tranny input shaft

I also want to address another point. The design of the BMW clutch is not really that unique. I have been working on VW cars for a while now and VW used the same transmission concept (maybe still, I have been out of the game for a while). Germans love their backwards stuff. The clutch components are more or less the same as oil head except they put ring gear that the starter engages on the clutch cover and not on the flywheel (funny looking). The pressure plate is bolted to the crank and the "flywheel is bolted to the pressure plate. Here are the components.


You can see the plate that the push rod engages(bottom right) that is very similar (same function as in the oilhead)



I have done 20-30 of these clutches and none were lubricated and none had spline wear. the input shaft did however go all the way through the clutch plate. Here is a quick picture. So there is no reason that the concept is wrong since it has been used for a while on millions of cars. Just to address the concerns of those that say "I have never seen the transmission without pilot bearing."




You can see the long push rod that goes all the way back to the cam (throwout bearing is under the cam) that is actually connected to the clutch lever (cable actuated).

esmir.celebic said:

Hello everybody.
I guess I am the epicenter of the input shaft splines now. I have been reading (and posting) in a few other threads and the short story is that for the past 7-8 years discussion has been going on that boils down to these facts (or assumptions)

Input splines fail and they always fail with the same pattern (the famous picture posted by Anton I believe)

Root causes and potential remedies are as follows:
Misalignment of the engine transmission ------ installing offset dowels done by one person so far (Mr Peterson, please correct me if I am wrong), no repeat failure, or buying new transmission + clutch parts, about $700-$1000, unknown success rate.

Shaft too short -----longer shaft designed by GSAddict (hello again, good to see you back) , one part no repeat failure.
Shaft too short ---- modified clutch hub -- unknown number of parts , unknown results.

...

Click to expand...

I appreciate your last two posts, boiling down what is known and not known and the Volkswagen example. At times the input-spline clutch-hub discussion resembles an "oil thread". Like you, I've had a difficult time finding enough data and measurements on which to base a decision on how to treat the hub/spline connection on my '04RT, which at the moment seems okay--1/16" play at the perimeter. In the absence of data, there are only a few people who have worked on enough Oilhead hubs and splines to develop an instinctive insight into what may be wrong. That being the case, I'm leaning toward these inputs: 1) Many: Check the hub-spline play through the starter opening periodically; 2) Paul: Clean and lubricate the inputs shaft and hub every 30,000 miles (I'm there now.); 3) Anton: When the input shaft has failed or worn substantially, replace it and the entire clutch assembly; and 4) GSAddict: Extend the clutch hub (can't get one of his shafts) to get a full hub/spline engagement, which I may do when I lubricate.

While I completely agree that the engine and transmission have to be precisely aligned, I see several problems: 1) You have to disassemble your transmission to make the measurement; 2) The measurement of alignment is difficult and affected by crankshaft play; and the repair requires a fairly high degree of skill.

The short, incomplete insertion of the transmission shaft into the clutch hub looks like a problem waiting to happen; and I believe in Murphy's Law, "What can go wrong, will go wrong."

Looking at the clutch disc below, I see a possible problem that hasn't been discussed: the dynamic action of the clutch hub attachment (the six metal spokes). Although they look plenty stiff with regard to their ability to resist the rotational torque of the engine, they appear to have flexibility in the perpendicular axis (front to back). The other pieces of the dynamic puzzle are the level of power transmission through the coupling (5-20HP normally, up to 80+HP under hard, high-rpm acceleration); oscillating torque forces coming from the once-per-revolution power pulse (20Hz to 120Hz); and the power difference between the Left and Right cylinders (that causes vibrations on some Oilheads). When I look at the wear pattern that we see in so many photos like the input shaft below, I wonder if there is a resonant, vibratory force at work, cause by an unstable, dynamic action. I've posted a video below of a dynamic structure, excited by a high power level, which shows graphically what I'm trying to explain. The bridge was plenty strong enough to carry its intended load, but a gale caused it to vibrate and collapse.
RB

<iframe width="420" height="315" src="//www.youtube.com/embed/j-zczJXSxnw?rel=0" frameborder="0" allowfullscreen></iframe>

Hello Roger.
I don't understand why there is such defense of the fact that the splines are not fully inserted. While Chriss Harris's colorful language and at times arrogant behavior could be a turn of to some it is unfair to assume that he is always wrong.

