Rennlist - Porsche Discussion Forums

Rennlist - Porsche Discussion Forums (
-   928 Forum (
-   -   Failed torque converter bearings or why do we have these flex plate issues??? (

Nicole 09-23-2005 09:36 PM

Docmirror: This is very interesting stuff, but again it does not explain why some people say that after a while there is no tolerance left for adjustment on the shaft. I think it was Bill Ball who told me that his flex plate clamp is quickly approaching the end of the adjustment range.

This can only happen when either
a. the shaft moves and stays forward
b. the shaft gets longer

How could you explain that?

Tails 09-25-2005 11:30 AM

Just a few thoughts on the forward migration of forward flexplate clamp assembly applying load on the after end of the thrust bearing via the crankshaft thrust collar.

A lot has been written on the subject including 10% extra torque on the forward flexplate clamp’s allen headed set bolt to prevent its migration, however, test have been reported to have been carried out on the frictional resistance of this extra clamping torque and the force need to overcome the frictional resistance of the clamp was nearly the same axial load applied to overcome the frictional resistance when the clamp was torqued up to the original specification.

Locktite 290 has been applied to the splines with claimed success on stopping this forward migration of the forward flexplate assembly on the drive shaft and others in the racing area have modified the clamp to a collet type arrangement.

I initially released the clamp on my 1990 928 S4 in February 2004 and it moved 3.8mm and the thrust bearing axial clearance was 0.008" or 0.20mm which was well within the tolerance given by Porsche AG of 0.110 to 0.312mm.

A week ago I released the clamp bolt and the flexplate sleeve moved approximately 2mm aft towards the transmission and the axial thrust bearing clearance is still 0.20mm measured with a dial indicator. I renewed the clamp set bolt and re-torqued the set bolt 10% over the specified torque. At this juncture no Locktite 290 had been applied, but I am considering it.

So why does this migration happen and what causes the forward flexplate clamp assembly to move forward along the drive shaft splines?

I suggest that the clamp does not move forward along the drive shaft splines due to ballooning of the torque converter or the increased length of the torque tube drive shaft as some have suggested, because if the drive shaft was to move forward via either of the two mentioned methods and the forward axial force acting on the forward flexplate assembly was strong enough to overcome the frictional resistance of the clamp then the flexplate assembly would move aft along the drive shaft splines towards the transmission. This does not happen, as all reports indicate that it moves forward toward the engine.

Let recap, the set up of the engine, torque tube, drive shaft and the associated flexplates fitted to the automatic transmission models (My 1987 to 1995) is basically one bolted up integral unit with the drive shaft connected to the engine flywheel and the torque converter by flexplates. The aft flexplate clamp should not move as is clamped and locked in position via the allen headed set bolt that is located in a groove through the splines of the drive shaft, so the forward flexplate is the only one that can move, as it is held in position via the clamp through frictional resistance.

However, it has also been reported that the after clamp set bolt is not a “fitted” bolt in the groove, so there is some room to move axially until the set bolt outside diameter abutts the surface of the groove if there is sufficient axial force to overcome the frictional resistance imparted by the clamp, so both clamp assemblies can move. It would be interesting to investigate whether the aft flexplate actually migrates in service (depending on the positioning of the aft flexplate in relation to the groove in the drive shaft, and whether it moves forward or aft along the drive shaft splines. If in fact it does move, its movement would be limited by the clearance of the groove around the set bolt.

The most plausible explanation, as to the cause of the migration of the forward flexplate clamp assembly is by the torque imparted into the drive shaft whilst the car is under acceleration.

However, any increase or decrease in the axial length of the drive shaft caused by ballooning of the torque converter, growth of the drive shaft or torquing up of the drive shaft during acceleration is contained by the crankshaft thrust bearing and the deep ball bearings just forward of the torque converter and in consequence any movement of the drive shaft should be taken up by the forward and aft flexplates, flexing.

