I have posted this as a reply elsewhere but I feel it is of sufficient importance to warrant the start of a new thread - happy reading!
There is currently much concern about the failure of the splined crankshaft on the S&S engine fitted in the Morgan Three Wheeler. I will attempt to explain the problems involved - they are many and complex. I apologise to those of you for whom the technology is a mystery but I hope that my writings will help you to understand the near impossible task of achieving a perfect repair every time.
Firstly we must understand that the output shaft on the S&S engine is designed to carry a sprocket for a belt or chain primary drive on a motorcycle. It is perfectly adequate for this application which uses sprockets and compensators of relatively light weight, causing little flywheel effect and all of the forces are within the length of the shaft with nothing overhanging. The application in the Morgan Three Wheeler fitted with the Centa coupling uses a relatively heavy outer rotor with increased flywheel effect and this rotor overhangs the shaft by almost double the length of the shaft. Add to this the fact that the wedging action of the rubber inserts in the coupling can produce powerful sideways forces and you will see that we have a much greater set of forces for the splined shaft to support than was originally intended.
Splined shafts have been in use for automotive drives for many years but recently designers have been using them in a way which causes me to be concerned. Splined shafts can be made to slide in the mating component or they can be static.
Sliding splines are always tough, hardened steel such as those used in gearboxes for the engagement of different gears. The mating parts are a relatively loose fit and slide freely against each other with no clamping required. Most gearboxes have a long life with no early failure of the splined components.
Static splines are those which have the two parts tightly fitted together and should not move in relation with each other. Problems arise when the fitting of the mating parts is not good enough and play develops. This is the design employed on the S&S Xwedge engine in our cars.
There are 10 splines cut in to the S&S shaft which means that there are 40 mating surfaces to be considered when fitting the two components together ( 2 flanks to each spline and between the splines, 10 areas of major diameter and 10 areas of minor diameter) it is extremely difficult to get all of these surfaces fitting well. Add to this the angular spacing of the splines (10 of them giving a 36 degree spacing) which also has to have a tolerance and however small this tolerance is, it will still give rise to some faces mating better than others.
Manufacturing techniques require tolerances of dimension - nothing is perfect and so, to achieve a tight fit. the female component has to be slightly smaller than the male. Manufacturers are aware that such items need to be installed by mechanics who may or may not be skilled in bench fitting and they aim for “off the shelf mating components”. The result is components which are sometimes borderline as far as fit is concerned. The S&S shaft is not fully hardened but “nitrided” - a process which gives a hard skin to a relatively soft component. A loose fitting hub will, therefore hammer the soft shaft till it becomes even looser and eventually failure occurs. It is becoming increasingly obvious that a tight fit is vital. Failures of shafts fitted with my standard alloy steel hubs seem to be running at about 4% which I find unacceptably high. I am working to reduce this figure.
If a hub goes easily on to the shaft it should be put aside and replaced with an under sized version - this will mean hand fitting of the undersized hub. Hand fitting means pulling the hub which is too tight on to the shaft as far as possible with reasonable force - removing it - examining the marks made by the shaft and carefully filing away some material from inside the hub bore. The process should be repeated until the hub fits tightly till it is 10mm from home and then pulled on the rest of the way with increased force. I would suggest torque figures of 100 ft,lbs for the initial fitting and 440 ft,lbs for the final tightening.
Such a fitting process would normally involve the use of engineers blue - a greasy blue compound applied to one face and then the mating parts put together. When the parts are taken apart some blue will be seen to have transferred to the other component and these blue areas are then filed or scraped down. The process is repeated til a uniform pattern of blue dots is seen on the mating face. Our problem is that when making a fit of parts which slide together the blue is scraped away by the assembly process and only the parts which do not touch have any blue left on them - exactly the opposite of what is needed. I use a piece of new crankshaft as a gauge with a light source behind it to get a view of the fit between the parts but this, of course, can not be achieved with the crankshaft mounted in the cases. You can conclude that the whole process is very difficult requiring much patience and skill.