While discussing intricacies of mechanical design is good exercise and lots of new facts and ideas are presented,why not go with the simple fact that anybody can understand. Splines are not fully engaged. Why not start there. How can anybody make a statement "full clutch engagement will do nothing" is beyond me. Even if there is misalignment the fact you have more surface area will definitely prolong the life of the parts in the least. And I believe the best argument is brought up by you saying "IF 75% engagement is enough, how much is not enough and what would the failure mode be?" I am paraphrasing but you get the idea.
Lets not disregard the fact that 99.9% of the times the right answer is the most obvious one, especially when all others are either prohibitively expensive and still inconclusive.
If you look at it logically the only solution that makes sense is to install modified clutch, and to make things simpler you don't have to touch the input shaft at all. I am currently conducting my own experiment in this regard, with actual parts on my bike and I am hoping to have some results in a couple of weeks. What I can say so far is that I am very optimistic.
I am sure we will talk more but there will be a lot of pictures coming from my end.

EDIT: While your resonance theory might be sound it is still speculation and has no solution (viable anyway), so lets focus on stuff that we can actually attempt to fix.

EDIT: Actually even the resonance theory failure could be fixed with modified clutch since that could throw it out of the resonant zone. (theoretically of course)

Esmir

My bike had 1/16 play as measured from the starter hole and I notice upon dis-assembly that clutch hub was beginning to wear. There was no damage to the input shaft but I could tell that there was color difference in the "wear zone" at the same spot where they all fail towards the transmission. I would like to think that I caught it in time at 23k miles. I am not sure how fast the process is but I am assuming it is exponential once the wear starts.
So the lesson is, without putting eyes on the parts it is not possible to judge the true condition.

Just wanted to post since my user name has changed. so esmir.celebic is no longer (having your full name out on display leaves me feeling naked)

Resonance requires one or two masses, a spring, and an excitation. I don't see an obvious non-torsional excitation. The torsional spring is very stiff, and the torsional mass/inertia is small. That's why I discount resonance. I suppose one could say the flywheel mass/inertia is high, but as a free floating system having two masses separated by a spring, the transmission inertia is low.

The VW clutch assembly is interesting. FWD cars are pressed for drive line length etc and many omit the pilot bearing. But the VW clutch has a torsional damper in the ID that may have some radial compliance that would spare the spline from radial loading. They may also have better clutch housing manufacturing quality control. It is interesting they don' use any apparent assembly lubrication. There's no scale on the drawing, but the spline proportions otherwise seem comparable to our motorcycles.

I suspect BMW used the longer hub clutch hub for something else like inventory convenience. It may even have been an undetected design error but the cost to retrofit field examples would have been considerable, and for what purpose. Properly aligned, the attenuated spline engagement is no problem, just a red herring. But I'm still suspicious of the clutch housing casting rigidity vs their manufacturing process.

nrpetersen said:

I suspect BMW used the longer hub clutch hub for something else like inventory convenience. It may even have been an undetected design error but the cost to retrofit field examples would have been considerable, and for what purpose. Properly aligned, the attenuated spline engagement is no problem, just a red herring.

Click to expand...

I'm not a mechanical engineer but from a shade tree mechanic's common sense way of looking at this, partial spline to hub engagement makes absolutely no sense to me whatsoever. I agree with Chris Harris on this and having seen a lot of VW clutches and a few North American clutches along with many other spline hub applications, unless the spline needs to slide in and out like on a drive shaft I have never seen any fixed partial spline hub arrangements. There may be some, but I've never seen one. Except on an oilhead. That's why I think the oilhead clutch to spline design was a mistake. Given the fact there is ample room for full engagement I can see no other logical conclusion.

