But when the bearing seizes, something mechanically abnormal will happen; it's a matter of what gives first.

The control shaft (not rod - the bulletin calls another part "control rod") is not the piston. The bulletin says "the control shaft passes through an outer shaft which forms part of the tail rotor hydraulic actuator piston". So the outer shaft is the piston.

Still no explanation of how the hydraulics act on the control shaft. Sectional drawing anyone?

Fig. 4 in the bulletin appears to show sealed bearings. Seal shown in pink at the outer faces of the bearing.

Same here, according to the bulletin the control shaft passes through a hollow piston. There is a direct mechanical connection, with the castellated nuts at each end. The detached nut allowed the control shaft to move beyond its normal travel. So how do the hydraulics act on this shaft?

Fig 4 appears to show one outer race, two separate inner races. If so, the bearing is in one piece and not possible to fit with one part reversed.

Pozidrive - picture it as a MR servojack, the moving part of the jack pushes upwards and pulls downwards on the swashplate and is a piston acted on directly by hydraulic pressure - the direction of travel is determined by which side of that piston the pilot valve/servo valve directs the pressure fluid.

It is a one piece element, fescalised to reduce wear and the TR servo jack (the control rod) is exactly the same.

https://cimg8.ibsrv.net/gimg/pprune....92340b7138.jpg

Tail Rotor Servo Jack
https://cimg9.ibsrv.net/gimg/pprune....2a7c16a8c9.jpg

Main Rotor Servo Jack
https://cimg0.ibsrv.net/gimg/pprune....324d959f20.jpg

Servo valve and piston

I think if you look very carefully, you will see in that diagram that the outer races are made with a deeper track on their inside faces so that they can carry a greater axial load in one direction. The diagram does show that it is designed to carry equal load in each direction but, offset. Obviously, there are a huge variety of options that suit different applications. I believe that the "preload" term may be something of a misnomer. The manufacturing tolerances are designed so that as the inner races are clamped, and move together, the clearances of the balls in their races are positioned where they are designed to run so, this part of the process is one of correctly positioning the bearing components. Also worth noting that, unless the parts are Murphy proof, the component bearings might be assembled in wrong order(s), with their characteristics corrupted. Depending upon the method of clamping and holding the inner races, the required torque load on the retaining nut will also be defined and this will be part of the overall design.

OAP

For those who might be interested, I've compiled a 'tutorial' from another source that hopefully explains duplex bearing pre-load. The duplex bearing that is the subject of this thread is a 'back-to-back' paired bearing and it is the inner race that is pre-loaded. As Crab mentioned, pre-installation, the spider case is heated and the bearing frozen to enable an interference fit on the outer race. The inner race is then pre-loaded with the nut.

From another source- "One of the final steps in the bearing manufacturing process is the assembly of the individual bearing components: the outer ring, inner ring, balls and retainer (or ball separator). When the bearings are assembled, it is necessary to have a controlled amount of internal clearance, or looseness between the rings and balls. This is referred to as radial play in most bearing catalogs.

In certain applications, this internal clearance must be removed for a pair of bearings to operate properly. The application of an axial load across a pair of bearings – for the purpose of removing free internal clearances – is called preload.

Benefits of preloading ball bearings include:

• Rotational accuracy and precise shaft positioning
• Elimination or reduction of ball skidding
• Control and reduction of axial and radial deflection under applied load
• Noise reduction
• Load sharing between bearings

Preload is critical in most high-precision and high-speed applications, particularly where rotational and positional accuracy is required.

If, under operating conditions, a bearing has radial play this means that one bearing race can be moved radially and axially relative to the other. With rotation, this looseness translates into wobble or non-repetitive runout. This motion is unacceptable in applications such as machine tool spindles, electric motors, optical encoders, flow meters, and high-speed hand tools.

The application of axial preload forces the balls into contact with raceways, establishing a contact angle which causes the ball set to rotate in a uniform circumferential plane.

