Zen and the Art of Gear Reduction:  Part III

Thrust Bearing Design Change


As hard to believe as it may be, even we make mistakes.  The input shaft thrust ball bearing shown in part II of this series turned out to be less than optimum.  I removed the redrive from my RV-4 in order to install the RD-1(A) and was shocked to see that the bearing was on the verge of complete failure.  A maximum-effort  investigation into the cause of the failure was started immediately.


On the left is the ball cage from the drive with a new one on the right for comparison.  The races (not shown) were equally bad.  Total time on the bearing was about 100 hours.

When I first selected this bearing to replace the needle bearing we started with, I had a gut feeling that something was not right even before I installed it. The bearing was made in foreign country (which will remain unspecified so as not to offend).  Anything I have purchased, that was produced in that particular country, turned out to be of poor quality.   I over-ruled this gut  feeling because the book specs on the bearing looked good and when I calculated the bearing life it came out to around 2000 hours.  

My gut reaction returned with a vengeance and plans were made to install a German made bearing and re-test but before I had a chance to do this, the investigation took an unexpected ugly turn.  The calculations on bearing life had already been triple checked  but going over them a fourth time uncovered a fatal flaw.  It turns out that a different exponent is called for in the formula when calculating bearing life for a thrust ball bearing as opposed to needle or roller bearings.   Failing to notice this, I had used the same one as for the needle bearing.  This difference is reflected in a specification called "Fatigue Load Rating" which I was formerly unfamiliar with.  When I looked up this rating it was obvious that the ball bearing used was doomed to failure.  At this point, I felt that my tiny company called Real World Solutions, inc. was headed for a similar fate.  While thinking of the gear drives that had already been shipped I had one of those rare moments when I was sorry I had ever started it.

Fortunately there is a happy ending to this story.  A roller bearing was found that would fit in the same outline as the ball bearing and would require only minor machining of the input shaft assembly to install.  Even though the numbers on load and life look very good (even when using the correct formula!), I have learned not to ignore those "Gut feelings" I mentioned before.  In spite of the higher cost, I specified a German made brand this time.  This is the bearing I installed in the prototype RD-1(A) and flew to Oshkosh with.  After a harrowing return flight (due to weather, not mechanical problems) I came down with a killer cold so the post flight tear-down and inspection of the re-drive has not been done yet but I do not expect any surprises.   I'll post the results here when they are done.

Sending out a notice concerning this change to all previous buyers of the RD-1 was the most painful business decision I've had to make.  The expense did not bother me much, but having to admit that I had royally screwed up is never any fun.   Fortunately, none of the shipped units had been flown yet.  The retrofit of the bearing was done at no charge. 

Here is the latest (and hopefully last!) input shaft thrust bearing.  The rollers are in a non-metallic polyamide cage which allows a higher RPM rating

Stay tuned for more.........

Hosted by www.Geocities.ws