Some other things also come to mind:
- IIRC GSAddict's recently stated his hybrid spline has been in service for about 65K now with zero issues. It engages fully.
- Brunos machine shop is making clutch hubs that fully engage. These guys have a lot of drive line experience and I would put faith in their assessment and fix for the problem.
- Gears, splines, universal joints, bearings... do you ever see any such mating components partially inserted? Full engagement of high pressure or high torque devices just makes sense.

roger 04 rt said:

I've posted a video below of a dynamic structure, excited by a high power level, which shows graphically what I'm trying to explain. The bridge was plenty strong enough to carry its intended load, but a gale caused it to vibrate and collapse.
RB

Click to expand...

That video clip is longer than any any I have seen before. It is a perfect example of aeroelastic instability (flutter like in aircraft). The design problem was the shape and size of the bridge cross section. Amman, the designer also did the Whitestone bridge in NYC which also been modified twice to meet current wind design standards. I understand the failure happened when the wind was so strong that the bridge deck vibration mode changed from a first heave mode mode to a higher frequency first torsional mode.

Fun to see THX

I was the engineer on the tuned mass dampers at CitiCorp and John Hancock Tower.

Yeah "Galloping Gerdie" is one of the more famous engineering mistakes out here on the Wet Coast. This was the Tacoma Narrows Bridge near Seattle WA. When you walk into the front door and lobby of the Civil Engineering building at the University of B.C. there is a kiosk type affair and this same video is running for students to watch. I guess it's a reminder of what can happen if you don't get the math right! Yikes....

We had a major bridge collapse here in Vancouver back in 1958. A combination of design error and construction methods plus a small wind caused the collapse as the center span was being put in place. 79 iron workers were tossed into the ocean and 18 did not survive. Later in the day a diver drowned during rescue operations.

I guess this gives some perspective on yer big mistakes vs the smaller ones?

nrpetersen said:

Resonance requires one or two masses, a spring, and an excitation. I don't see an obvious non-torsional excitation. The torsional spring is very stiff, and the torsional mass/inertia is small. That's why I discount resonance. I suppose one could say the flywheel mass/inertia is high, but as a free floating system having two masses separated by a spring, the transmission inertia is low.

Click to expand...

I hear you. I mentioned the possibility and included the video clip because sometimes the best laid plans and designs have causes that aren't at all obvious or expected. Because we have very little measured data, there are a lot of untested hypotheses about the root cause. I'm sure the designers at BMW know exactly what is wrong and they've done a brilliant job keeping it a secret. All we're left with is our best guesses. Then again, maybe the problem is completely obvious: some misalignment, some lack of flywheel/clutch-housing flatness, and an insertion that is not quite long enough.

It is my carefully considered opinion that clutch hub spline failures, like final drive failures, do not have just one cause. I have personally seen axial runout which I know is a cause sometimes. I have personally seen lack of lubrication which I am sure is a cause, sometimes. NRPeterson is sure he has seen radial misalignment. I don't doubt it. The partial engagement of the spline may contribute, but who knows how much spline engagement is needed or enough if nothing else is wrong? I don't think there is a single universal cause and I think trying to establish that there is such a thing is futile. Like a lot of things, there are too many variables.

...problem is completely obvious: some misalignment, some lack of flywheel/clutch-housing flatness, and an insertion that is not quite long enough.

Click to expand...

My theory is that due to the incomplete engagement alignment tolerances window got really narrow and what should be tolerated by the design is no longer since one of the variables changed (in my opinion as a mistake).

Now speculations.
Gearbox is built by Getrag, clutch by Sachs.

Mistake has been either unnoticed or noticed but no real failures have been seen during the testing phase. It takes a while we all know that.

Engineering: We noticed that the spline is 5mm short.
Finance: How much will it cost to fix it.
Engineering: Probably a lot given that we built a lot of them.
Legal: If we change the design now it is an admission of guilt.
Engineering: But it is not right. The part should be fixed.
Warranty: How long will it last, enough to get it out of warranty?
Engineering: Probably in most cases.
Legal: OK, then we cover the ones that fail in warranty but under the grey area. Those are the customers we want back anyway. Do we really care about saving the money to the guy that buys used bike?
Parts: Not a bad deal for me. I sell clutches and input shafts. Fine with me.

