li Arc's Cochenille 85' T16

[li Arc]

That's the plan, we'll see what's what when I start pulling it apart tomorrow (hopefully). I double checked the accuracy of the machinists' parallels today with a Mitutoyo micrometer, and it falls in at about 0.001", both parallels. The bottom parallel was around 0.0005", and the gauge block was about 0.0002" deviation. It's good enough for me.

li Arc

[S900t8v]

You don bang them out. I tried that. You shim them if you are going to reuse. Or use the euro kit(I'm sure you'll go this method?)

So you used a used crush collar? How much preload did you put on the pinion bearings I kgf? Did you need heaps of torque on the nut to achieve preload or was it really easy when you did it?

[li Arc]

Yeah, I'm sure if you shim them out, that would work well too, though I would probably only re-use them once or twice max. But no, I didn't shim them, just re-used it as is, so I'm sure that's where it died. I put enough preload on the bearings to require 15lbs of rolling torque to rotate the bearing housing, as shown in the manual. That was probably one of the most highly torqued nuts I have ever had the misfortune of working with, including head bolts. It was not at all easy, and the thought of applying even more force to crush the collar is scaring me; the heavy duty vise I was using last time was having problems keeping the whole assembly stationary. If you're insinuating that I hit the crush collar when I put it together last time, I think that's probably not very likely with a previously crushed collar. I also imagine you would be able to tell when the effort required to turn the nut was like suddenly hitting a brick wall and trying to overcome that; I don't recall any experience of the sort.

I will be ordering the bearing kit with the crush collar and shim from eeuro once I've verified the bearings need replacement. I also need to get a whole bunch of other stuff from them as well, such as the transmission gasket kit, which is currently out of stock. I used Timken bearings last time around at some people's suggestion, but eeuro's kit is an SKF BR52 bearing instead. At this point, I don't think it's that big a deal esp. at <$17 a bearing vs. something like ~$45 I paid locally for the Timken ones, as long as the rebuild is done correctly.

li Arc

[S900t8v]

No I was saying you don't bang out a used collar you shim it. Banging it out to return normal height to the collar to re use doesn't produce an even surface and changes the temper.

Base on what you said li arc I don't think the pinion preload or crush collar is your cause of failure. I've pulled down 5 badly worn trannys one had 10mm of pinion play. The crush collar was not mangled and was 0.4mm less than standard height (original crush). FOr the record the discussion I had with Jim recently talks about the impossibility of the bearing races getting closer together to damage a crush collar without torquing the nut. If you have play it's from preload not being achieved initially or pinion bearings moving apart (outer races or nut spun) also for the record the actual Saab manual says to pre crush the collar with the press (when assembled between the bearings) until preload is felt to make the job of torquing the nut easier.

So what if the crush collar failed (don't know how) the recent discussion on this might be worth a read I'll post the link later. When you pull it down of there is play on the pinion bearings but the pinion nut hasn't spun then the races weren't in the housing Properly.

Before you tear it and you remove the side covers I reckon lever the main gear stack to see if there is play there. That would mean the gesrstacj components weren't seated properly on assmbly. (what I reckon) but I'm sure you know all this.

Interested to see what yOu find.

[li Arc]

Originally Posted by S900t8v

No I was saying you don't bang out a used collar you shim it.

Originally Posted by li Arc

Yeah, I'm sure if you shim them out, that would work well too, though I would probably only re-use them once or twice max.

You'll note that I was agreeing with you.

I have also been saying that yes, there will be gear stack play, but it is an effect, not the cause. In fact, I am very certain there is stack play because this would explain why driving the pinion gear and having the pinion gear be driven by the wheels gives different symptoms. If you stop and think about it, the gear stack was originally shimmed correctly. If there were play initially in the gear stack, it would not have lasted 5km, much less 1500km. If conversely, I simply shimmed the gear stack to take up any play as it is, it would not solve my problem, and instead only temporarily removes the symptoms; the root cause will return and there will yet be play in the stack. Something changed inside after 1500km, which most likely caused gear stack play, and that this cause is most likely pinion bearing failure since something needs to account for this increase in space; the amount of particulate in the transmission fluid points to bearing wear.