It has always been reported the forward flexplate clamp migrates forward not aftwards, this therefore eliminates the growth of the drive shaft or the ballooning of the torque converter as the reason for the forward migration. If the drive shaft did increase in length or the torque converter expanded (ballooning) it would cause the forward flexplate to move aftwards if the axial force applied overcame the frictional resistance imparted by the clamp .

The forward movement of the forward flexplate assembly has been verified by, measuring the bowing of the flexplate towards the engine via a straight edge, the painting of the flexplate clamp assembly and the spline area of the drive shaft that gives a clear visual area at the rear of the flexplate sleeve where there is band of no paint on the drive shaft splines, and when the clamp bolt is slackened the flexplate splined sleeve moves aft toward the transmission.

From the above I believe that the migration forward is caused by torquing up of the drive shaft via the transmission of power reducing it length.

Hypothesis of flexplate forward movement.

When a rotational force is applied to a shaft the shaft twists and the angle of twist can he measured and this angle is directional proportional to the quantity of horsepower (torque) applied to the shaft (as a matter of fact this is the method used to determine the horsepower transmitted by the propeller shafts in large marine steam turbine installations in ships). As the shaft is elastic it will return to its original position and length once the driving force has been removed unless the elastic limit of the material is exceeded by excessive twist.

This angle of twist can he calculated by applying the maximum torque (rotational force), delivered by the engine (in ft lbs on N/m), the length of the drive shaft, its diameter and its modulus of elasticity. Once the angle of twist is known the shortening of the shaft can be calculated.

The problem of migration appears to be from MY 1987 with the introduction of the S4 automatic model until the 928 was discontinued in the GTS form in 1995. From the three dimensional view of the various torque tubes from the 3 speed auto and the introduction of the 4 speed auto transmission in the WSM and PET, I cannot determine whether there has been any major modification to the forward flexplate assembly, so could this phenomenon of migration been caused by the increase in horsepower of the S4 through to the GTS autos or a change in the elasticity or diameter or the drive shaft?

If we accept this forward migration of the forward flexplate assembly is due to the driveshaft shortening then the possible explanation/hypothesis as to why the forward flexplate migrates forward along the drive shaft splines is as follows:

The automatic transmissions fitted to all 928s are fitted with torque converters. Torque converters allowed for the removal of the clutch on automatic transmission automobiles, which allows for smoother take off, adjustable smoothness of gear changes and also for idling whilst the vehicle is stopped (probably not all the reasons, but sufficient for the argument).

When the throttle is applied, the take up of the driving force is initially by the torque converter due to the increasing of the sheering force of the oil circulating with the vanes of the torque converter and as the rotating speed of the forward rotating element of the torque converter is increased so does the velocity of the circulating oil and in consequence the force to continue sheering the oil increases until the aft rotating driven element speed nearly matches the speed of the forward driving element. realising that there is always a small amount of sheer. The force to sheer the oil at high rpm is adequately demonstrated when torque converter stall speed test are carried out as the S4 stalls out at around 2,400rpm from memory.

The action from the torque converter on the drive shaft is a gradual twisting up of the drive shaft as it absorbs the horsepower being applied until the reaction of the drive shaft, angle of twist, equals the horsepower applied (Newton’s 2nd Laws of motion (I believe) “For every action there is and equal and contrary reaction”). The drive shaft angle of twist reaches it equilibrium point when the car reaches it cruise speed on a flat consistent surface road with constant wind resistance etc. Unfortunately this point of equilibrium is never constant due to road surface conditions, bends in the road, hills, overtaking etc.

If throttle is depressed all the way (pedal to the metal) for fast acceleration from a standing start the torque developed through the gears follows the torque/speed curves and the angle of twist of the shaft follows these curves. These different loads are applied to the drive shaft as the transmission changes gears and, as there is also a ‘kick down’ switch installed, this causes more cyclic loading during aggressive acceleration. In addition, depending upon the setting of the pressure modulation valve within the transmission it is capable of varying the mode of change from an almost undetectable change (via slippage) to a harsh “kick in the back” type power change.

