So I recently replaced the steering wheel angle sensor in my 2008 Trailblazer w/~95k miles. It was intermittent and causing a Service Stabilitrack error code when cold. They appear to be a common failure point for the Stabilitrack system at what I’d define as a premature failure. There are plenty of Youtube videos describing how to do the repair so not going there but in looking at why it failed, it becomes somewhat apparent that the design is lacking.
The absolute position sensor is a simple variable resistor (rheostat) and that is what is failing. If you look closely, you can see why. The carbon tracks are being erroded and casting off carbon particles and some metal flaking from the wipers.
Take a closer look at the wiper. They are like rakes, pressing the bent edges into the carbon tracks. If you feel the edge of the wiper, it is very sharp like would happen if they were sheered and the edge not dressed. Any wonder they are causing premature wear of the carbon tracks?
At first it might appear that a continuous rubbing contact will cause the element to rapidly wear out, but with careful attention paid to lubrication, contact design, and correct composition and curing, resistive sensors will remain fully functional after more than 1 3 109 dither cycles and many miles of contact travel.
Apparently, careful attention to contact design wasn’t involved here…
While I think that 95K is a bit short for the life of the sensor, I don’t think it is an outrageous one. You might have a minus 2 sigma life sensor for a number of reasons. The edge condition of the wipers, for example. Or maybe you correct the steering at a higher cycle rate than the average driver. Dunno.
Similar sensor design, from the analog part, is used for the wheel position sensors used on the Magnetic ride control systems and their design life is 11 million cycles.
I’ll bet I can design a sensor for that, that lasts the life of the vehicle and doesn’t cost any more money
Look at the dither expectations in the article and think about normal usage over 95k miles (BTW- many people getting far less usage than I did). No way I approached that number!
It’s true the majority of wear you see is concentrated about the center point but it may not show in the pics- the wear on one track is far more than the other across the swept range. Not sure if the failure mode is shorting from cast off or lack of contact pressure on one wiper when very cold ambient conditions. At any rate, that amount of cast off material means the contacts are very abrasive. Normal variable resistors use an arc bent wiper that rubs on the track, not digs in. I also think they messed up the formulation of the carbon tracking, it is far too soft a material to match the wiper design. The article speaks to that aspect as well.
These sensors were the topic of many hours of debate by the Stabilitrak engineers as well as the suspension folks. Getting good fidelity from the sensor was really important to both systems’ operation. The digital portion gives good fidelity for control loops needing steering wheel velocity and rotational acceleration. while the analog helps to determine wheel position. There was endless debate how to determine on-center at startup since you never know if the wheel is at zero or 360 degree left or right. The mechanical durability may have taken a backseat to the output fidelity.
Funny, I had exactly the same thoughts when looking at it. How would I determine absolute position from worst case, power cycled start up.
Another possibility- the sensor is not hermetically sealed. So they may have been quite concerned with oxidation or other contaminants affecting the continuous connection between the wiper and resistor. So they decided to use a highly abrasive wiper and sacrifice lifetime.
You should see the bracket holding it in place- 1/8" thick steel. Crazy.
Very interesting to hear the inside scoop- makes sense from my experience on how the discussions evolve…
I can understand rigidity on the bracket but 1/8" might be a bit of overkill! Or just laziness on the part of the engineer.
I used one of the analog-only sensors for data acquisition on my race car. My bracket wasn’t 1/8" thick. The accuracy was very good to measure angle. Differentiating to get rotational speed also worked pretty well for my purposes.
Stabilitrak, or any ESC system is balancing a fine line. Anytime the brake is applied it wears - makes sense. Applying it early in a stability event is desired for smooth control but it greatly accelerates brake wear, so you want accuracy from the sensor but a rather large “deadband” where the system doesn’t intervene. You really don’t want the brake to apply unless you really need to.
That’s where electronic shocks come in. If you ignore the ride part when a handling even occurs, you can stiffen or soften an individual shock to help reduce the oversteer event (soften rear shocks, stiffen fronts) or reduce the understeer event (stiffen rear shocks, soften fronts) without the brakes even being needed. The steer sensor signal leads the car’s response so you get lots of time (100msec - yeah, lots!) to help the car without needing an individual brake to apply to rotate the car back straight.
That sensor is pretty darn important to the whole system. Probably more than the accelerometers!
I guess they need the resistance method to back up the optical encoder for fail safe safety purposes. I concur, that doesn’t seem like it would provide a lifetime of the car part life expectancy. The gadget that seems to fail on my inexpensive radios and audio players is usually the same technology, the variable resistor for the speaker volume. My higher priced audio units use an optical encoder method for the volume control, and none of those volume controls have ever failed. Interesting post, thanks.
Actually, it provides absolute position information. The optical encoder can only tell direction of turn and rate of change. If you start off with wheel cocked to one side, it can’t tell. The optical encoder has much finer detail/resolution than the variable resistor. Data fidelity as mustangman appropriately defined it. The control system needs both inputs to function properly…
Excellent failure analysis and great information. A tip of the hat to you.
There’s a concept in reliability engineering called Failure Mode Effect Analysis (FMEA) which analyzes the effects of a failure, how it propagates through other systems when it fails and what other failures it becomes the root cause of. One (me) would think that with an FMEA analysis by the vehicle manufacturer, which must have disclosed the possibility of steering failures (I can only assume here) if the steering angle sensor failed, a more robust device would have been required from the component designer/manufacturer.
But, than, I need only think back to the original Hubble lens to realize that my thinking doesn’t always reflect reality. .
It never should have happened.
Perkin-Elmer’s own measurements showed flaws in the mirror before it shipped, and subsequent (to the telescope failure) failure analysis showed flaws in the Perkin-Elmer measuring system that should have been detected. The whole mirror-polishing process was one huge screwup.
I had the honor of working on a piece of analytical equipment that was to go on the shuttle. Nobody is perfect, but there are plenty of analysis, redundancies, tests, retests, backup tests, and oversight built into the work to prevent such a problem.
In my mind it’s the Perkin-Elmer name that should have gotten the “black eye” for the Hubble screwup. Unfortunately, NASA ended up with the 'black eye". The buck stopped there.
From a hazard and risk assessment perspective, the initial risk level values might be quite high but because the detectability is so predictable and the feature is not a primary system, the end result is acceptable without any further mitigation. A warning message is presented to the operator indicating service is required. This addresses the safety aspect but not VOC (house of cards…I mean quality… )
Considering the large number of electric power assist steering systems on the road, maybe they used their FMEA and designed a better one for E-steer?
I’ve had 2 cars with E-steer, a Honda I sold with 108K and no steering issues plus my current 58K mile Mustang. Both E-steers have worked perfectly but I can imagine the problems if the sensor starts to fail. There must be a torque sensor in the mix as well so it reads my effort as well as the position and rotational velocity.
I hope a rigorous FMEA was done on the E-steer system so it fails in a soft, friendly manner - like no power assist. The last thing I’d want is the car taking a hard right when I am on the highway.
While FMEA is solely focused on failure effects, I do think that E-steering systems will(have) evolve(d). Like any technology in its first years, I’m sure there’ll be (have been) some bugs to work out. Nothing technical began in its current form.
.
)
