The motoring media was granted a rare insight into Toyota's product testing methods last week. Participating journalists from around the country were invited to take part in a round of testing for the Camry Hybrid against a competitor set of three medium segment (VFACTS) peers: the Honda Accord VTi Luxury, Mazda6 Diesel Sports hatch and the Subaru Liberty 2.5i.
A team led by technical consultant Graeme Gambold -- and including rally aces Rick and Neal Bates -- set up a series of tests at the motorcycle track located outside Broadford, north of Melbourne. The tests -- five in all -- were designed to assess vehicle dynamics, broadly covering roadholding, handling, steering, acceleration and braking. Data collected from in-car telemetry would be collated and used to paint a repeatable comparative picture of the Camry and its rivals.
"We do a lot of product evaluation, leading up to a [new] model," Toyota's senior director of sales and marketing, David Buttner explained during his welcoming address.
"[And] we do a lot of tuning in Australia... but we also do a bit of mid-model cycle testing. For us, this is really an integral part of our normal cycle... where we get out and observe vehicles -- and see how our vehicles perform against those and feed that data back into the system for future generations of cars."
Buttner described the Accord, Liberty and Mazda6 as the Camry's "level two and level three CBU competitors," meaning that the cars were Completely Built Up imports -- as against the locally manufactured Camry.
According to the Toyota executive, this sort of testing is usually conducted by the company's in-house engineering staff, with some help from Toyota Technical Centre Australia and all overseen by product planners. For this occasion (whether due to the presence of notoriously cynical journos, Buttner didn't say) the parts of the engineers, TTCAu and product planners would be played by Gambold and his staff, the journalists and a couple of Toyota staff -- Buttner himself and colleague, product planning chief, Peter Evans.
"Today's really the first time ever we've done this sort of back-to-back testing, inviting [the media] along -- not just to observe, but to participate," Buttner remarked.
Gambold delivered a preliminary briefing to the drivers, outlining the five tests and what they were designed to achieve. The first test was a simple slalom. Drivers were to approach the witches' hats at just below 80km/h indicated and then turn the car into the slalom, ensuring no brake pedal pressure applied and no use of throttle to correct a slide.
For the second test, drivers were to launch the car from a standing start using full throttle and turn into an uphill right-hand bend, maintaining an arc close to the inside of the corner. The wide-radius bend turns in on itself near the end of the test section, forcing the driver to take another bite at the corner -- just in case the vehicle's stability control didn't already have enough to do.
Immediately after, the driver would bring the vehicle to a full stop for the combination of third and fourth tests -- acceleration and braking.
The Racelogic onboard telemetry equipment would time the vehicle's acceleration to 100km/h from the standing start, but not beyond that speed. Similarly, it would measure the car's rate of deceleration, distance travelled and elapsed time from 100km/h to zero. So the idea was to launch the car and accelerate up to an indicated speed of 110km/h (allowing for the 10 per cent speedo error permitted by law under the auspices of the ADRs) and then brake for all your worth, bringing the car to a complete halt. The driver was to hold the car stationery until it had even stopped rocking on its springs, as the telemetry might construe that as continuing travel and screw up the data accordingly.
Accelerating through another uphill corner was the final test, but this one was a left-hander with a sting in the tail. At the exit from the corner were two impact-absorbing posts in high-vis orange.
While this was also a test of the sophistication and response of each vehicle's stability control, unlike the second test, this one was aimed at assessing how quickly the stability control system could change from modulating tyre slip while accelerating to ironing out yaw with the power off. For this test, the driver would accelerate full-tilt through the corner and then snap off the power at the top of the climb as the vehicle passed through the gates. Typically, the sudden weight transfer from rear to front in long-wheelbase front-drive sedans should result in trailing-throttle oversteer
We were each allowed three runs in each of the four cars. Two of the cars were self-shifters (Camry with CVT set-up and Accord with a conventional epicyclic automatic transmission), but the other two (Liberty, Mazda6) were manuals. This could/would have some bearing on how the cars fared against each other. Furthermore, one of the cars, the Liberty, was an all-wheel drive -- but this didn't proffer any sort of advantage over the front-drive vehicles.
So there was some variation in the different vehicles' specifications. Gambold was also concerned that with so many drivers taking part -- and doubtless drivers of varying levels of ability -- there were two points of variation that could influence the data output. For this reason, he stressed that the exercise wasn't to see who could get through the slalom the fastest, but rather who could approach the sort of repeatable results that the professional drivers were achieving.
He estimated that the variation between the non-professional journalists and his own test drivers might change the results by as much as 10 per cent. At the end of the day, he was pleased to report that the driver-related variation had been as low as "three to four per cent."
>> Seat of the pants -- Test one
Test one, the slalom. Gambold's team of four co-drivers (including the two Bates brothers) were sitting in the front passenger seat to monitor the VBox telemetry and advise the drivers. When the co-drivers said to keep the approach speed below 80km/h, there was good reason for that.
In the first attempt (in the Accord), the speed was slightly above the indicated 80km/h and the Honda turned in with all the grace of a pachyderm on blade skates. Missed the first cone -- and the second -- but the car couldn't make up its mind whether it wanted to oversteer or understeer by the third and blurred recollection suggests it took out the fourth cone as well.
For the second attempt, in the Camry, a different tactic was called for. Accelerate up to 80km/h indicated and then, about 20 or 30 metres prior to the first cone, back off the throttle altogether. Result? Not much better, if any... The co-driver explained that the car effectively continues accelerating slightly, even after the driver has lifted the right foot off the pedal. This seems to be due to something like engine flare and the nature of the CVT combined. Whatever the reason, the Camry was still travelling too fast as it entered the test section.
Right... The next car was the Subaru. Only accelerated up to about 75km/h (holding third gear) and turned in, taking a very wide line. The point about these cars with their stability control is that sometimes they need to be taken by the scruff of the neck before the active safety system will intervene. That means more steering lock and/or more lateral g-force.
This time, the Liberty (possibly with the aid of some engine braking through the manual box) cleared all the cones. Exultation was quickly overtaken by disappointment when the co-driver advised that the line taken was "too wide."
Lastly was the Mazda. Speed right, line right, some engine braking from the diesel powerplant -- and plenty of arm wrestling with the wheel. The car once again negotiated the slalom without smoking any cones.
The writer attempted this test three times in all four cars. By the third time, the Camry completed the course without hitting any cones. Both the Liberty and the Mazda6 managed to complete the course perfectly on all three occasions. The only car that I couldn't set up correctly (speed-wise) was the Accord.
>> Test two
Test two, uphill corner with secondary apex. All four cars managed to complete this test with reasonable precision, the stability control intervening as the driver kept the boot into it and turning in tighter as the corner wound in on itself.
The front-wheel drives generally coped better with this test than the all-wheel drive Subaru did, the Liberty exhibiting a touch of oversteer just past the second apex. Rather than blaming the Subaru's basic underpinnings (and its 'passive dynamics'), the co-driver suggested the Liberty's problem possibly lay with an older generation of stability control program strategy -- one that was slower to respond to a changing situation.
Oddly, the Subaru required a different driving style for this test. Where the diesel Mazda was taken up to third gear and left there with the foot flat to the floor, and the auto pair of Accord and Camry were just held flat in Drive, the co-driver recommended keeping the Liberty in second, with the engine tagging the rev-limiter as it crossed the finish line for the test -- tail yawing left slightly. What would have happened with an upshift to third?
After the Accord, the Camry seemed also to be an understeerer, but the stability control kept the Toyota's track fairly neat and consistent, nonetheless. The sense of understeer came through the wheel rather than from the seat, indicating that the car was tracking more or less correctly, regardless of how it felt.