The various torque loading on the drive shaft are always more excessive during the ‘change up’ or accelerating mode than the ‘change down’ mode when slowing down mode while the engine is applying a braking force.

During the acceleration cycle assuming the end play clearance of the thrust bearing is around 0.20mm or 0.008” (as with my car) and assuming the aft flexplate is firmly locked in position by the groove in the drive shaft, then the forward flexpate is flexed in the aft direction during torque up. In addition there could even be a shock loading imparted momentarily when the thrust bearing collar is bumped into the forward end of the thrust bearing during acceleration. If the flexplate has a pre load (as initially required by Porsche AG or the flexplate has already started it migration) this shock loading will always be present.

The shock loading and the torquing up force could be of sufficient magnitude to over come the frictional resistance of the clamped flexplate sleeve journal to the drive shaft and the forward flexplate assembly moves a very small amount, a matter of 10th of a mm or thousands of an inch during each cycle of the above, that is at every gear change.

You can imagine the number of torque up cycles that occasions in a daily driver and the distance the forward flexplate sleeve migrates can soon mount up. Some posts have mentioned that when the flexplate clamp sleeve starts to move wear has started as well, so the frictional resistance will also start to decrease. This migration forward on the drive shaft can ultimately cause excessive load to be applied to the thrust bearing and in some case can result in the thrust bearing failure.

Please note that the collar type thrust bearing is possibly the least efficient type of thrust bearing and when the pressure gets too great the oil film will break down causing face to face contact, excessive heat generation and subsequent bearing failure especially if the faces are kept in continuous contact with excessive load. Reciprocating engines are not usually designed to accept continuous crankshaft axial loading.

The manual transmission 928 only have cyclic loads applied during clutch operation, which only load up the thrust bearing for short intervals and once the clutch is unloaded the oil will cool the thrust bearing, whilst maintaining the oil film.

Why do some cars suffer failure and others do not?

Some cars may suffer from a double loading effect especially after a torque tube change. As previously stated, the aft flexplate assembly is held in position by a clamping set bolt via friction and also by the groove location. If the set bolt, when assembled, was located with its outside diameter abutted to the aft end of the drive shaft groove and the above mentioned migration also occasioned with the aft flexplate, until it takes up the clearance, this could possible give an initial double load on the thrust bearing.

The cyclic torquing during acceleration gear changes causes the drive shaft to be drawn further out from the flexplate assembly until a resident state is achieved. This resident state could happen when the flexplate moves forward sufficiently that the flexplate sleeve follows the shortening of the drive shaft aft without any further migration forward. This could occasion when the aft axial force from the bowed flexplate matches the residual frictional resistance remaining via the clamp. This reciprocating movement should occur mainly during acceleration and would cause a fluctuating load on the thrust bearing and when the car has reached it cruise state it will maintain a continuous load on the aft face of the thrust bearing via the collar on the crankshaft.

This residual state seems to occur around 4mm load up of the forward flexplate and this load may be sufficient to cause failure of the thrust bearing in some engines over time, as the load could cause the oil film to fail with face to face contact, heat generation and bearing failure. The failure of the oil film could also be dependent on the type of oil used in the engine and the type of driving, aggressive compared to milder acceleration.

If this hypothesis is indeed correct then the flexplate clamp should be released at least at each service period ( every12 months) and re-clamped to the specified torque plus 10%. More frequent observation is possible by painting the forward flexplate clamp sleeve and drive shaft and at regular intervals checked for migration by viewing through the hole at the bottom of the bell housing using a small 12 volt globe on the end of a probe (this can be made up very simply). I would suggest that any migration detected should be released immediately.

As a further precaution some owners may wish to apply the Locktite 290 to the clamp sleeve splines and splines of the drive shaft to stop this forward migration. If this is done and is successful in stopping the migration, then the thrust bearing will only suffer fluctuating loads during acceleration and when there is no migration there will be no residual axial load whilst cruising. This state will actually emulate the manual gear box operation.

Tails 1990 928S4 Auto.

Vilhuer 09-25-2005 12:40 PM

Originally Posted by Tails
The most plausible explanation, as to the cause of the migration of the forward flexplate clamp assembly is by the torque imparted into the drive shaft whilst the car is under acceleration.

First of all thank you for very good arguments. Unfortunatelly I have to disagree a little. Since it seems about 4mm is max. amount of flex on front plate, in order of any theory to successfully explain TBF phenomenon it needs to be able to explain this 4 mm flex also. I haven't done calculations on how much TT center shaft would contract but I find 4mm quite a lot. Also we must keep in mind that flex plate will resist clamps movement and flex somewhat before clamp movement happens. This means center shaft would need to contract at least few mm's more than 4mm to be able to create 4mm offset on clamp. I would expect this movement also to happen at once, not little by little over longer time period.

I'll throw in one more theory I read once. Not sure if this would solve the mystery or not. It could be one more piece of the puzzle where many different phenomenon work together and create TBF as end result.

If we think crank and torque converter as end points and static we have two separate systems that will expand and contract when temperate of the parts change, one inside other. On outside there is aluminium clutch bellhousing, (at least partly cast) steel TT outer tube and aluminium torque converter housing. Inside are flywheel, front flex plate, TT center shaft and rear flex plate, all made of different steels. When both front and rear clamp are tightened all parts have their respective dimensions on temperate of the workshop. This locks them together from front and rear ends. TT bearing in the middle don't really play major part in this as center shaft moves inside them quite easily compared to what it takes to seriously flex flex plate.

When car is driven, parts will heat up and/or cool down as result of outside temperature, heat from engine cooling system, warm exhaust, etc. What effect these temperature changes have to above mentioned parts? They can be seen as two rods. One consisting of aluminium-steel-aluminiun and other steel-steel-steel-steel. Will temperature changes into these seven parts be enough to cause one set to expand or contract more than other?

I don't know what effect if any temperature has to all this. Just one more thing to consider when calculating what's going on.

The manual transmission 928 only have cyclic loads applied during clutch operation, which only load up the thrust bearing for short intervals and once the clutch is unloaded the oil will cool the thrust bearing, whilst maintaining the oil film.
I don't see manual having quite same loads as clutch plate bore can move very easily on clutch shaft. It is always position of least resistance by far and should never require any real pressure to move.

Schocki 09-25-2005 05:25 PM

2 Attachment(s)
Just wanted to add-on my latest findings:
I replaced the torque converter bearings last night ad they had quite some play. Each bearing had a 0.4mm for/aft movement. This adds up to a maximum of 0.8mm movement.
For my understanding the situation is clear:
With brand new bearings and no axial movement the front flex plate does not move. Just because there are no forward/rearward vibrations present. As soon as the bearings start to wear the flex plate is slowlly pushed forward. Finally when there the infamous "rear axle" vibrations are present at 3000 rpm or higher speeds, the bearings are close to be worn out ad need to be replaced.

I think Porsche understood the problem and allowed to torque the pinch bolt to 110% in an attempt to stop or minimze the flexplate movement. It is to bad that the pinch clamp assembly can't be bought seperately. Otherwise I would have tried to use to clamps @ 110% torque to stop the movement. I will use Loctite 290 and painted markings on the shaft. Below some pictures of my bearing measurements the a feeler gauge.

borland 09-25-2005 07:39 PM

Way to go Schocki !

The WSM procedure includes heating the bearing housing in an oven. The bearings will fall right in with that heat applied. But you still need a press to remove the old bearings and put the shaft back in place. That's the way I did mine, per the WSM.

During the assembly, I lubed my splines with moly grease (per WSM) and torqued the pinch bolts to the original WSM spec. With the new bearings, I had zero movement of the forward flex plate when I checked it after 2000 miles.


Schocki 09-26-2005 12:16 AM

5 Attachment(s)
I used an electric heat gun and heated the housing with it. Worked very well, no problems.
If you don’t have this there is als an opposite way.

Put the bearing in the freezer!!!

