Background on unit:

 

Drive had very little testing on installation (ground only).  Builder reported some oil leaking from rear seal and requested a new seal which was subsequently sent to and installed by builder.   Engine experienced various problems during testing (coolant found in rotor chambers) and was rebuilt. The builder was not able to identify the source of the stray coolant.  Builder decided to examine drive while it was off during engine rebuild and noticed that the input shaft thrust bearing had failed (roller cage was melted and all rollers were loose in drive).  Unit was returned to RWS for analysis and repair.

 

In addition to the thrust bearing, the following problems were found during inspection:

 

1.  Input shaft gear drive pin had sheared off and the gear had spun on the shaft causing the thrust bearing oil passage to clog.

2.  Sun gear pilot bearing had seized on the prop shaft pilot and stayed there when the drive was disassembled.

3.   The rear input shaft thrust bearing race had spin welded and seized on the shaft with evidence of extreme heat in that area (blued metal).  I noted that the race was more or less in its normal position on the shaft and had not seized due to moving forward onto a larger diameter portion of the shaft as would be expected when the rollers and cage vacated their normal location.  This indicates that the rear bearing race seized prior to the destruction of the roller bearing cage.

4.  The seized race had spun against the adapter plate causing the aluminum to heat and extrude into the oil seal cavity.  This allowed the input shaft to move further back against the adapter plate.

5.   The oil seal was completely destroyed by the heat of the spinning race.

6.  The moly impregnated nylon snubber was crushed on one side with evidence of heating due to friction. 

7.  The loose thrust bearing rollers had bounced around inside the planet carrier housing leaving the housing and adapter plate with the appearance of being shot peened.   Surprisingly, it appears that none of the rollers got past the planet carrier bulkhead and into the gears.  There was no evidence of damage to any of the gear teeth.  

 

There are fairly obvious explanations for each of the individual failures observed but the likelihood of all of them being present and manifesting themselves at the same time is very small.  The analysis of the root cause reduces the problem to finding a single factor which would fit all of the observations and failures.

 

The key piece of evidence in this failure turns out to be the pinched nylon snubber washer.  It would be possible for an error in setting the input shaft end play at time of manufacture to cause the snubber to be crushed when assembled but this would result in a symmetrical crush around the entire circumference of the snubber.  Operating the drive in this condition would cause the snubber to quickly overheat and melt until the interference was eliminated.  It is probable that the drive would operate normally at this point except for the harsh metal to metal contact between sungear and ring gear bulkhead under negative (abnormal) torque conditions such as engine backfire.  Insufficient endplay alone would not explain any of the other failure points.

 

The only possible explanation I could come up with that would explain the asymmetrical force sufficient to deform the snubber (it takes a force measured in tons to do this) is for the rear race of the input shaft thrust bearing to drop off the pilot on the shaft and get trapped between the adapter plate and input shaft.

 

  This can happen for a number of reasons.  Failing to push the input shaft all the way through the damper spline and into the crankshaft pilot is the most obvious.  Too much grease in the crankshaft pilot hole can also prevent the shaft and bearing assembly from seating firmly against the adapter plate.  It is a close tolerance fit between input shaft pilot and the crankshaft pilot hole and the grease will not easily extrude past the clearance.   Even if none of this has occurred and the shaft assembly is in fact seated against the plate when the assembly is performed, the close fit of the pilot and the grease will compress air in the cavity causing it to push the shaft back out if the builder is not observant and does not hold the shaft in long enough for the air pressure to bleed off.   Whatever the cause, if the shaft and bearing is not against the adapter plate, the rear race can fall off the shaft and get stuck between the shaft and adapter plate.

 

In the event that this happens and assembly continues, there is one more chance for the builder to notice that something is wrong.  When the main gear housing is slid onto the 12 mounting bolts, it will not quite seat all the way onto the planet carrier housing.   This might be hidden by the recess in the planet carrier housing if not checked closely.  The 12 mounting bolts will however, have plenty of exposed thread to fasten the nuts to.   When they are tightened, they will exert a force of many tons on the pinched bearing race.  This in turn causes the input shaft to be pushed forward into the snubber seated against the ring gear bulkhead.   Making matters worse, since the pinched race is only on one side of the shaft, it is cocked sideways causing the sungear pilot bearing to contact the prop shaft pilot at an angle.   The stage is now set for a tragic set of events when the engine is started. 

 

The exact sequence of events is in question but in order of most likely occurrence, the following things happen.  The snubber is caught between two surfaces rotating in opposite directions and is abraded in the contact patch area.  This relieves some of the pressure but it is not enough to make up for the extra 4 mm thickness of the pinched race.   The engine fires and rpms increase rapidly.  The badly cocked sun gear pilot bearing seizes on the prop shaft momentarily causing the drive to lock-up.  The firing chambers of the engine and inertia then shears the sun gear drive pin which serves the same purpose as the propeller drive pin in an outboard engine,   it saves the drive train and engine from catastrophic failure in the event of a prop strike.  At some point, the Loctite sleeve retainer holding the pilot bearing in the sun gear looses its grip and the gear starts spinning on the OD of the bearing.   The input shaft is also now free to spin inside the sun gear causing the oiling passage to be clogged with metal and Loctite debris, shutting off oil supply to the thrust bearing.

 

The engine is now free to unleash its power on the hapless thrust bearing with its pinched race behind the input shaft.   This assembly thrashes about until friction causes the race to stick in various places which causes the input shaft to machine itself into a crude tapered pilot and at some point causes the race to self-center on the spinning input shaft. The compressive forces on the assembly are finally relieved.  Things appear to work normally for awhile.  However,  the thrust bearing pilot on the input shaft has been badly mangled at this point and is no longer a precision fit inside the rear race.   In addition, it has been deprived of the forced lubrication because of the blocked oil passage in the input shaft.  It soon friction-welds itself to the input shaft and begins spinning against the adapter plate where it generates large quantities of heat.  The first thing to feel the effects of this heat is the oil seal which is pressed into the adapter plate directly behind the race.  It is soon destroyed but the welded race probably keeps much oil from escaping, especially since not much oil is being sent in its direction anyway.

 

  The very last thing to go is the thrust bearing roller cage.  It is made of a high temperature polysulfide material but it finally succumbs when enough heat is transferred from the race spinning against the adapter plate.   The rollers are flung out of the cage but at this point nothing changes other than the peening of the surrounding housing from the flying rollers.  The rear race is firmly welded to the shaft and is not depending on the rollers to support it.   The final event would have been the eventual bouncing of one or more rollers into the gear train but it was spared this due to the fortuitous timing of the builder's inspection.

 

End of report.

Tracy  (will take about as long to repair drive as this took to write )