>> Test three and four
Straight-line acceleration/braking... Here's where the Camry really made the other cars look amateurish. Boasting six-cylinder levels of performance, the Toyota whipped up immediate torque from its electric motor to supplement the four-cylinder petrol engine. The CVT helped also, constantly juggling the torque from the petrol and electric drive units under full-throttle acceleration. By the third run, the Camry was pretty much pulling up to a halt at a point on the track where one of the cars was still accelerating to its ceiling speed on an earlier lap.
An odd quirk in the Subaru was thrown up by the braking test. The co-driver approved throwing the clutch at the same time as jumping on the brakes. It would make no difference to the car's braking performance whether the clutch was engaged or not, he said. The Liberty drew to a halt with the engine still revving at around 3000rpm -- without any pressure on the accelerator.
"It's not you," the co-driver counselled. Perhaps the engine picks up revs during emergency braking to ensure a strong supply of hydraulic pressure for the braking system.
And for all those conspiracy theorists out there, currently focused on 'jamming' cruise control systems, Graeme Gambold was able to point out VBox telemetry that measured each car's rate of acceleration at no higher than 0.5g, whereas deceleration through braking exceeded 1.0g for all four cars.
In other words, stomp on the brake pedal and you'll slow a car with a sticking throttle -- even if the throttle is jammed wide open.
All the cars proved adept where straightline braking was concerned, but despite eliciting a chirp from one of the wheels on launch, it was the diesel-engined Mazda that seemed slowest reaching 100km/h. That conclusion was based on the distance travelled over this test. The Mazda's poor showing was attributed to turbo lag and diesel power delivery by the co-driver.
In the slalom, it was the Accord that had seemed slowest, but this test strongly pointed to the Mazda taking the wooden spoon.
Fallible human gear-changing probably didn't help the Mazda either, although it didn't seem to hurt the Mazda as much it did the Subaru. At least a couple of drivers wrong-slotted into fifth looking for third during the Subaru's acceleration run...
>> Test five
Final test, lift-off oversteer. This was the test in which the Mazda6 -- one of our favourite handling mid-size cars -- blotted its copybook. The test was intended to see how a car would react to suddenly lifting off the throttle at the end of a left-hand turn uphill, with plenty of inertia already pushing the car towards the outside of the corner. This test represents how the average driver might react to an emergency situation.
We were told to rev the cars out in first (for the manuals) and then short-shift at 5000rpm (in the Liberty) from second to third. The two self-shifters were left to their own devices and for the diesel Mazda, every shift is a short shift anyway. All of this as the car is veering left and climbing a short ascent.
Both the Camry and Accord coped respectably well with this scenario.
On the first two runs, the Liberty didn't put a foot wrong, but according to the co-driver, that was largely due to "rolling off" the throttle, rather than snapping it closed. On the third run the Liberty built up such a head of steam that it was understeering quite a bit before even reaching the point of closing the throttle. The car's offside mirror clipped the post on the exit gate as the Liberty sailed through. This was one of those occasions when the nut behind the wheel should have hauled on more lock.
But all of that was as nothing compared with the Mazda. Whether due to the extra weight in the nose and the truckload of torque from the diesel engine, throttle held fully open once the car was in third ensured the Mazda6 was embarking on a scenic tour of the grass verge on the outside of the corner long before the throttle was to be snapped shut.
In point of fact, it never quite reached that expeditionary point, but the car did take out the bendy pole on the right side of the exit gate during the first run. On both subsequent runs there was no other choice but to back off well before reaching the exit gate.
>> Objective data
In a debrief at the end of the day, after we had assimilated the 'low-mu surface'-type jargon*, the media was presented with a series of bar charts to show how the four cars had compared when assessed for different dynamic characteristics.
Looking at just the results for the third test session (presumably the most accurate, since the journos had their act together by then), the Accord posted the fastest time in the slalom, but generated the second lowest lateral G-forces. That result, according to Gambold suggests that the Honda was taking a shallower route and getting from points A to B faster, knocking over cones along the way. Perhaps the journos didn't completely have their act together after all.
