How to tell which wheel bearing?

Not as small as you might hope, and that limited how many directions I could choose for clipping them onto those bolt heads and not interfering with the wheels. They really look a lot like the standard-issue stamped spring clamps on cheapo 12V battery chargers, right down to the red ribbed plastic handles. I've got no doubt that's exactly what they are, and JS Products just buys them from Acme Cheapo Battery Clamp Company and jams audio pickups (very thin, probably piezo) under the springs.

I was sort of wishing they had used the next smaller size of cheapo clamp, but then I'm sure you'd run into something chunkier you wanted to listen to and the jaws wouldn't open enough.

-Chap

 

Ok, I'll see if I can get to that in the next couple days.

Once this bearing's replaced, I wonder if I'll need to add a Vehicle Proximity Warning tone....

-Chap

 

Ok, here's what a pickup and transmitter look like.
Here's a new bearing fresh out of the box. Everything's nice and quiet now, so the left front was definitely the one. That means the temperature test really wasn't effective for tracking down this problem: the hub on the bad side actually showed up a couple K cooler than the one on the other side. I was aiming an IR thermometer at the hubs through the wheel center holes. Maybe it would have been better to crank the steering and try to get behind the wheel and aim at the knuckle near the speed sensor, I don't know. There always seems to be a little drag with disc brakes, and that might just heat the rotors enough to mask any bearing effect unless it's really bad.

By comparison, the multichannel sound pickup was dramatically effective in identifying the bad bearing.

By the way, what do the rest of y'all do to stake a new hub nut to the shaft? Any better idea than a hammer and drift? It seems odd when the bearing makers all write so much about handling the bearing carefully, not subjecting it to any avoidable pressure or shock so as not to brinell the races, so you press it ever-so-carefully into the knuckle (only by the outer race!), and likewise the hub into the new bearing (only by the inner race!), and gingerly set the whole thing back on the car ... then crank down the hub nut and bang a hammer on it. Seems like there ought to be a better way....

I was curious about the inner details of the bearing, wondering whether the sound I heard in the cabin (at a frequency around 12.8× the wheel revs per second) might be a reliable characteristic. Conveniently, when the machinist pressed the hub out, one side of the split inner ring stayed with the hub, leaving the bearing open to view:

He didn't save me all the balls that got away when it came apart, but it's easy to count the empty spaces in (what's left of) the plastic cage, and see this is definitely a 15-ball proposition. (Per side - this is an assembly of two matched bearings back to back, in a one-piece outer race.)

My quickie measurements are 5.55 mm for the ball radius, 34.65 mm radius of the outer race, giving 23.55 mm for the radius of the inner race (which I didn't get back from the shop to measure directly).

In this application the outer race is fixed, the inner race goes around at wheel speed, and the ball cage goes around in the same direction as the inner race, but more slowly. In an earlier post I was half-convinced that the cage speed would be ½(wheel speed)(inner race radius)/(inner race radius + ball radius). Now I'm more than half convinced. Think of spinning the inner race clockwise by some amount, for convenience let's say one radian:

The ball cage rotates less far, so there's a difference (shown in red), the distance the inner race got ahead of the cage (by rolling under the balls). The green distance is how far the cage advanced along the (nonmoving) outer race, again by the balls rolling. In a good world the balls don't slide on either race, so the red and green distances are equal.

We spun the inner race by one radian, so the cage went along through some angle θ < 1. The green distance is just θ(outer race radius) and the red distance is (inner race radius)(1 - θ). For outer race radius substitute (inner race radius + 2 × ball radius), set the distances equal, and find θ = ½(inner race radius)/(inner race radius + ball radius).

Plugging in the dimensions of this bearing, the cage goes around at ½(23.55)/(23.55+5.55) ≃ 0.405 × wheel speed. There are 15 balls, so if there's some defect on the outer race, it's going to get whacked at (15)(0.405) or 6.075 × wheel speed. That's not very close to the 12.8 I was hearing.

But what if I was hearing the second harmonic of it? 12.15 is sort of in the ballpark. And there's some room for slop, seeing as I got my 12.8 by ear matching pitches with a tuner, going by the speedometer display and the nominal revs per mile of my tires. But I'd prefer not to have to handwave away more than 5% of slop....

But there's one thing about this bearing I've left out. It's an angular-contact ball bearing. The races are shaped with higher shoulders on one side so the bearing handles thrust along the shaft axis, and not just radial loads. That means the balls actually don't roll in the same plane as the bearing, but at an angle to it:

And that means, effectively, the outer race radius is a little smaller, and the inner race a little bigger, than what I physically measured. The radii that matter are where the races meet the green line.