As for the physics behind it, yes, I did follow the thread you're talking about, but I also understand that there are additional forces that act upon the pinion shaft than what you're all talking about. What that thread failed to account for are two major forces that could cause pinion bearing failure: longitudinal forces (the ones that Jim could not explain that push the pinion bearings closer together), and cross axial forces (that cause the pinion shaft to pivot at the pinion bearings).



The hollow arrows denote parts that are driving the assembly, while solid arrows denote parts that are being driven, in a normal forward-gear, accelerating condition. Arrows with 'R' in them indicate rotational direction, and arrows are paired so that each part shows a direction of rotation (as seen from the diagram) and a longitudinal force that is being generated. In this diagram, 3rd gear is engaged and is being accelerated by the engine. It is these longitudinal forces that can crush the pinion bearings together, depending on how much torque is being applied. How is there a longitudinal force? Because the gears are not straight gears, they are helical, like an inclined plane. In physics, using an inclined plane to push another inclined plane of the same angle will generate forces that can be broken down into two components: a vertical force (that in our case rotates the gear, and may also cause cross axial forces) and a horizontal force (that in our case pushes or pulls the gear stack toward or away from the pinion gears). If the inclined plane is at a 45 degree angle, the horizontal component will theoretically be equal to the vertical component of the force. Ultimately, in the diagram the input shaft (red), which is longitudinally detached from the main pinion shaft, is being driven towards the left, the input companion gear section (blue) is driven to the right, the cluster gear (green) is driven to the left, and the 3rd gear (purple) is driven against the pinion bearing towards the right; the pinion gear will drive itself and the shaft to the left against the other pinion gear when it becomes loaded in a forward motion against the wheels (diff), hence crushing the pinion bearings.

In the event that the crush collar isn't present, I could very well see the loads exceeding bearing specifications, and the races and rollers will wear prematurely, forming enough of a gap to manifest as gear stack play. This would support the amount of bearing wear that is apparent in the fluid.

There are 13 bearings in the transmission that could exhibit excessive wear: since the diff and inner driver bearings are unaffected, that eliminates 6 bearings. The symptoms also are inconsistent with input shaft and chain drive bearing problems, eliminating another 3 bearings. We are left with two pinion bearings, a fore, aft, and mid cluster shaft bearings. While the cluster shaft bearings could possibly show wear and perhaps generate some type of cluster gear play, the fore cluster shaft bearing (with the rollers in the input companion gear) rarely wears from others' experiences, and would also not be consistent with these symptoms, and the aft cluster gear bearing was re-used after an inspection of the cluster gear shaft and the aft bearings showed low wear (ie. it wasn't wearing before, don't think it'll wear now). We have input bearings and mid-cluster shaft bearings left. Something has worn, we know that for a fact. And what wears most in a transmission is almost always bearings. If they are pinion bearings, we will see gear stack play.

We'll find out tomorrow.

li Arc

[S900t8v]

I see what you are trying to say but you have the direction of forces incorrect.

You're talking about longitudinal forces - axial forces and longitudinal are the same thing. (if you read the other thread you will see me talking about axial force a great deal) I don't think the forces are cross axial either, but rather radial forces.

When going forwards the input shaft/(mainshaft gear that is engaged) spins anticlockwise and the cluster gear spins clockwise. The helical cut on the cluster gear creates axial forces that push it towards the diff end (hence the primary thrust washer being at this end of the cluster gear). Consequently the helical cut on the mainshaft gear causes axial forces that push the mainshaft towards the primary end loading the rear pinion bearing - the bearing closest to the diff (this is why saab upgraded the rear pinion bearing and why it wears out first!!!!!) . I should also note that under acceleration the cut of the ring and pinion gears (spiral beval) creates an additional axial force that tries to push the pinion gear away from the ring (loading the rear pinion bearing as well!)

Because the pinion bearings are at a fixed distance, when you load one bearing, the load decreases on the opposing bearing, the very reason preload is used is to ensure that the unloading of the opposing bearing does not compromise the tapered bearing setups ability to tolerate radial loads. That is why when pinion bearings wear out and play develops the subsequent radial loads result in destruction of the gearset/total catastrophic failure.



See the above image from a gearset design book I have.

The saab gearset design when in the forward loading (acceleration) corresponds with the top right image, and when in the reverse loading (deceleration) corresponds with the 2nd from the right.

Comparing to the image you posted (which accurately depicts the helices on the mainshaft/cluster gears) you will note that in the top right hand imagewhen under acceleration (mainshaft anticlockwise, cluster clockwise) axial/thrust forces (depicted by the dotted arrow line) on the mainshaft are pushing it towards the primary end and the axial forces on the cluster gear are pushing it towards the diff end.

You will also note that that there are no OPPOSING forces on the mainshaft or cluster gear. the axial forces only go in 1 direction when under 1 load characteristic (accel/decel). There are no forces pushing from both ends at the same time. The very principle reason why opposing taper roller bearings are used is to be able to handle axial loading in both directions. As a side note a single taper roller bearing can handle small radial loads, but when two opposing taper rollers are used the radial load handling goes up dramatically. only one bearing is taking the primary load at any one time, the second is acting as a support bearing to help limit radial play and to take the primary role when direction of axial force changes (when reversing or under deceleration)

I don't know how I can make it any clearer. That is out of a book written by gear designers/engineers. There are no forces pushing the pinion bearings together, even if there were even forces that were 'trying' to do this (THERE ARE NOT!) - the forces would dissipate on the pinion nut (which would load the front pinion bearing and unload the rear). An example of the effect of axial loading towards the diff end is when there is gearstack play and you see smashed stack height shims from the gearstack hammering against the pinion nut - it happens on reverse loading/deceleration I have torn down trannys with this phenomena, but because the pinion nut and pinion gear face are immovable objects no forces can push the races closer together and the crush collar has no 'extra' damage.

I'll say it again lol - because the pinion gear head and the pinion nut are FIXED there are no forces that push the bearings closer together, any forces aimed at moving the pinion bearings act as an axial force on the pinion shaft and can only go in 1 primary direction at any one time. The forces countering the axial loading are - one of the two opposing taper roller bearings - not both at the same time. When one bearing takes primary axial load, the opposing takes primary radial load and vise versa. The taper bearing cup angle and the distance between the two bearings plays a role in how much radial load the bearing can handle, shallower cup angle ( more squat looking bearing) = more radial load, steeper cup angle = more thrust load.

The helical gear design cannot create opposing forces at the same time, you would need herringbone gear principle to create opposing axial forces you describe - and IF they were herringbone gears then the forces would cancel out and there would 0 loading of components and still no risk of the pinion bearings moving closer together.


On an unrelated note, did you lubricate the pinion bearing with assembly lube or oil, and were the bearing well oiled before you took preload readings (and did you rotate the housing 20+ times to seat the rollers?). When I did my preload it was around 8.5kgf until I rotated the housing a few times in each direction with oil. It then dropped down to 6.5. Then finally at 5.8. Right in the middle of the 4-7kgf spec.

[li Arc]

Originally Posted by S900t8v

I don't know how I can make it any clearer. That is out of a book written by gear designers/engineers.

I couldn't agree more. And you are contradicting yourself. The book you cite is indeed correct, under acceleration, the upper right gear set depicts the axial loads that the gearsets are under; however, there is no diagram that depicts the axial loads under deceleration. If you look more carefully, the second diagram from the right which you indicate shows deceleration in fact shows the same gears but driven in the opposite direction, which does not occur (unless in reverse, and only then, the reverse gear). If you compare the input shaft and companion gear sets and axial loads, the 'engineering gear book' shows exactly what I dictated in my diagram (red and blue components). If you look closely, the input gear is driven this way, exactly as you say, counter clockwise. By extension the cluster gear and third gear rotation and axial loads are as I state in my diagram (purple and green), which cause 3rd gear to place an axial load against the fore pinion bearing. At the same time, the pinion gear driving the load (diff + wheels + weight of car) is pushing the gear against the aft pinion bearing. Hence accel = pinion bearings are crushed against each other.