Times have changed.

From my experience in dealing with similar situations.

cele0001 said:

My theory is that due to the incomplete engagement alignment tolerances window got really narrow and what should be tolerated by the design is no longer since one of the variables changed (in my opinion as a mistake).

Now speculations.
Gearbox is built by Getrag, clutch by Sachs.

Mistake has been either unnoticed or noticed but no real failures have been seen during the testing phase. It takes a while we all know that.

Engineering: We noticed that the spline is 5mm short.
Finance: How much will it cost to fix it.
Engineering: Probably a lot given that we built a lot of them.
Legal: If we change the design now it is an admission of guilt.
Engineering: But it is not right. The part should be fixed.
Warranty: How long will it last, enough to get it out of warranty?
Engineering: Probably in most cases.
Legal: OK, then we cover the ones that fail in warranty but under the grey area. Those are the customers we want back anyway. Do we really care about saving the money to the guy that buys used bike?
Parts: Not a bad deal for me. I sell clutches and input shafts. Fine with me.

Times have changed.

From my experience in dealing with similar situations.

Click to expand...

Interesting, but it fails to address the fact that some of the R1150 bikes go very long times with minimal, "normal" spline wear. If the short spline was a fatal flaw none would last a long time.

And this is where my theory of window of tolerances comes from. if you take it to the extreme case of minimal engagement needed to transfer torque then any and I mean 0.0000000001 misalignment would make it fail since it is already at the edge. But again from the standpoint of doing and not thinking, full clutch engagement is the first step (and obvious one) that should be taken. IMHO

cele0001 said:

And this is where my theory of window of tolerances comes from. if you take it to the extreme case of minimal engagement needed to transfer torque then any and I mean 0.0000000001 misalignment would make it fail since it is already at the edge. But again from the standpoint of doing and not thinking, full clutch engagement is the first step (and obvious one) that should be taken. IMHO

Click to expand...

I don't disagree at all - in the abstract - but somebody decided to use an "off the shelf" Sachs disk.

Of equal importance among the "they should have" things would be drill the alignment holes in the right places and don't bend the flywheel when you build the bike.

Oh, and pick a decent lubricant suited to the purpose and apply it properly!

nrpetersen said:

But I'm still suspicious of the clutch housing casting rigidity vs their manufacturing process.

Click to expand...

It is not very rigid at all - and as far as I recall not even a casting. A cast or forged ring gear on what appears to be stamped steel fingers and a hub. I think it is more rigid than on the 1100s however.

The parts themselves are probably pretty uniform as they are made by machines. There is one other major variable however and it is the human assembler and the human riders. I've always thought the way a bike is ridden might be a big factor here too.

happy wanderer said:

The parts themselves are probably pretty uniform as they are made by machines. There is one other major variable however and it is the human assembler and the human riders. I've always thought the way a bike is ridden might be a big factor here too.

Click to expand...

But according to my reading on all forums related to this issue this is not clear either. In my automotive repair career (and I would like to think that I was better than most) one rule was always "fix the obvious" regardless of customer complaint. This is my approach and I am hoping that soon we will have another long term test (my bike).

cele0001 said:

But according to my reading on all forums related to this issue this is not clear either. In my automotive repair career (and I would like to think that I was better than most) one rule was always "fix the obvious" regardless of customer complaint. This is my approach and I am hoping that soon we will have another long term test (my bike).

Click to expand...

Again - hard to argue with, except that longer shafts and longer hubs together cost $1,500 plus the labor to install the shaft in a transmission ($500 to $1,500 depending on what else is done while you are in there) plus the labor to take the transmission out and put it back in (another $1,000 to $1,500 unless you do it yourself). So the simple "fix the obvious" sets a person back from $1,500 to $4,500 and doesn't address at all the possibility the clutch housing is bent, or the cases are misaligned, and the possibility that those other clutch parts ought to be replaced too.

A person could get $6,000 into a fix for a $5,000 bike in a heartbeat!