The worst situation comes when there is a difference between the sizes and angular offset of the splines - this can result in, for example, only three or four splines being a tight fit. The mechanic installs the hub thinking it is a good fit but six of the splines may be loose leading to eventual failure. It is near impossible to detect such a situation. A good example of this was a mechanic who recently pointed out that a hub which had become loose was tight if rotated 180 degrees on the shaft - assembling the components in this situation would have given a tight feel to the assembly but many of the splines would have been loose and failure would have been inevitable.
I have several solutions to spline failure and I list them here in order of severity of the damage being addressed.
1/ ensure that the splines on the shaft are not severely damaged and hand fit a first undersize hub as described above.
2/ if the splines on the shaft are more severely damaged then fit a second undersize hub but filing the damaged shaft splines in order to recover the original shape as far as possible - it should be noted that in this case the nitrided hard skin will be compromised and so a good tight fit is even more important
3/ if the damage to the shaft is too much for the above to work then the shaft can be “re cut” using the spark erosion process and a custom hub made to fit - the hard skin will also be compromised in this process.
4/. If the shaft is not recoverable then it can be repaired by grafting a new shaft in to the crank web. My preferred design is a larger diameter shaft with a single keyway. This enables a full and measurable interference fit to be achieved on both diameter and key ways. I prefer to use a key which is fitted top and bottom as well as side to side.
As may be obvious the above methods are also listed in order of increasing time and expense.
Many people have expressed the belief that the reason for failure is the nut coming loose on the shaft - nothing could be further from the truth. Regardless of how tight and well secured the nut is, loose splines, allowing movement under the nut, will shake it loose in a matter of minutes which is why the nut is always loose when a failed shaft is dismantled. The purpose of the nut is to create a compound bar of the assembly. To illustrate the compound bar phenomenon imagine a dining chair - easy to push over until someone sits on it when the legs come under compressive stress and the chair is much more difficult to move. Then imagine, instead of the person sitting on the chair we drill a hole in the seat and pass a threaded bar down through a corresponding hole in the floor - fit some nuts and tighten the assembly - the tighter the nuts the more rigid the assembly becomes. In the same way the 1.125” dia shaft on our engines becomes a 2” dia shaft and has even greater strength when the nut is fully tightened bringing the shaft in to tension and the hub in to compression. This is very important when we consider the previously mentioned forces in play on the crankshafts of our cars. A tight nut will not secure loose fitting splines.
People have expressed surprise at the force required to tighten the nut properly ( I use a full nut rather than the half nut fitted as standard). The correct torque setting for a 7/8 UNF nut is 440 ft.lbs - that means 440 lbs exerted on a 1ft long bar or 110 lbs exerted on a 4ft long bar. I use a 3 ft bar requiring 146 lbs to be exerted for the same result. I weigh in a a bit under 200 lbs so 3/4 of my body weight is needed. When people accuse me of swinging on a 6 ft scaffold tube they are not quite right but I understand their concern. To gain a better understanding of the principles involved in torquing down the nut look here :-
https://www.norbar.com/Portals/0/downloads/TorqueValueGuide.pdfWe do not have figures for the tensile strength of the shaft so there is always some guess work required.
A further cause of failure has been the shaft breaking off completely. There are multiple reasons for this type of failure. Firstly the 10 splines are cut in to the shaft and at the ends they set up stresses within the shaft - such details are often referred to as “stress raisers” . I have described these stresses as being like the ribbons on a maypole where they spread out to the 10 dancers and examination of failed shafts shows exactly this shape in the broken ends. Secondly on early and unconverted machines the old Harley Davidson style Cush drive caused these breakages because it was heavy and did not fully engage with the length shaft splines. Thirdly, the splines, when cut in to the shaft, reduce the effective diameter of the shaft making it much weaker.
In conclusion I would say that the modern tendency to see multiple splines as being stronger than a single keyway is flawed due to the stresses set up in the shaft - the reduction in effective diameter and the inability to achieve a good fit. A single key in a larger shaft gives a much stronger shaft with only one stress raising point and the ability to create a good interference fit between shaft, hub and key using simple measuring techniques.
Both I and those working with me to install upgrades to your cars, will do our upmost to ensure a successful result but an occasional failure must be seen as inevitable due to the poor initial design concept. Whatever the failure and it’s cause and regardless of the seriousness of the failure, we will endeavour to make repairs possible. This is especially important at this time when replacement parts seem to be in short supply.