If you still don't understand, perhaps a simple force diagram:



Let's say you have two inclined planes moving toward one another. If force 4 is immovable, ie. it's lying on the floor, force 1 will upon contact cause forces 2 and 3 to form. Force 1 is a driving force, and it will cause the green block to move to the left, while causing the purple block to move to the right. The magnitude of these forces depend on a variety of conditions, but the directions are correct, and the sum will equal the magnitude of force 1 (minus any frictional losses). So if instead of force 4 being stationary, let's give it half of force 1. Now the magnitude of force 2 and force 3 will equal half the magnitude of force 1 (minus any losses), while the other half of force 1 is transferred to force 4.

This example in fact shows the point of contact between 3rd gear (purple block) and the cluster gear (green block), where the cluster gear is driving the 3rd gear. Force 3 drives the gear stack against the fore pinion bearing when there is a load (ie. force 4 < force 1). If that's not clear enough, I'm not sure what is.

Thus far, you haven't been able to account for the bearing wear, nor the gap that formed causing gear stack play, whereas I can explain both. You think that the pinion nut is absolute, that it is immovable, which is incorrect. The way a nut or even a bolt works is that it becomes immovable (ie. torqued tight enough) when it causes a tension buildup between the nut or bolt head and the contact surface. That's right, an immobile contact surface will indeed cause a tension buildup that will generate enough pressure between the contact surface and the fastener that static friction will be enough to keep the fastener in place. Now suppose that contact surface were to move away from the fastener, releasing fastener tension. What's to keep it in place? Nothing. In the case of the pinion nut, maybe the stake or loctite, but with the amount of coaxial rotation energy, the stored inertia in the nut, the heat generated, and the impact of the pinion bearing and the gear stacks effectively breaking it loose, it may as well not have been staked or have thread locker at all. With the normal direction of rotation of the pinion shaft with pinion bearings that have moved together, the nut will most likely re-tighten against the pinion bearing, creating a gap for gear stack play. This is why the gap was non-existent when the gearbox was first put together, and operated fine for 1500kms before it went south.

I take the time to try to explain the physics so that no one else repeats my mistake. However, if you choose to remain steadfast in your theories, then I am done trying to explain, as this is taking an incredible amount of time to write up and elaborate. At this time, all theories are irrelevant, as I'm no longer interested in fantasizing about the problem, and am more interested in solving it. We will soon enough have all the evidence to show what I'm seeing. Since I have not completed the teardown yet, I will keep from prematurely stating a cause of failure, but I will post my findings when I have conclusive evidence.

As for the pinion bearing preload, I used assembly lube on all parts, and similar to the input shaft, rotated the housing in each direction something like 20 times before I would continue tightening. I did this each time I took a measurement. The final measurement was 15lbs (6.4kg) required to roll the housing, with a string attached to the circumference of the pinion bearing housing. I did not have a torque wrench to measure with.

li Arc

[S900t8v]

I think we have missed a major concept here Li Arc.

I think we are both wrong. You said the forces are reversed under deceleration, but if the forces created by mainshaft/cluster gear were reversed under decel then by the very same principle so would the ring/pinion forces and the end result would be what Ive been saying all along and that is that the axial forces on the pinion gear are only traveling in 1 direction at any one time not into each other. WE agree the cut of the helical gears and the spiral bevel gears are what create axial forces, and so by that principle (shared and common between the two) you describe an event then BOTH components must act the same.

(I was wrong about the 2nd image from the right, that shows reverse only)

The thing is -I'm so sure about this because torque on the gears is reduced the axial forces created are reduced, to the point where they may become close to neutral (no axial force), but if the direction of rotation remains the same the forces cannot be reversed to go in the other direction, you opened my eyes to this, the helix/cut has to change for the forces to be reversed and the helix is not effectively 'different' unless in reverse direction of travel, just as the image shows.