Duplex bearings are matched pairs of bearings with “built-in” preload. The inner or outer ring faces have been ground to a precise dimension known as the preload offset. This offset corresponds to the rings axial movement when a specific axial preload is applied. When the bearings are clamped together at assembly the offset faces abut, establishing a permanent, rigid preload in the bearing set.

Duplex bearings have increased radial and axial rigidity. There are three common preload configurations. DB (back-to-back) and DF (face-to-face) can handle bi-directional thrust loads. DT (tandem) can handle very heavy unidirectional thrust loads. At higher speeds, these bearings can run hotter due to the rigid preload. These bearings are commonly used in machine tool and other spindle applications, due to their low deflection rate, minimal runout and ease of assembly.

How much preload should be applied?

In general, provided the design requirements are met, the least amount of preload is desired

What are the benefits of preload?

Rotational accuracy and precise shaft positioning, elimination or reduction of ball skidding, control and reduction of axial and radial deflection under applied load, noise reduction, load sharing between bearings

Does preload have any effect on a bearings operational life?

Bearing life decreases as preload is increased

What other side-effects of preload should I know about?

With increased or excessive preload, stresses are higher and excess heat is generated "

nodrama ,PM for you

Sorry, nodrama and crab, I still don't get it.

"The control shaft passes through an outer shaft which forms part of the hydraulic actuator piston" If that description is correct there must be more than one component, and the control shaft is not the piston.

Nodrama - great post, very informative :ok: Thank you.

Pozidrive - I think you are getting too hung up on the wording in the report. My diagrams are of a 139 TR setup which is near identical.

if you replace the word 'which' with the word 'and' then it might make it clearer.

Well, I would not like to see that - this is a critical bearing (like swashplate bearing) and is quite exposed, eventhough protected with boot.
For instance: EC-135 has practically no grease lubricating requirement anywhere on airframe, as it was a goal of designer to make it low maintenance as much as possible. Swashplate bearing is still regularly grease lubricated with grease gun (comes about once a year/400FH)
Just a side note: A109Power had originally T/R duplex bearing grease lubrication schedule every 600hours or 24 months. To put grease in, the whole assembly had to be removed and disassembled to access the bearing (including removal of Tail rotor, as duplex bearing is "around" T/R driveshaft, connected with set of levers and scissors - entirelly different design than AW169)
Then in 2009 a mandatory SB came, requiring change of bearing housing to one with grease nipple and lubrication interval shortened to 100 hours / 6 months.

nodrama, many thanks for clarifying duplex bearing to everyone - nice piece of machinery engineering knowledge.

I still do not know, how exactly the control shaft and hydraulic power are interacting on T/R servo of AW169 :(

Pozidrive - I can't explain it any better than Crabs last diagram.... the one with the pressure switch in red. Also, replace 'swashplate' for 'spider' in that diagram.

nodrama-email for you

nodrama,

"Duplex bearings are matched pairs of bearings with “built-in” preload. The inner or outer ring faces have been ground to a precise dimension known as the preload offset. This offset corresponds to the rings axial movement when a specific axial preload is applied. When the bearings are clamped together at assembly the offset faces abut, establishing a permanent, rigid preload in the bearing set"

I think that this quote from your tutorial is probably the important bit. IMO, the "preload" in this case is the achievement of the design running clearance and this is mainly done by the machining process. Where your tutorial refers to preload, I believe that it means precisely this, the amount of bearing clearance (or interference) when the assembly is clamped together, achieved through the machined dimensions of the components. Yes, the assembly will come with a design clamping load or torque setting to achieve and maintain the integrity of the assembly in use.

Where your tutorial goes on to say these comments:

"How much preload should be applied?

In general, provided the design requirements are met, the least amount of preload is desired

What are the benefits of preload?

Rotational accuracy and precise shaft positioning, elimination or reduction of ball skidding, control and reduction of axial and radial deflection under applied load, noise reduction, load sharing between bearings

Does preload have any effect on a bearings operational life?

Bearing life decreases as preload is increased

What other side-effects of preload should I know about?