Took the torque tube today and measured my old torque tube against the rbuilt unit frim Mark Anderson.
I tried to be very precise and here is what I got:

Rebuilt: 136.9 cm
Old: 137.08 cm

That is a total difference of 1.08 mm or 0.0425197 inches.

After 110.000 miles this a neglectible differce!

I do not believe that the torque tube shaft extends in length over the years. If it is getting “longer”, it is due to worn torque converter bearings and a little bid of movement due to tolerances in the rear clamping assembly.

I also found that the front flexplate is NOT made out of a bi-metal! It is simply made of to high grade steele metal sheets. These are bolted togther in the center by the clamping shaft bolts and later when the flexplate is mounted by the 6 flywheel bolts. Other than this the sheet metal is not coected or welded together. This is the reason why the flexplate ALWAYS comes back to it’s original completely straight form. If it would be a single sheet, it would bend over the years and would not completely release all tension. Very simple but a very effective design.

I also checked my crank for thrust bearing wear and measured 0.3mm. The factory maximum is 0.4mm. I blame this primarily on my previous owner(s) who most likely never released any tension for the first 100.000 miles. I think I was lucky…

My rear crank seal is also leaking badly, finally it is good that I do this job now. Because it would have been a big disappontmet if a new oil pan gasket wouldn’t have resolved my oil leaking engine…

borland 09-26-2005 02:21 AM

Your rebuilt is missing the guide sleeves on the rear mating surface. Make sure you transfer them before returning your old tube.

They will pound out, but they are spring loaded, so will fly upon being dislodged.

Schocki 09-26-2005 09:23 AM

Originally Posted by borland
Your rebuilt is missing the guide sleeves on the rear mating surface. Make sure you transfer them before returning your old tube.

They will pound out, but they are spring loaded, so will fly upon being dislodged.

Thanks for the heads up!

heinrich 09-26-2005 12:02 PM

Certainly a discussion as old as the mountains. And though very plausible, this is the first time I've read that particular theory Borland puts forth.

Is it correct? Possibly, possibly not. Is it important? Possibly, possibly not. What is important, is that it happens.

I believe as has been discussed over and over again that the torque tube acts as a very large elastic band. if you twist it, it shortens. If you release twist, it lengthens again. And what happens inbetween, is that the front clamp on the flex plate (and maybe rear clamp too) slips (or more precisely, the tube slips out of the clamp). When the tube lengthens again, the clamp is pushed forward.

Personally I believe the way to solve this is to design soemthing that does not clamp, but slides, or U-joints.

Am I correct? Possibly, possibly not.

Is it important that I be right or wrong? Not so much. Problem exists. Here is the more interesting question: Why does it happen almost exclusively in S4+ cars ... torque?

heinrich 09-26-2005 12:11 PM

Originally Posted by Vilhuer
.... we have two separate systems that will expand and contract when temperate of the parts change ..... When car is driven, parts will heat up and/or cool down as result of outside temperature, heat from engine cooling system, warm exhaust, etc. What effect these temperature changes have to above mentioned parts? They can be seen as two rods. One consisting of aluminium-steel-aluminiun and other steel-steel-steel-steel. Will temperature changes into these seven parts be enough to cause one set to expand or contract more than other? ....

Hmm ... and right under that flex plate and that pinch clamp sits a device that is usually glowing red hot .... the cats. Interesting thought.

Schocki 09-27-2005 12:37 AM

This is just an attempt to verify what's going on inside the transaxle.

If the shaft twists and shrinks or if vibrations are the reason, the outcome is the same. The flexplate is pushed against the flywheel

Question: If twisting causes the sahft to shrink, why is there no movement in the flexplate with NEW Torque converter bearings? The shaft is still twisting and the torque converter bearings have nothing to do with this. Why do so many people have bad vibrations in the rear end of the car and new bearings make everything smooth again? Just asking...

