2002 power steering shudder fix

I have not been able to test this on the steering but I did test it with a noisy accelerator.

Take a 9V battery and jumpers. Disconnect the PS ECU (located behind the glove box.) Identify the torque sensor pins. Put the 9V battery on the torque sensor pins and turn the steering wheel manually. Reconnect and test handling.

We found the accelerator encoder appears to suffer from 'tin whiskers'. I suspect the torque sensors suffer from the same problem but the challenge is how to get rid of them. A 9V battery provides more than enough voltage and current to vaporize the whiskers. Turning the wheel sweeps the torque sensors as far as they can move and hopefully reaches the 'buggers.'

These 'tin whiskers' grow on their own and there is no know cure if a batch of sensors has the problem. Burning the existing ones out does not prevent new ones from forming . . . this is KLUDGE . . . but it should provide relief until you can get a permanent fix.

The resistor-network works by reducing the signal coming from the torque sensors. In effect, turning down the gain so the 'noise' has less effect. A more advanced form would use a resistor-cap network to filter out the high dV/dt, noise. But this is also a proposed fix. If it happens to our 2003, I'll have a test article and can tweak it. Use Wiki to search for 'electronic filter' to see more technical details.

Bob Wilson

 

What did you try? The steering is pages 164-165 in my (2001) wiring diagram - probably very similar in yours except maybe page number changes. The only place where the connector's easy to reach is at the EMPS ECU, where the connections are pins 1, 4, 2, and 3 of connector E5, respectively white, red, green, and black, carrying: positive reference voltage, torque sense 1, torque sense 2, and ground reference. Section K in the diagram will show what connector E5 looks like and how the pins are numbered.

If you want to see how those are related to the inner structure of the torque sensor, I think some good explanations/diagrams are in the New Car Features Manual ... or you can go for the retro ASCII art in my 'steering gear jitters' thread.

Bob is probably the man to go to for suggestions as to which combinations of pins to hit with the battery/wall wart.

-Chap

 

All combinations with a swing back and forth on the wheel while parked. Tin whiskers are small, metal 'fuses' and you want the battery/wall wart to vaporize them. But if you want to do it right, to measure it first, we'll need an amplified speaker and small, high resistance voltage.

We would rig up the amplified speaker with a small voltage, enough to hear any changes. Then connect to each pin-pair and sweep the steering wheel and listen for the noise. Doug Schaffer pioneered this approach when he developed the accelerator cleaning procedure. I used something different.

I used a sensitive VOM and measured the voltage at difference offsets. In the case of the torque sensors, I would rig up a torque source and measure the resistance at different applied steering torques. But then I don't have the part.

For a repair, just use the 9V/wall wart and sweep the steering wheel. Hook it up and take it for a test drive.

Bob Wilson

 

Yes. The problem is worse than tin whiskers. Sad to say, the encoders are not in a place where mechanical cleaning is an option.

There are no cheap and easy options at this point that I am aware of:

1. Route steering ECU power to a switch - keep off except when power steering is needed for parking and live with PS light.
2. Replace rack, used
3. Replace rack, new
4. Reseat connectors (very, very seldom works but cheap)

Bob Wilson

 

Well, crud. I was lucky enough to have my rack replaced under the warranty extension, but I was still holding out hope this technique would be effective. Not just for other people too late for the warranty, but also assuming even my replacement rack will probably need the treatment eventually, as I don't think its design is any different.

Now it's back to looking for other solutions. Phooey.

If I remember right, just leaving the torque sensor unplugged worked for me, and (oddly but conveniently) didn't seem to generate the warning display.

Just to make sure nothing simple is overlooked ... when you unplugged the sensor connector from the ECU behind the glove box, and applied voltage to the connector pins, you were applying it to the pins in the disconnected cable end, right, and not in the ECU socket? Please don't take the question wrong; you just didn't exactly say that, and if I didn't ask, there could be something obvious we'd miss.

-Chap

 

So, one nice thing about getting a shiny new rack back in 2012 was being able to take torque sensor resistance measurements when it was one day old.

Now that it's 2½ years old, I might just go out and repeat the same measurements for an idea of how it's holding up.

-Chap

 

One thing, if you have a chance to inspect a used rack before buying, is you could measure its resistances with an ohmmeter. I've got two sets of "good" measurements now, from my replacement rack when it was a day old and now at 2½ years, and also the "bad" measurements from my old one. The bad one had some values that (a) were noticeably higher, and (b) never quite settled down on the meter. Readings on the good one have changed very little, and the meter is rock steady when reading them. So that might give you a way to tell a decent used rack from a flaky one.

-Chap

 

Sure would, wired up the right way, but I'm not sure you'd even need it. My 2001 ECU, anyway, would light up the screen with a warning if it got enough weird input from the sensor, but seemed to be perfectly happy to tool around town with the sensor unplugged outright.

-Chap

 

... but if you need to, it's probably easier to just build a simple voltage divider that's in the same ballpark and provides 2.5 V on each of VT1 and VT2, than to try to match exactly what's in the steering gear.

What's down there is, to the best of my knowledge, a sort of hidden resistor network that we can only learn about by measurements between the four accessible points. If we could see the network, it ought to look like the drawing in the New Car Features Manual - unless Toyota has done something more clever since then to address the problem, and at the moment I'm not sure because the measurements I've taken differ from what that picture had me expecting.

If it is as the picture shows, there would be four resistance values to find, and none of them is directly present in my measurements. The last set of measurements I took would be easiest to start from, since each gives one equation involving only two of the unknowns in the formula for parallel resistances. Solving for them is still an exercise (I didn't plan to say "left for the reader" but at the moment it is, 'cause I haven't done it yet).

-Chap

 

White and black supply a 5 volt-to-ground reference provided by the ECU. What you would want is to choose appropriate resistors to build a voltage divider that returns half of that onto the sense lines VT1 (red) and VT2 (green).

Ordinarily a divide-in-half voltage divider would be two equal resistors, but you might need to build that, try it, and measure what voltage it actually puts onto VT1 and VT2. The rack, as I measured, does not have equal resistance from sense to +5 as from sense to ⏚, and yet it does put 2.5 volts on VT1 and VT2 (when you're not trying to steer). That could mean that something down in the rack is more elaborate than just two parallel pots as the manual shows (in which case bets are off). Or it could just mean that the ECU inputs have quite low impedance and the voltage divider has to account for that in the choice of resistor values.

In any case, just because of imperfect tolerances on things, any circuit you build up will probably not supply quite exactly the same VT1 and VT2 voltages as your rack does, and you'll have to go through the programming steps in the manual to teach the ECU "ok, this is straight ahead" or the car may be slightly prone to driving in circles. That's the same programming that has to be done if a rack is replaced.

-Chap

 

A pair of 2k ohm trimmer resistors should do the trick:

The challenge will be to adjust them so the ECU remains bias neutral, not applying power to the motor.

Bob Wilson

 

Once they're anywhere close, just telling the ECU "what you're seeing right now is what zero torque is" can be done pretty easily with the Ts and Tc jumpers, and even more easily using Techstream.

-Chap