Anyway -the rear pinion bearing wears out first, saab upgraded this bearing, that is because (and you agree - the far right picture) in the forward direction whether accel or decel the axial loading on the mainshaft primarily loads the REAR pinion bearing. The primary forces on the cluster gear push it into the diff end (hence the thrust washer and the upgraded thrust bearing put here in later transmissions)

The final point is I don't get your theory, you have pinion bearing play but somehow the crush collar has crushed down and the pinion nut has spun? Both of those things would be awesome for the gearbox and would prevent pinion bearing play developing, it's play that destroys a gearbox, not too much preload (unless the bearings spun which they would not have IMO) Too much preload can shorten the bearing life expectancy, but is a positive thing as it introduces stiffness into the pinion bearing assembly, the play will be what has destroyed the gears. Furthermore the reverse gear is splined, and cannot exert rotational force on the nut, there is no way for that nut to come undone other than vibration or a spun inner race (improbable if not next to impossible)

Regarding gearstack play, when pinion bearing play develops it the axial forces pushing the mainshaft push all the mainshaft gears into the cluster gears. Because the cluster gears are pretty much immovable this transmits to extra loading on the thrust surfaces of the mainshaft gears. The mainshaft gear thrust surfaces wear and subsequently play develops in the gearstack. The loading/unloading of the mainshaft by axial forces causes the reverse gear to be hammered into the shim stack destroying the shims.

You know how I know this is what happens? Because in transmissions with pinion bearing play the axial force that is transmitted onto the thrust surfaces on the mainshaft gears (instead of disappated by the bearings) causes the 4th gear bushing to weld to the 4th gear. We know the pinion shaft moves violently as you often see damage on 4th gear guide ring on mainshaft as it grinds against the side face of 4th cluster gear.


I only offer this counter argument because I don't agree with your theory. It's a public forum I am open to being wrong but nothing you have said is really conclusive IMO.

And away from this discussion and trying to be helpful, just my point of view on a side note from a family friend who rebuilds race engines, he said he doesn't like assembly lube because he built up a couple of drag car motors with it and warmed them up (with oil) and spun EDIT ROD (bah not crank lol) bearings. he started using medium-thin oil for assembly and never had the problem (he'd assemble, put oil in, fill with oil, then prime before starting) It's one mans opinion but google finds other people with similar stories?

Also I don't get how you could have got assembly lube evenly over all the rollers and all the race surfaces etc? Isn't it like a paste when compared with oil? if you subject bearings to preload and then try and roll them when they aren't evenly lubed they will skate and this causes microabrasions /micro spalling which sets the bearings up for rapid failure. Also I think assembly lube has different friction characteristics to oil, and so using it when setting the preload might have caused you to set the preload way to high which lead to rapid bearing wear and failure?

IMO this seems way more likely as a possible cause for Pinion bearing failure. Just my opinion so don't bite my head off lol

[li Arc]

Originally Posted by S900t8v

...in the forward direction whether accel or decel the axial loading on the mainshaft primarily loads the REAR pinion bearing.

This is where you seem to get confused. If under accel the axial loads exerted on say 3rd gear pushes it to the right towards the pinion bearings, then under deceleration the axial loads will push the gear away from the pinion bearings.



This diagram is split into two sections: Dia 1 and 2 show the green teeth driving the purple teeth, while Dia 3 and 4 show the purple teeth driving the green teeth. In this example, the green teeth represent the teeth on the 3rd gear companion gear on the counter shaft at the contact point, while the 3rd gear itself is represented by the purple teeth. The helical directions are representative of the actual gearset as shown in my original diagram.

1) In diagram 1, we see under normal acceleration the cluster gear applying a source force, Fsc, against a stationary 3rd gear; the cluster gear is driving the 3rd gear. Upon contact, an axial load, Fc, is produced, which acts to push the cluster gear toward the front of the car. Fc is proportional to Fsc and the helical cut angle.

