steering gear jitters

I've been a little jittery recently wondering whether I'm on the way to the Dreaded Steering Gear Torque Sensor Failure. My car is not in the VIN range affected by the Y05 steering gear campaign, but two or three times in recent weeks, usually while steering at low speed into a parking spot, I've felt a fleeting, strange little wubba-wubba-wubba through the steering wheel. It actually reminds me of what a Macintosh does when your password is wrong, for those who've seen that. No lights, no codes, and the rest of the time I can't seem to make it happen.

Then today as I was rolling to a stop the ICE shut off, and I realized that sometimes the ICE shutdown is a little lumpy, and if it should happen exactly when I happen to be steering into a parking spot, it might feel like what I'm feeling. So I guess I'll watch carefully for a while longer and see if anything is getting worse, or it's just a false alarm.

But at least I did find out today that Cardone offers the steering gear remanufactured. I didn't recall seeing that option mentioned here before and I wonder if anyone has experience with it. If it is really as difficult as I hear for an end user to take the thing apart and clean the torque sensor contacts, it seems like the next best thing to have a well-equipped remanufacturer do it.

(Cardone doesn't sell directly; it'd be something you'd look to order through the neighborhood parts place.)

Now it seems like the really forward-looking approach (looking toward 625k lifetimes like Bob Wilson speaks of, for example) would be for some remanufacturer to find a way to retrofit a different type of strain gauge into the unit, maybe with a little electronic adapter to present the output in the two-complementary-voltage form the ECU expects. Hmmm....

-Chap

 

It might, but that's not the situation every time, so I'm probably not going to be that lucky. hwell:

At least there are a few more fun facts to discover:

• For faster tidier access behind the glove box, leave the inner side stops alone, pop out the hinge pins at the bottom, and let it drop and hang from the side stops (instead of removing the side stops, flipping the glove box down, and dumping everything on the floor).
• The EMPS WL signal is easier to probe at the combination meter connector than at the ECU behind the glove box. It's normally low and active high, which makes the diagram correct where the non-THHS DTC-reading procedure is first shown, but might not be what you would think based on the diagram for DTC C1556.:huh:
• Not only does the EMPS continue to provide assist power for a few seconds after IG-OFF (nicer to turn the wheel to a locking position that way), but after that it continues to provide 5V reference to the torque sensor for about 3 minutes - that provides a nice opportunity to move the wheel and measure torque sensor response without complication from the assist motor. This isn't just a happy accident, caps discharging, or the like - the ECU actually holds the EMPS relay energized that whole time.

I did have DTCs C1511 and C1513, but I read them after doing a lot of unusual wheel twisting and backprobing. So I cleared them and will see how long they take to come back without me doing weird things.

I did not, so far, manage to capture any smoking-gun scope trace of a glitchy torque signal.

Anybody have any experience with the reman steering gears?

-Chap

 

Thanks everybody for the info. I would guess about turning the wheel to clean the potentiometers that it wouldn't matter much whether the wheel is turned lock to lock - the pots are mounted to the shaft as a torque sensor so the signal shouldn't depend on how far the wheel is turned, only on how hard. Just leaning back and forth on the wheel with the ignition off (so no assist), and without actually pivoting the tires much on the pavement, ought to do the trick. If you twist the wheel hard enough to swing the torque sensor signal between say 0.5 and 4.5 volts a bunch of times, that should be exercising the pots over their full range of travel. I actually did that a large number of times while trying (unsuccessfully) to capture a glitchy signal on a scope, and it does seem to have reduced the severity of the symptoms, probably temporarily.

I have this lovely stuff called Jiffy Bath that works wonders on pots with this problem, only it seems like there won't be any practical way to try to apply it.

-Chap

 

... but in ACC of course, not with the steering locked ...

-Chap

 

I don't have photos because the thing is internal to the steering gear. It sounds as if tochatihu may be the only one on PC to have dismantled one deeply enough to get there (and from the available diagrams I infer that some bearing pull/press work was involved in getting that far, not to mention that tochatihu didn't find an obvious way to expose the pots without cutting wires).