What that angle is, I can't easily measure. A Koyo document gives standard contact angles for their bearings as 15, 30, or 40 degrees. If I had to guess, I'd guess 40 - the shiny path on the race goes pretty far down the shoulder. So what does that do for the dimensions?

I'd better start by taking the cage radius to be my originally measured outer race radius minus the original ball radius (no correction yet): 34.65 - 5.55 = 29.1 mm. But from here out, my 'effective' ball radius is going to be (5.55)(cos 40), or 4.252 mm, and 'effective' inner race radius 29.1 - 4.252 = 24.85.

Now the cage speed will look like wheel speed × ½(24.85)/(24.85+4.252), or 0.427. There are still 15 balls, now making whacks at 6.405 × wheel speed. I still think I could have been hearing the second harmonic, which is now, hmm ... 12.81. Now that, I don't feel a need to handwave away.

So for those keeping track, if you find yourself with a speed-dependent growly tone around 12.8 times wheel speed (or g, c, and d, slightly flat, at 30, 40, and 45 MPH on stock tires), I'll vote for a front wheel bearing. I never got to take my old rear apart to see if the 10.8 I heard there was also characteristic of its dimensions, but I'm thinking probably yes. That'll be e, a, and b at 30, 40, and 45, again slightly flat.

-Chap

 

Thanks! I didn't happen to remember where I'd seen that or re-read it for this job, but I definitely remember reading it a while back. All Hobbit's write-ups are great. I notice he needed a right front and it was louder turning left; mine was left front and louder turning right. If that pattern holds for more than two anecdotes, it could be another early hint for which bearing is going.

His write-up is for an NHW20, and from it we see that the NHW20 has got "generation 3 hub bearings" at all four corners. The NHW11 has those at the rear, but "generation 1 hub bearings" at the front. (Those terms seem standard across the bearing biz; the Toyota-supplied bearings are from Koyo but I gave links to SKF's pages instead just because they've got much more informative pages on their site. Honestly, Koyo seems to have the least informative site of all the bearing players.)

Having now done one rear and one front on my NHW11, I'll offer a generalization:

• Generation 3 hub: replacement labor ⬇, parts cost ⬆(just one part, but it's ~ $250)
• Generation 1 hub: parts cost ⬇, labor ⬆(bearing is ~$40, other misc. parts, multi-step teardown, press work)

The rear job is really nothing more than (0) wheel off, (1) drum off, (2) four bolts out, (3) swap new hub for old, (4) torque four bolts, (5) drum on, (6) wheel on, (7) confirm ABS signal. Step 3 is the one that's just as easy at it sounds if you're not in the rust belt, and otherwise consumes 90% of the time and cussing for the whole job.

Front job: (0) unstake and crack axle nut loose (1) wheel off (2) caliper off (3) disc off (4) axle nut off (5) ABS sensor out (6) tie rod end out (7) ball joint from lower arm (8) axle out of hub (9) knuckle from strut (10) hub pressed from bearing (11) snap ring out of knuckle (12) bearing pressed from knuckle (13) new bearing pressed in (14) new snap ring (15) hub pressed in (16) knuckle to strut (17) axle to hub (18) ball joint to lower arm (19) tie rod to knuckle w/ new cotter pin (20) ABS sensor (21) new axle nut on firmly (22) disc (23) caliper (24) wheel (25) axle nut to 159 ft lbs and staked in place.

For step 10 I carried the knuckle, new bearing and new snap ring into the neighborhood Carquest shop and came back after step 15. The cussed-rust-belt steps are 10, 11, and 12, which I didn't have to deal with. I think I can guess the number of pieces the old snap ring came out in from the number of pry-tool gouges in the old bearing's inboard seal when I picked the stuff up. Too bad I missed out on that part.

All the steps I did myself went smoothly and nothing was difficult from corrosion. All the nuts and bolts took a touch of persuasion with a breaker bar and then came out smoothly with really clean threads. I don't know what Toyota makes their nuts and bolts out of or plates them with, but whenever I get around to building my underground world-domination lair, I'm using their bolts.

I tightened my puller moderately on the tie rod end and walked off to find my safety glasses and a hammer, and when I got back it had popped itself out. I never had to disturb the ball joint from the bottom of the knuckle and neither did the machinist; it's stubby enough to not get in the way of the press work. Where it attaches to the lower arm it's just two nuts and a bolt and nothing tricky to separate. I did confirm that it didn't feel loose and moved smoothly while I had the chance.