The Mazda was next fastest, but only marginally so against the Subaru -- and the Liberty posted far and away the highest lateral G-force numbers. Gambold said that other data shows the Camry and Accord are more capable in the hands of inexperienced drivers, but to us, the bald data suggests the lighter Liberty is the one you would want to be driving in a motorkhana.
The exit speed from the slalom tells a similar story. While the Accord, at 49.6km/h was far and away the fastest, there's a strong suspicion that at least one or two drivers had given up trying to circumnavigate cones and just ploughed over the top of them.
At 44.1km/h, the Liberty's exit speed was marginally slower than the Mazda's at 44.3, but given the Subaru's faster time through the slalom, it appears to have washed off more speed on the trailing throttle. The Camry posted the slowest time and the lowest exit speed (40.5km/h), but was second only to the Subaru for lateral G-forces generated.
Not sure what to make of that other than perhaps the Camry's regenerative braking was providing more retardation than the conventional engine braking in the other cars?
Supporting the subjective view that the Camry was quickest of the four in a straight line, the basically unquestionable acceleration data showed the Toyota reaching 100km/h from a standing start in 8.47 seconds -- nearly 1.5 seconds ahead of the next fastest, the Accord at 9.91 seconds.
Hang on, the Accord was second fastest? It didn't feel like that. The difference between the Honda and the Liberty (10.17 seconds) or the Mazda6 (10.38 seconds) was likely due to the Honda's swifter-changing automatic transmission, versus the manual boxes in the Subaru and the Mazda. That was clearly shown in the acceleration time from 60-80km/h: Camry still in the lead with a time of 1.83 seconds, but followed in order by the Liberty (1.98 seconds) and Mazda6 (2.20 seconds) -- Accord following the rest of the field with 2.29 seconds. In that speed range, there was only one gear change at most for the manual cars.
Also confirming that manual gear changes detracted from acceleration figures, the peak G-forces were lowest for the fastest-accelerating Camry at 0.46g, significantly better (0.52g) for the Accord, but better still for Mazda6 (0.53g) and Liberty (0.55g).
What this seems to say is that with the Camry's two drive units working in unison, but torque peaking independently, acceleration for the Toyota was closer to a linear rate. The other cars -- particularly the manual jobs -- did actually accelerate faster at times in gear.
All four cars posted similar distances and times for braking from 100km/h, but the Honda took longest to stop and required the greatest distance. Still, all four pulled up in less than three seconds and while the Accord's stopping distance was 40.54m, it was only about two metres longer than the best performing car, the Liberty on 38.49m.
The Mazda was a smidge slower stopping than the Liberty and the Camry placed third.
Taking into account its median-like braking and its outstanding acceleration, it was no surprise the Toyota aced the combined result for 0-100-0km/h. Honda was second, narrowly pipping the Subaru and the Mazda further behind by a significant margin.
Data from the second and fifth tests were combined to present a measure of each car's traction control system capability and stability control performance. Corner acceleration forces (lateral G-force) actually placed the Camry (0.567g) just ahead of the Liberty (0.563g), with the Mazda6 further back (0.552g) and the Accord following (0.530g).
While this data, on the face of it, appears to speak to each car's roadholding, the structure of the test favoured a car that could accelerate faster than others -- the Camry. It's not specifically a measure of roadholding, in isolation. The faster a car can accelerate, the more G-force it will meet on a turn of equal radius.
There's a counterpoint to illustrate this -- cornering deceleration. In this instance, the diesel-engined Mazda, with its high compression ratio and manual transmission was well ahead of the other three cars.
For cornering exit speed, the Mazda6 (94.86km/h) was marginally ahead of the Camry (94.21km/h) and there was barely daylight between the Accord (91.97km/h) and the Liberty (91.36km/h). These results are purely data-based and tell nothing about the car's attitude on the exit -- like the Mazda's excessive power-on understeer in test five, for example.