What we do know from the available diagrams and docs is that the shaft is split within the housing into an upper and lower section joined by a torsion element, and one section carries a resistive element while the other section carries two wipers, diametrically opposed. The ECU supplies ground and +5 across the resistive element, which acts as two voltage dividers, "read" as voltages VT1 and VT2 by the two wipers. Both are near 2.5v with no force on the wheel; driver torque to the right drives VT1 toward 0 and VT2 toward +5, and vice versa. Of course this whole assembly also rotates together as the shaft goes from lock to lock, and the way +5, ground, VT1 and VT2 get from it to the stationary enclosure (tochatihu finds) is through a clockspring-wound ribbon cable.

It's just pretty uninviting to play with this because you can't even get there without investing all of the labor of steering gear R+R, and doing that much work to try an unproven fix and not know how long before having to do it again, doesn't sound like fun.

I guess the curious could buy a salvage gear to experiment with. It would be nice to have a good test jig to apply known torque, measure VT1 and VT2 behavior, and get an idea how glitchy the signal really is in a symptomatic one compared to a new or reman'd one. One would want to think that Cardone has done a lot of this.

-Chap

 

hmmmm ... tochatihu, when you disassembled one of these units, did the compartment with the torque sensor seem to be pretty well sealed, or was there a plausible way for it to be affected by weather conditions? Mine, too, is misbehaving more at some times than others (I haven't clearly pinned it to weather yet, though).
I wonder if exposed wiring could be implicated somehow....

-Chap

 

I wonder how possible would be some sort of Hall-effect retrofit, not unlike what Toyota decided to go with for the next generation? The ECU sends you +5 and ground, and you need to send back VT1 and VT2=(5-VT1), with VT1=VT2=2.5 at no torque.

I've got no estimate of the difficulty myself - I can picture the page on Hall effect in my college E&M book, and I know it's how my DC current clamp works, but there endeth my knowledge.

-Chap

 

Checked for DTCs again and nothing new since I cleared the C1511 and C1513 on August 14. This despite at least one instance of wheel shaking like a wet dog during a low-speed right turn. I wonder what the ECU's threshold is for weird input before it actually logs a DTC??

I don't know when it logged the ones I saw before. Definitely since I bought the car because I had it THHT'ed at a dealer first, but other than that I don't know if they were logged in normal driving, or caused by my recent attempt to measure the sensor voltages, or left over from a previous time when I unplugged the torque sensor to see if my generic scantool could read steering DTCs (it can't). I didn't do the full clear-DTCs dance then, I just opened the battery circuit and the red triangle was gone after that, but maybe it remembered the codes anyway.

Because it's a pain to read steering DTCs without a THHT and I didn't want to keep pulling the glove box and backprobing the ECU as shown in the manual, I added an LED to the combination meter. The ECU WL signal goes all the way to C10 pin 11 at the meter, there's just no light there to display it. So I just took the LED-resistor combo shown in the book and stuck it permanently to the meter, tack-soldered to the backs of pin 11 and ground (pin 14). Reassembled, it's just visible through the bottom-right notch in the combo-meter lens, after popping off the dash bezel, which is easy enough to not be annoying.

You can see there's even an unused idiot-light socket (three, even, counting two you can't see in the pic) so it beats me why they didn't USE one to show the steering WL. I thought of putting the LED there and printing a new version of the symbol card on transparency, with PS below the water-temp symbols, but I'd have to do much more futzing to get the colors to look good off my printer, so I just went with the LED stuck on the bottom with clear plasti-dip.

Rats ... caught myself inferring more than I know. Of course there's no reason they'd have to be diametrically opposed, and there's a diagram here that looks more 120-degree-ish.:redface:

-Chap

 

If you don't have a THHT, you can jumper TC to CG at the diagnostic connector, and the LED will blink out the last two digits of any steering DTCs (so 13 means C1513).

The book tells you to drop the glove box and back-probe your LED into the connectors at the steering ECU, but that gets old by the second or third time (and always in the back of my head is that one day my hand'll slip and I'll short something).

Most of the other ECUs can already cough up their DTCs by blinking something (the check-engine light, ABS light, door light, airbag light, A/C light). It's kind of fun to zap TC to CG and watch how many things start blinking (and the MFD pop into the familiar diagnostic mode, but in Japanese). But for some reason they wired the steering ECU's blinky circuit all the way back to the combination meter and then didn't put a light there.

-Chap