With increased or excessive preload, stresses are higher and excess heat is generated "

Again, these refer to the machined-in preload, not the amount that the retaining nut is tightened. Obviously, the installation / assembly instructions for the components will be specific.

OAP

Hoistop - if the bottom diagram in my last but one post doesn't explain it then I have run out of ideas to show you.

The input lever from the yaw controls moves the pilot/spool valve which allows hyd pressure to one side or other of the piston - that piston is part of the control rod.

The feedback link stops the spool valve directing fluid once the piston has moved in the appropriate direction.

It is that feedback link that came undone on the 189 meaning that the follow up action was removed and the piston did what the spool valve asked it to do - and it kept motoring until it reached full travel.

OAP - you may be correct? It is still possible to relate this sentence ......

"The inner or outer ring faces have been ground to a precise dimension known as the preload offset. This offset corresponds to the rings axial movement when a specific axial preload is applied. When the bearings are clamped together at assembly the offset faces abut, establishing a permanent, rigid preload in the bearing set." ...

..... to the image I posted at #1105 to understand how torquing the nut effects the Duplex bearing.

And maybe this sentence.....

"Duplex bearings are matched pairs of bearings with “built-in” preload"

......should finish "built-in preload offset".

I had already looked carefully and saw that the "load lines" on the diagram show equal angles once the pre-load is applied. I'm familiar with Duplex bearings to some extent (I needed to research the subject in the past before buying some for my various vehicle projects).

And now it makes sense :ugh:, many thanks nodrama for your professionalism and to crabs for the diagrams :ok:

Just trying to clarify an earlier post which said the control shaft was the hydraulic piston. Also to understand why manual control isn't possible - in spite of a direct link from lever to spider via the control shaft.

Pozidrive - The input lever is attached to the end of the control shaft (with the nut that did come undone) - then the shaft runs through the TR servo assembly with the piston section as an integral part of the shaft - the shaft then runs out the other side, through the TR drive tube, through the duplex bearing and is attached to the spider with the nut that didn't come undone.

A TR control input moves the input lever, but only enough to displace the pilot/spool valve in the servo - this directs the hyd fluid to move the piston (which is part of the control shaft) which moves the spider and changes the TR pitch.

The feedback/followup movement of the control shaft recentralises the spool/pilot valve in the servo, fluid stops being directed and the TR pitch stops changing.

The whole system is designed to have hydraulic assistance and that allows the autopilot to create inputs to the spool/pilot valve without the TR pedals needing to be moved.

I don't think the input lever has enough mechanical advantage/leverage to move the control shaft in a manual mode, even if you and the control runs were strong enough to move the pedals.

Thanks, crab - I understand well, how follow-up hydraulic servo control mechanism works, believe me.
I am (still) wondering if piston is trully part of control shaft, or control shaft passes thru the hydraulic cylinder and is clamped/attached... to (hollow)piston somehow - in other words, did rotation of (undone) control shaft strain seals of hydraulic cylinder in a way they are not designed to or not? It is an important question.
If answer is yes, it means that in case this failure occured in cruise and pilot managed to somehow maintain (some) control, he might be soon faced with massive hydraulic leak - loss of hyd. system as a consequence. If this servoactuator is powered by both hyd systems, (two pistons one on top of other and dual spool valves as in standard dual servoactuators) both systems would probably soon leak hyd. fluid, rendering helicopter uncontrollable.

Hoistop - as I mentioned earlier, look at a MR servo jack - the piston is part of the control rod that moves the swash plate - why would you make it more complex by having to clamp a piston to a control rod when you can just machine it in one piece?

If there had been evidence of a hydraulic leak, it would have been mentioned in the initial report.

Hoistop - As I've said before, I've never seen the inside workings of one of these servo actuators.
The diagram is a schematic of what I think the TR actuator looks like judging by the shape of the one fitted to an AW169. I've also provided an AW169 'technical' description that I have found. I am assuming that 'connected by a flange' is describing how the cylinders are connected together.

https://cimg3.ibsrv.net/gimg/pprune....dabbc1eda4.jpg
"A Linear Tandem Hydraulic Actuator, consisting of a single jack with two pistons housed in two different cylinders mutually connected by means of a flange; the two cylinders receive the control flows from the two control valves and provide the force necessary to drive the pitch control mechanism of the tail rotor blades."