>> What was learned from the day
Heaps... But much of it counter-intuitive! For a start, 'inexperience' beats 'experience' in cars equipped with stability control -- soon to be mandatory for all new cars sold. Drivers facing an emergency should turn harder and brake with as much force as possible applied to the pedal.
Why? Because the car will undoubtedly recognise the driver is grappling with a serious situation and respond promptly.
Those of us who learned to drive in pre-antilock brake times will endeavour to moderate brake pedal pressure, possibly wind off lock and back off the throttle gently. Drive in a way that's more like motor racing than avoiding a crash and the stability control may not do what you hoped.
By now we're mostly aware that there are stability control systems and there are stability control systems. Some cars, particularly those promoted as sporty models, allow the driver considerable latitude before the electronic safety aid will override the driver.
Understeer is a subjectively safer handling trait for the majority of drivers and the car companies face a major conflict between the needs of developing a stability control Mu estimation strategy that will suit most drivers and "not be the fun police," as Graeme Gambold describes it.
It was the view of Gambold and his team that cars like the Camry and the Accord are better in this regard, whereas the Mazda and the Subaru err more on the side of sports calibration for their respective systems. There's room somewhere in between for all to meet.
There are very few areas of vehicle development that are quite so dependent on the car company knowing its buyer demographic as stability control programming. Stability control calibration can and should be a case of accurate market research meets engineering precision.
If you're a company selling thousands of cars to fleets every year, it doesn't seem logical to supply those cars with the sort of stability control system that allows the tail to step out about 15 degrees on a trailing throttle before it applies the stoppers. Nor do you want an over-reactive system in a sports car.
Gambold praises the electric power steering in the Camry. Not only is it a fuel-saving device, but it's not subject to the oil surge in a hydraulic power steering system, creating what Gambold calls 'wall effect' in a test like the slalom. With centrifugal force acting adversely on the hydraulic fluid in the system, the steering becomes heavier in one direction than the other.
Toyota claims that the Camry's stability control program pulses at 100Hz a second, which compares very favourably with some systems on the market, operating closer to 50Hz. To demonstrate this, the Camry and the other three cars were driven around a circular array of witches' hats set on a wet section of track (the infamous low-mu road surface) with the drivers invited to keep the accelerator pedal nailed to the floor.
The Camry was the only car that would allow this and maintain a tight turning circle. As the Toyota circled the cones, the body pitched fore and aft every couple of seconds as the stability control reduced torque and independently applied braking pressure to hold the turning circle.
By comparison, the other cars couldn't hold a constant line without the driver backing off the throttle long enough for the nose to tuck back in. The Subaru had the hardest time of it, spiralling further out on each pass until even the most determined driver was forced to ease up as the car headed towards the grass by the side of the track.
Granted, this is not the sort of test that an inexperienced driver would attempt, and it's also doubtful that an experienced petrolhead could be bothered trying it either, but it does illustrate that in a completely unrealistic test the Camry Hybrid has a party trick the others can't emulate.
If 'party trick' sounds harsh, consider this: a dumbed-down test in a contrived situation to demonstrate a car's singular prowess doesn't mean that the car will never be called upon to deal with a similar situation arising from a real-life emergency. Or, for that matter, the Camry's ability in this scenario may also have wider application elsewhere in the real world.
And as a final observation, while the track day was illuminating overall, the results wouldn't stop us buying a Mazda6 or a Subaru Liberty if we really liked those cars.
What is mu
* 'mu' is the Roman representation for the 12th letter of the Greek alphabet ( 'µ' ) and is used to denote the coefficient of friction.
1.0 mu is the coefficient of friction for bitumen,
0.7 mu is the coefficient of friction for wet bitumen,
0.5 mu -- gravel,
0.3 mu -- snow,
0.1 mu -- ice.
Read the latest Carsales Network news and reviews on your mobile, iPhone or PDA at www.carsales.mobi