2) In diagram 2, at this same point in time, we have the 3rd gear "pushing back", according to Newtonian physics, with a load force, FL3, that equals Fsc (while the 3rd gear remains stationary). This load force will generate an axial component Fa3, which pushes the 3rd gear the opposite direction as Fc, into the pinion bearings. Because FL3 is proportional to Fsc, Fa3 is thus proportional to Fsc and the helical cut angle.

3) In diagram 3, we now have a normal deceleration condition where the 3rd gear is driving the cluster gear instead. A force Fs3 will now generate an axial component F3 when this 3rd gear tooth comes into contact with the cluster gear, which actually reversed the previous axial load and now forces 3rd gear to push the remaining gear stack in front of it toward the front of the car. Again, F3 is proportional to Fs3 and the angle of the helical cut.

4) Finally, in diagram 4, the engine will now generate a load force FLc that pushes back against Fs3, generating a component Fac that moves the cluster gear towards the rear of the car. Again, because FLc is proportional to Fs3, Fac is proportional to Fsc and the helical cut angle.

You'll notice the direction of rotation is always the same, the helical cut is always the same, but changing the driving component vs the driven component can reverse the axial loads, given everything else is constant. Thus, under acceleration, there is an axial load (fore of the pinion bearing assembly) that pushes against the fore pinion bearing, while under deceleration there is no axial load on the fore pinion bearing.

Originally Posted by S900t8v

Both of those things would be awesome for the gearbox and would prevent pinion bearing play developing, it's play that destroys a gearbox, not too much preload

Acutally, this is correct and incorrect. When the pinion bearings are crushed against each other exceeding specifications, they wear prematurely, which then causes too little preload when these forces are relaxed. With axial loads that change directions when driving or being driven, the gear stack will move back and forth within that gap range along the pinion shaft, causing dynamic shifts in pinion shaft location. Pinion bearing play can be the root cause for gearbox failure, but it is the effect that this has on the alignment of the gear stack against the cluster gear that ultimately leads to failure. With loose pinion bearings, there can be cross axial movement of the gear stack or stack play, either of which will ultimately misalign to the point the gears will hit each other unsynchronized, causing force spikes and stresses that the housing cannot contain and everything blows up.

Originally Posted by S900t8v

Furthermore the reverse gear is splined, and cannot exert rotational force on the nut, there is no way for that nut to come undone other than vibration or a spun inner race (improbable if not next to impossible)

Again, the other gears aren't "turning" the nut, the nut is already loose without the pinion bearing to torque against. The gear stack and the pinion bearing are now free to cause impact on both sides of the nut, and along with heat can easily free the nut even with a stake in it. The pinion shaft turns counter clockwise in normal acceleration, and thus with some inertia the nut can tighten down against the now loose pinion bearing, creating the gear stack gap.

Originally Posted by S900t8v

Also I don't get how you could have got assembly lube evenly over all the rollers and all the race surfaces etc? Isn't it like a paste when compared with oil? if you subject bearings to preload and then try and roll them when they aren't evenly lubed they will skate and this causes microabrasions /micro spalling which sets the bearings up for rapid failure. Also I think assembly lube has different friction characteristics to oil, and so using it when setting the preload might have caused you to set the preload way to high which lead to rapid bearing wear and failure?

This seems like a snake-oil argument as I doubt the pinion bearings failed on account of assembly lube. I'll admit I can't verify the effectiveness of assembly lube on bearings, but once you load the bearings and rotate they won't 'skate' and the lubrication will be complete. The assembly lube is more viscous than motor oil so that it can adhere to assembly surfaces, but it is not a paste. There are those who don't lubricate bearings at all as they already contain lubricant from the factory, who assemble and operate without problems in flooded environments. In our cases, the transmission fluid will overcome any assembly lube within a couple minutes of operation, and the assembly lube is designed to dissolve and integrate into whatever oil it is washed away in.

Anyways, take what you will from this, but I'm not interested in a debate. I'm going to find the problem and deal with it. You'll forgive me ahead of time if I fail to respond to any further attempts at inciting a debate.

li Arc