After all the excellent information about duplex bearings and preload offsets , we are still left to find a reason for the bearing failure and the excess torque on the spider end nut.

Unless the split pin at the spider end had come adrift, the control rod can't have rotated relative to the nut to tighten it - is the initial report wrong when it says that the torque is a result of that rotation and if so, how?

Unless there is a clear explanation of that process you come to the conclusion that the nut had been overtightened either on assembly or during servicing. Whether that excess torque directly affected the operation of the duplex bearing or just exacerbated an existing problem with it is one of the questions the AAIB will hopefully answer.

"Elimination or reduction of ball skidding"


These people should be producing underpants.

Surely the bearing failed first?, so the inner race turned on the shaft, tightening the nut and shearing the pin? A bit of extra torque on the nut would not make any difference to the bearing clearance unless it deformed the inner races, which is unlikely. As above, the clearance is set by the machining process, the nut only has to apply enough pressure to hold the inner races firmly together.

I don't know the inner workings of exactly how the inner shaft is coupled to (and therefore pushed/pulled by) the hydraulic actuator and whether rotating the inner shaft could cause a leak, however the Hydraulic system #1 has an automatic TR shut off valve (TRSOV in the diagram). If pressure drops at the TRA pressure switch, or the system level drops to a pre-determied level the TRSOV closes to maintain the remaining system #1 hydraulic fluid within the main rotor actuators to maintain main rotor control.

I.e. a catastrophic loss of the TR hydraulic servo would not in itself lead to a loss of main rotor hydraulics. Such a leak would however lead to a loss of system #2 pressure as there is no TRSOV on that system, so you would lose MR hydraulic redundancy.

https://cimg7.ibsrv.net/gimg/pprune....abafe42b6f.png

It was the pin and nut on the other end - not the spider/duplex bearing end - that failed and the excessive Tq was on the spider end nut which didn't fail and the split pin was intact. How did the excess Tq happen?

Aucky - just like the 139.

Crab,

The excess torque could be for a number of reasons and it is only the effort required to remove the nut.

It is easily conceivable that the clamp up was correct.

At first glance in the picture in the report, the TR end of the control shaft there appears to be burnt grease that has migrated along the shaft.

It woudnt surprise me if it migrated to the nut end as well. As you can see it can get in to very small places as there are traces between the outer part of the inner race and the shaft.

The thread seems to be full of it as well. You dont need much friction in the thread to change the torque requirement in the nut. Ever used Loctite? It doesn't have much inherent strength but spread over the area of the thread?

Nodrama,did you get my email??

Yes. I replied by email yesterday?

I'm not really one for speculating, so the picture below is inferring nothing (except that I drew on it with my foot!). The diagram of the spider assembly to the pitch change rod (shaft) in the report isn't entirely accurate. It doesn't show the metal spacer that is between the nut and the duplex bearing. My picture shows that- nut, washer, spacer, duplex bearing. Note the step, where the spacer couldn't move any further despite what torque was applied to the nut. Maybe this is purposely designed to prevent crushing of the inner race?

https://cimg5.ibsrv.net/gimg/pprune....2f33802d48.png

RVDT - I see what you are saying.

Agree. Already covered in my post #1096

OAP

nodrama,
The step is probably not designed as a stop, it is far too narrow. If you look, there also is a similar step undercut at the inboard end of the bearing seating. This is probably designed so that the inner races very slightly overhang the seating at each end and so, there is no chance of the inner race being wedged up on a radius at the inner end and at the outer end the spacer can also fit against the whole outer end face of the inner race. These bearing races are very hard but quite brittle. Any unequal or point loading can cause fracture.

OAP

Thanks OAP. That makes sense to me.

So what caused the failure?

Since these duplex bearings are widely used in TR assemblies, their failure rate should be very low unless they are not installed properly.