Crank oil seal findings: Removed the old seal and after the flywheel was off, the reason for the leak was visible. The seal was installed tilted by the factory The upper portion of the seal was pushed in further than the lower portion. This caused more wear on the lower part of the seal and finally a leak.
Pretty amazing how good the seal looks overall and how much material is gone due to the increase in pressure at the bottom of the seal.

heinrich 09-27-2005 01:47 AM

Yes this is how I understand it. Schocki if it is a fact that new bearings == zero tbf then I am happy with your explanation but is that certain?

Originally Posted by docmirror
Well, I have the answer to the 'why' questions from above. Unfortunately it's long, boring and not worth the electrons to fully explain. Let's see if I can use a simile and then apply that to the system to see where Porsche went wrong;

Think about it this way - You take a towel, get it fully wet, then lay it down and measure the length. Now, in an effort to dry the towel, twist it from both ends into a spiral by applying TORQUE to it in opposite directions. Once twisted into a spiral, set it back down and measure the length again. It's shunk by quite a bit right? Take the towel, untwist it, and lay it back down and measure it again. It should be as long, or longer than the first measurement of the wet towel.

That is called torsion modal shrink. Now, if you can think of the towel as your torque shaft, inside a torque tube, where the ends are rigidly fixed, I think you can see the problem when torque is applied to the shaft repeatedly within an undamped system.

What they could have done differently; Most engineers would use a coupling at one end to relieve not only the coupling moments, but also that slight modal shrinkage that occurs. A Lovejoy or other type of elastomeric element at either end would have absorbed this torque and allowed the shaft to shrink and grow as the torque is applied then removed. The "flexplate" performs this function to a certain degree, but apparently they did not allow for enough longitudinal compression to stop the clamps from moving.

It appears that the clamps on one end allow the shaft to shrink, and then when torque is unapplied, that shrinkage is not taken up by the clamped area, but forces the flexplate to bow in as viewed from the crankshaft. This then preloads the crankshaft, and the whole process is started all over again when more torque is applied. Do this a few thousand times, and it looks like the take up exceeds the play in the shaft, and starts eating into the bearing by too much preload.

Why autos more than stick? The answer is left as an exercise for the community. Hint; think about how an auto car is used on occasion. How much torque twist does the shaft encounter? I don't have the specs for it, but I would estimate under full load, worst case, short moment twists of probably 150-200 degress end-to-end. The air cooled VW trans input shaft experiences this, and it's much shorter. Of course, VW put splines on the shaft, and torsion springs on the clutch plate for billions of hours of good service. That, or the 40 HP just doesn't put much stress on it :)


borland 09-27-2005 02:26 AM

The shaft does not change it's length under normal use (elastic deformation). It's just not a consideration when designing rotating machinery. Here's some engineering discussion on the topic:

Schocki.... Did the flywheel come free without unbolting the clutch housing?


Schocki 09-27-2005 09:51 AM

Originally Posted by borland
The shaft does not change it's length under normal use (elastic deformation). It's just not a consideration when designing machinery. Here's some engineering discussion on the topic:

Schocki.... Did the flywheel come free without unbolting the clutch housing?


No, the upper part of the bellhousing had to be removed too. A 19 mm wrench with an extension did the job. Little tricky setup, let me know if you want to see a picture. The bolts are super tight and hard to reach.

Thanks for the link!

Bill Ball 09-27-2005 02:02 PM

As Tails argues, I've always accepted that winding up of the TT shaft during accelertion shortened it a bit and with the help of vibration, pulled the shaft back through the clamp, accumulating the 3MM of preload most people find whenever they check and release the clamp. Others have done the calculations and say the twisting and shortening just is too small to be the explanation. Also, on my 89, the exposed spline has gradually gotten less and less after each release, suggesting the shaft is moving forward or stretching. I really don't know what is going on here. Since my torque converter bearings are starting to whine, I'm headed towards a converter bearing and TT job soon. I plan to install the earlier type shaft with the shims and circlip to eliminate shaft migration back through the clamp and then just check the flexplate for flatness every so often.

All times are GMT -3. The time now is 06:56 AM.

© 2019 MH Sub I, LLC dba Internet Brands