduced very substantially.
But what about in higher speed applications? A double lane change test was set up, and the steering inputs again logged. In addition to requiring less twirling of the steering wheel, noticeably less steering corrections were required with the variable ratio steering - the subjects steered the vehicle much more smoothly.
The engineers also tested the vehicle on a slippery surface. The test consisted of pulling on the handbrake at 40 km/h on a road with a low coefficient of friction. The time taken for the driver to recover from the tail slide was measured, using 15 different subjects "having various driving experience".
The diagram here shows the results of this testing. The graph relates the maximum steering rate (ie how quickly the steering correction had to be made) and the time taken for the correction to be successfully finished. The red dots show the data for the variable ratio steering system and the black dots the data for the comparison conventional system. When equipped with the variable ratio steering, drivers generally both recovered faster and also had to input steering less rapidly.
The BMW Z22 research car was built in 1995, however it did not become public until 2000. The car exemplified many drive-by-wire technologies, including its steering.
With no mechanical connection between the driver and the steering wheels, it's possible to have effectively any steering ratio at any road speed or degree of steering lock.
The steering system also allows the implementation of variable feedback. Quoted in Automotive Design and Production, Dr Jürgen Guldner, project manager, BMW Technik Drive-By-Wire Systems, said, "The steer-by-wire system on the Z22 was my area of responsibility. We measured the forces through the tie rods, which gave us a pretty accurate picture of what was happening at the contact patch." A non-linear amount of feedback could then be generated - for example, the slight lightening of the steering that occurs as the front tyres reach the onset of understeer could be amplified so that the driver could more readily feel the front-end slide starting.
Furthermore, using a speed-dependent variable ratio approach, "you can create a very agile, very responsive vehicle - even if the vehicle is a large one. So your limousine can feel like a sports car."
Steering inputs also do not need to be confined to coming from the driver - in the same way as Stability Control can yaw the car (ie pivot it around a central vertical axis) by braking individual wheels, the system can also have direct inputs into the steering. In addition, steering inputs that are inappropriate (eg 90 degrees of steering wheel movement very suddenly applied by the driver at 200 km/h) can be ignored or reduced in magnitude to an amount that allows the car to perform a safe turn.
Drive-by-wire steering technology gives the greatest flexibility in the relationship between steering wheel input and the angle that the wheels adopt.
Next week: non-linear throttles
While once it was common to have steering 'quick racks' available, they appear to have all but disappeared for current models. However, the availability of variable ratio racks, where the rack ratio spread and gradient can easily be tailored to suit a sporting driver, makes this an obvious avenue for the aftermarket to explore. The sort of rapid change is steering ratio that is shown in the experimental Bishop steering rack would be ideal - and it could be made faster at all steering angular inputs. Two turns lock-to-lock without nervousness at the straight-ahead position would be a reality.
Once electronically variable steering systems are introduced, it become ridiculously easy to make modifications. As this diagram of the Honda prototype system shows, there is only one sensor input into the variable steering system - road speed. Therefore, altering the speed input would alter the ratio of the steering that is available at that speed. Reduce the speed that the ECU sees and the steering will get sharper. A variable quick-rack will be as simple as adjusting an interceptor module....
Speed sensitive variable weight power steering systems can have their assistance very easily altered, as our article "Modifying Speed-Sensitive Power Steering" last week showed. The potential driver benefits in altering steering weight really needs to be felt to be believed - an aftermarket interceptor dedicated to this function would sell very well for some cars that as standard have overly light power steering.
So there's certainly plenty of opportunity for OE steering technology modification equipment to enter the aftermarket.