• Bearing 90080-36136 $38
• No-reuse snap ring 90521-79002 $2
• No-reuse axle nut 90080-17238 $5
• No-reuse tie rod end cotter pin 95381-03020 62¢
• Shop work $35
• Knowing which bearing to replace: PRICELESS

While I didn't end up needing the ball joint out, in case you ever do its slightly larger cotter pin 95381-03025 is 66¢.

At those prices I went ahead and bought parts for both sides. Like Hobbit, I'm not planning to actually do the other side until needed--after all, this bearing didn't go until 50,000 miles after the first one--but the parts are cheap and why pay twice for shipping? By contrast, at the price of the rear hub assembly, I'll leave buying that till I know I need it.

There are two tool sets made, the OTC Hub Tamer and the Ken Tool Hub Shark, to eliminate the work at a press and do steps 10-15 right on the car (also saving you steps 6, 9, 16, and 19), but they're expensive for DIY and none of the parts stores in town had them listed among their loaner tools. The OTC is also sold in the KD brand for a less premium price, but I could still take ten of these to the shop for that money and my car hasn't got that many wheels. Anyway you still have to get the old snap ring out and, seeing what a fight it gave the machinist, I'm glad I wasn't trying to do the same thing in close quarters with the knuckle on the car. These tools are specific to the generation 1 hub style so if my next car's likely to have the generation 3 style all around, I can go without.

I did buy a Schley Products 65420 axle nut unlocking tool ($17) for dealing with the staked axle nut. There's a 65400 kit providing that and a nice heavy 12 pt 30 mm socket if you don't have one (12 pt is required for the nut). The socket is impact-ready but all I needed was a breaker bar and a 1.5 m persuasion enhancer to get it loose.

I ended up doing some brake and suspension work while I was at it, for more than I spent on the bearing job itself.

-Chap

 

Is this a trick question? Maybe it's an icy field and the bulldozer is skidding? Or it's in an airplane? After all, it's only traveling "over" the field.

Umm, start with the answer to the first part. Subtract the vector velocity of the possibly skidding or flying bulldozer. Take the dot product of this with the unit vector in the direction the dozer happens to be pointing. Negate. Now add the dozer's overall velocity back in.

Veterans of this quiz get a certificate allowing them to think about the ICE, MG1, and MG2 in the Prius transaxle.

-Chap

Edit: on second thought, the dot product shouldn't be necessary. Or put another way, if it ends up making any difference in the calculation, there are bigger problems than a skidding bulldozer.

 

Hmm, I'd have thought their velocity component in machine direction goes from zero to twice machine speed over half a sprocket revolution (machine speed + (sin θ)(machine speed) for θ from -π/2 where the sprocket catches up with it on the ground to π/2 where it leaves the sprocket on top), and back down to zero the same way where the other sprocket lays it down. Of course there's more going on, like going abruptly from having no centripetal acceleration to having some. My head hurts, I need to relax and think about bearings.

Two of three falls for who sits in front.

-Chap

 

I have to admit I didn't try that (though I did spin the unloaded wheels by hand). Still need another method for the non-driven wheels though.

Honestly, now that I have the wireless stethoscope, I think I'm a convert, and will probably break it out first thing next time, skipping other methods. It gave me an unmistakable answer pretty much instantly, with everything else fuzzier or more equivocal (I've even seen posts reporting exceptions to the 'opposite your turning direction' rule).

-Chap

 

For the front, you buy a "generation 1" type hub bearing (I gave the part number in the post above, 90080-36136, about $38) plus the snap ring, cotter pins, and axle nut listed - did you read the post?

For the rear, what you're buying is a "generation 3" type hub bearing, which comes in the form of a bolt-on assembly with the ABS sensor, and runs around $250 or so.

-Chap

 

In this case, it's the bearing outer race that presses into the knuckle. The drive shaft through the inner races is not an interference fit. Toyota's instructions are to go ahead and use a shop press without any special attention to heating first. The real wild card in how nicely it comes out will be corrosion of the bore and snap ring.

The really best way for many people will be to drop off the knuckle, new hub bearing, and new snap ring at your local auto machine shop and say "let me know when I can pick it up." BTW, the ball joint doesn't really get in the way of press access to the bearing - at least, my local machine shop had no trouble doing the press work with the ball joint in place, so that doesn't need to be removed unless it seems worn and you want to change it.

-Chap