Aftermarket OBD tools

It's a pity that kind of machine learning is too expensive to run the tpms system.

I had a play with SZviewer this morning.

It doesn't read the steering angle on the PS screen, only the torque sensor output and the motor currents.

It does display the steering angle offset on the ABS/ESP screen,  (mine is -2.5 degrees) but you can't zero it.


And I had 4 DTCs set, one for each tyre low pressure from December (rather sneaky remembering a low pressure, no doubt so someone can deny an insurance or warranty claim), so I cleared them

 

mlines wrote:

lightning wrote:

Thank you for the advice, l will take a look. However the steering self-centre seems to have adjusted itself and is no longer pulling to the right slightly!

l wonder if that is possible? Maybe the computer had compared the position of the steering wheel with the difference in rear wheel rotation speed
And thought to itself, hey, the steering is self centred slightly off, let's adjust it slightly.

With modern technology it's possible.
 

According to the manual, the steering "self centres" on the basis of simple averages. On average you are going to be steering in a straight line, even if you are constantly Steering left and right, the overall average of your steering will be the centrepoint (unless you are driving the Indi-500). Therefore the sensor can determine where the centre point is after a period of driving.
 

Slightly addition to this: it learns by looking at values of the steering angle sensor when the rear wheels are turning at a similar rate, rather than just an overall average for most of the time. If it sees reasonable difference in the front wheel speeds side to side it also doesn't use those times even if the rear wheels are turning at the same rate to stop it learning in certain circumstances.

That's why the manual has the warning about not rotating the rear wheels with the front wheels not turning at all and the steering turned off centre, as it can learn a new wrong centre.

That's also my working hypothesis about the wheel alignment shops screwing it up. Picture this as a scenario.
1. Underpaid worker is brought to rapidly put the Jimny on the rack to do an alignment
2. Worker drives onto the rack, since it won't take long and it'll be a road test they leave the car running (or they switch it off, but then have the ignition in ON position so the steering lock doesn't acrtivate)
3. Worker bee goes "oh cool, rear wheels have no adjustment, there's my datum".
4. Worker grabs the wheel triangle mounts to set the datum for their fancy 100k spared-no-expense laser alignment and levelling system
5. Worker is a good worker and understands that wheels can be bent, so they have the rear wheels slightly off the ground (or on rollers - have used them before to make it faster!) and spins the wheel they're using to set the datum.
6. Open diff spins the opposite wheel in an opposite direction, like open diffs want to do, Open diffs also are really good at sprinning the opposite wheel not just backwards but backwards *at the same speed* as the wheel you're rotating
7. The car's wheel speed sensors do not differentiate forwards and backwards movements. Just counts of tick, tick, tick... so if you spin a wheel backwards at the same speed that another one is spinning forwards, the car thinks both wheels are
8. Since the steering wheel is off centre because they haven't yet adjusted the drag link, the car learns this new position off to one side as the central position as the centre it is seeking.
9. Alignment is completed, steering wheel dutifully centred and toe set (even though it was probably ok beforehand, especially after a lift is installed). Now the steering wheel thinks it's turned a bit to the opposite side.
10. Customer picks up car, complains that the car power steering now forces the steering over to the angle it was before someone played with the alignment.

Almost all steering angle systems do self learning like this because sensors do drift over time and this saves customers needing them to be recalibrated every x km or whatever.

Chops, that is good info, and I see your logic about how a pro alignment can go wrong.

But are we sure that the car is using the PS to force the wheel back to (what it thinks is) centre? I have never seen that kind of behaviour in any other car I have worked on.

Facade, excellent info on SZ Viewer capability, thanks!

Shame it doesn't allow one to zero the steering angle sensor, but at least with a readout it will help with diagnosis. And one could always use the readout to manually centre the sensor -- bit of a faff but better than nothing.

Motacilla wrote:

Chops, that is good info, and I see your logic about how a pro alignment can go wrong.

But are we sure that the car is using the PS to force the wheel back to (what it thinks is) centre? I have never seen that kind of behaviour in any other car I have worked on.

 

Yeah, I'm confident for a couple of reasons. First off the Jimny does do some 'slow speed' self centering:

P/S control module determines basic assist power based on steering assist control and improves feeling of the steering by using steering return control and steering converging control. Further, the P/S control module controls motor current to harmonize target current with actual current in accordance with computing result of each control and maximum current limit control.

In any case, the other reason why I'm confident: if you yank the power steering fuse on a car which is exhibiting this behaviour then it totally stops doing it. Plus the angle the steering wheel gets to when it is trying to correct itself is the angle it was kinda 'at' before someone screwed with things.

It's still always hard with what are black box modules to know exactly what the decision making is; if I had more time on my hands then sure I'd love to make up a test rig to emulate the behaviour. The other thing that'd be good is actually have a good log of what has happened to the car including every step the people playing with the suspension took. People get themselves lost with these things.

Pretty much I've only seen it happen with pro alignment shops, too. People doing lifts at home and adjusting the drag link without actually doing stuff on an alignment rack (save for one example which appears to have been just getting stuff a bit unhappy and it eventually came good - presumably learning saved it there) always seem fine. It's always a pro alignment place and it isn't consistent, so clearly there's a way to do it and not screw it up with an actual alignment, too.