Gen 1 steering shudders

The torque sensor in the steering gear is just one of the many sensors in Gen 1 that were built out of potentiometers and eventually become electrically noisy. It's the one that's the most arduous work to get to, though.

It generates the signal for how hard you are turning the wheel, which the computer uses to decide how much motor power to send to assist you. When the sensor starts producing noise, the computer sends power to the motor in the direction it thinks you're turning the wheel. Because you're not really turning the wheel, that results in a torque signal in the other direction, so the computer switches to assisting in that direction, but you're not turning the wheel that way either, so that leads to a sensed torque in the original direction again. So it's a feedback loop, and in the right circumstances it can set up a self-sustaining shake.

While mine had the problem, I found I could minimize it by just keeping a relaxed but secure two-hand grip on the wheel. It's not about gripping hard or trying to 'fight' the shakes, they're too fast and nobody has the reaction time to do that. It's just more basic physics than that: with a nice secure two-hand grip, your arms become part of the mass getting shaken, and that changes the system's resonance away from the region where it oscillates the most.

Toyota extended the warranty by several years and would replace the steering rack for it; that's how mine got replaced at around 194,000 miles and completely solved the problem. But the limit on that program is past now, so it would be out of pocket to replace the rack, either with a new Toyota one (haven't checked if they're still sold) or a third-party reman (Cardone sold one, I think it was 1A-1000, haven't checked if they're still sold), or cleaning the noisy sensor in the one you've got.

Bob Wilson was an advocate for a non-invasive technique using electric current to burn off 'tin whiskers' that may have formed in the potentiometers. I don't know if anybody ever reported success with that, but it would be something to try without the labor of taking the rack out of the car.

Most of the other Gen 1 noisy-pot syndromes can be fixed by just opening the offending item and wiping the pot contacts and traces with contact cleaner (case in point: the heater servos), and I have little doubt it would work for the rack also, but of course in the case of the rack, that still requires getting the wretched thing out of the car and reinstalling it after cleaning.

I did once buy a junk one just to figure out how it comes apart to get to the sensor, which was rumored to be hard to do, but turned out not, after finding the trick.

Details in this thread and this one.

The torque sensor consists of two potentiometers arranged so that either direction you turn the wheel, the voltage from one goes down and the other goes up. The computer looks for that mirror-image relationship to confirm that things aren't wonkers. If it sees enough noise in one drive, it will log codes and turn off the power assist, and you will have manual steering, which is a difference you might not even notice at cruising speed, but for neighborhood speeds or parking it's a workout. Turning the ignition off and restarting will reset it, until it shuts off again. When the problem has just started, it might never shut off except on long multistate drives, because it depends on how many errors are counted in one drive.

Chart record of the sensor outputs while driving around my neighborhood. Old bad rack first:



You can see how noisy the signal (especially on the black pen) is, and how it's not always mirroring the red one. (On an old-skool recorder with real pens, you have to kind of ignore that everything the black pen does looks a few millimeters 'later' than the red one, since the two pens can't run on the same rail. But even allowing for that, you can see the sensor above is messed up.)

And then the good replacement after Toyota installed it: