Since both shaft and inner bearing ring are normally made out of steel, no thermal expansion problem occur. However it is of high importance that the inner ring is prevented from creeping, and wearing the shaft.
In all commercial engines the inner ballrace ring is retained on the shaft by a more or less heavy interference fit, or worse, by a sliding fit (Rossi, Super Tigre, K & B).
A tight fit, like in the Nelson, makes putting the shaft in and out a specialists job (same advice as before!) and it also calls for high clearance bearings.
Bearings with normal clearance won't accept interference fits of more than 0.001 - 0.002 mm interference on the shaft without loosing their play.
Sliding fits on highly stressed T.R. engines will cause all sorts of distortion and vibration problems, increasing friction and eventually ending up with a cold forged, oval shaft.
In quite a few cases we found remarkable, if temporary, improvement just Loctiting the shaft to the bearings!

In the FMV the construction shown in fig. 3 is used. A bush clamped between the inner rings of the bearings by tightening the prop nut will prevent both inner rings from turning. For this reason we can easily use a non-hardened shaft (material DIN 100Cr6, En 31 (G.B.), E 52100 (U.S.A.), the same as our front housing and bearings), with 0.001 - 0.002 mm interference fit on the bearings. This bush is quite common in "real" machinery and has been used for many years in Russian TR-motors. It can be adjusted to such a length that the bearings are set to give the desired axial play to the shaft.

3.3 Axial and radial play of the shaft.

Apart from turning the crankshaft can, and probably will, move (or better vibrate) in axial and radial direction.
These vibrations are directly induced by the ignitions of the engine and are strongly influenced by the play of the ballraces. They will increase the friction, heat generation and wear of the bearings.

In ball bearing technique it is normal to control the play of bearings by "preloading" them. This means, that in the case of a shaft supported by two bearings, both axial and radial play are controlled by setting the inner and outer rings at slightly different distance (See fig. 4 and fig. 5).

Our bearings when new, have an axial play of 0.06 - 0.08 mm.
We found this little too much, causing setting and overheating problems as well as fast wear of ball bearings. By shimming, we make the bush between the inner rings about 0.03 mm shorter than the distance between the outer rings, resulting in 0.03 - 0.05 mm axial play. This seems to be a good compromise and will stay constant over a large temperature range in an all steel set-up.

The reasons that no less play can be used safely, are the following:

• During running the inside of the motor (shaft) will always be warmer than the outside (front housing) simply because there's cold air outside and there's no other way for the generated heat to go.
  Therefore the inner ring is expanded and radial play (being around 0.003 mm in our case) will decrease by 0.001 mm with every 10° C rise of temperature difference.
  This effect, helped by a too well cooled front housing, committed quite some bearing ravages in our Rossi front-exhaust, the test-bed for many experiments with steel front housing. That's why in the FMV model direct cooling to the front housing is prevented by using a spinner (see fig. 16)
• Play is necessary to allow the shaft to bend during ignition. Calculations show that about 0.4° rotation at crank pin location occurs.

With an aluminium front housing controlling radial play by means of axial preloading is nearly impossible. This is because the bearing play induced by the radial expansion of the house will be much greater than the play reduction due to the lengthening of the house for a given temperature variation.

Setting a preload for running temperature will ruin the bearing when the engine is cold. In well fitted engines, like the Nelson, preloading is controlled by the interference fit of the bearings and the different expansion of the crankcase and the shaft, just giving the right radial and axial play at running temperature. (See fig.5).

3.4 Locating and sealing of the crankshaft, lubrication of the front bearing.

Because of the limited axial movement of both ends of the conrod it was essential that the middle of the crank pin was located exactly (within 0.01 mm) on the center line of the cylinder.
The measurements were carried out with the equipment drawn in fig. 6.
From fig. 6 it is clear that, by measuring the distance difference to the dummy con-rod with the dial gauge in top and bottom position, the perpendicularity between the shafts-axis and the cylinder axis can be checked with the required accuracy of about 0.05°
The axial play of the shaft can be checked with the dial gauge on the crankweb by pushing and pulling the shaft softly.

In the case of the FMV the position of the shaft is set by putting a separate part into the front housing (See fig. 3)
The thickness of the rear ring determines the exact position of the shaft.
This part also gives the sealing around the shaft.

A difficult point has always been the clearance necessary around the shaft to prevent touching. Since we want the clearance as small as possible to prevent excess fuel leaking through the front, calculations were made of the radial movement of the shaft at different places caused by the ignition of the motor.
Fig. 7 shows the approximate movements, for an 8 mm shaft and 8-19-6 mm and 7-14-3,5 mm bearings.
With a stronger shaft and bearings like the Nelson and Bugl have, these movements will be slightly less, but still of the same order of magnitude.

In any case it will be clear, that the best point of sealing the shaft is between the rear 1/3 and half the ball bearing center distance, where the movements are smallest. This is unlike most commercial RV-engines having the sealing right in front of the main bearing.

A clearance of 0.06 - 0.08 mm in diameter between the sealing ring and the bush around the shaft over a length of about 10 mm has proved to prevent any touching and gives barely enough leakage to keep the front bearing wet (1 or 2 drops per tank).
This lubrication is forced by average crankcase pressure through the small gap around the shaft.
In most other engines this guidance of lubrication doesn't exist, because after passing the sealing there's no controlled way for the fuel to reach the front bearing.

3.5 Balancing.

Since we always found that vibrations (think of a not fully tightened pan), cause losses in performance, economy and consistency, we are convinced that keeping the dynamic forces, (the cause of vibrations) down to minimum is one of the ways to make a better engine.
A great deal is known about balancing, one of the things being that it is impossible to balance a single cylinder engine completely, without bringing the reciprocating masses down to zero (except by using auxiliary rotating counterweight). One logical way of improvement is to lighten piston and con-rod as much as possible.

Is there any other reason, that aluminium pistons work so well??

The normal way to "balance" single cylinder two-stroke racing engines is to counterbalance the shaft in such a way that first order horizontal and vertical reciprocating forces are equally strong, which means that the dynamical first order forces are minimized. (See fig 8 for explanation.)

In the presently marketed 2,5 cc engines with iron pistons only 15 - 20% of the piston's weight is balanced.
In ABC or AAC engines this is up to 30 - 35%.

To get the FMV up to 50% balance with a Meehanite piston two things were necessary: firstly the piston weight should go down as much as possible, secondly a counterweight was put into the crankshaft.
the piston construction is to be described later, the hollowing and filling of the crankweb is drawn in fig. 9. The counterweight is made out of tungsten-carbide with a density of 15.1 gram/cm^3. (about 1.5 times that of lead).
A standard type cast iron piston, including pin, about 2 grams heavier than ours, would have made a counterweight out of gold or platinum necessary. Financial problems then forced us to design a super light piston. With AAC piston/liners expected to come this year, the problem of balancing will be solved more easily and with better result, with only a small counter weight in the shaft. (By the way, the FMV shaft with counterweight and bushing is as heavy as a standard Rossi RV shaft).