A New Approach To Compensation?

Stephen Harris   Drawings by author

Some years ago my very first scratchbuilt loco, class 08 0-6-0 diesel shunter, had "popped" its clogs and so had been laid to rest in the railway mortuary pending decisions on it's future. Also, from a later batch-build of 3 chassis for further class 08's, I had one incomplete chassis which had never been used - after progressing the 3 mechanisms this far, I had decided to concentrate on 2 locos only, as the consumption of hours was so great as to make the completion of any of them seem an eternity away! So about 18 months ago I was keen to combine the original 08's body, perfectly salvageable with a little modification with the spare chassis into a working loco. However the spare chassis had been built as a rigid form whereas recent experience of compensation on an 0-4-0 loco had fully convinced me of its great benefit, to the extent that I would now never go back to rigid construction.

So what was I to do?

I remembered that some time ago John Greenwood had written an article describing his use of compensating beams between pairs of axles set between the main chassis frames. This approach was an absolute non-starter in my case as all available space between the frames at axle level contained either drive mechanism or lead ballast weight - no room for anything else. A fresh thought was required. I needed a system, which allowed individual wheels to rise and fall relative to the frames s variations in rail levels dictated, but with minimal modification to the chassis.

Instead of describing the thought processes I went through to resolve this, I will now jump ahead directly to the solution and explain how it works. It is important to bear in mind that, prior to setting the old grey matter to work, I carried out a simple test (see figure 1) to determine the position of the centre of gravity of one of the other complete and operational 08's. It was about 2-3mm in front of the middle axle. The fundamental significance of this piece of information should become apparent.

Sketch of balance point

Figure 1. Push Plasticard along workbench, rolling it on the drill, until the loco is at balance. The drill is then directly below the centre of gravity. The lateral position of the centre of gravity will normally be close on the centre line of the loco.


Figure 2 shows diagrammatically the final arrangement. Axles 1 and 3 have hornblock bearings, which slide, vertically in slots in the frames. Each end of both of these axles is lightly sprung downwards by a piece of phosphor bronze wire to ensure that the wheels maintain contact with the rails. The top of the frame slots for axle 1 are dimensioned to prevent this axle from rising higher than its notional level position (hereafter called datum level) relative to the bottom of the frames. It is however, free to drop (by up to 0.5mm) below datum level. I shall call this a PARTIALLY RIGID AXLE. The frame slots for axle 3 are taller and allow the axle to rise or fall (+ or - 0.5mm) from the datum level. I shall call this a FLOATING AXLE. Axle 2, the one driven by the gears has no hornblock, but instead has circular bearing buses which can be locked into place by cuttings of wire slotted though sections of small brass tube soldered to the bottom of the frames. The axle cannot move from its datum level and I shall call this a RIGID AXLE. It has to be so in order to maintain the meshing of the gearwheels.

General arrangement of fixed and floating axles
Figure 2

Photo of chassis
Photo 1 - photo by Mark Fielder

The incomplete and spare class 08 chassis, shown prior to it's conversion. The worm shaft is mounted on two ballrace bearings. A simple wire clutch arrangement allows removal of the Fleischmann motor without upsetting the worm wheel mesh.

How it works

Knowing that the centre of gravity of the complete loco lies between axles 1 and 2, axle 3 could in theory be dispensed with and the loco would balance on axles 1 and 2 alone, like an off-centre 0-4-0. I say in theory, as clearly this is not acceptable for obvious aesthetic reasons! So axle 3 is retained, still driven by the coupling rods, but it carries almost no weight at all and therefore contributes nothing to the tractive effort. Both wheels of axle 3 remain in contact with the rails by means of the phosphor bronze springs, and they therefore assist with electrical pick up. Because these wheels carry almost none of the loco's weight, axle 3 can be ignored when considering the behaviour of axles 1 and 2.

Consequent on the centre of gravity being close to axle 2, this rigid axle will support the majority of the weight of the loco, and the remainder will be supported by axle 1.

Now consider axle 1. On dead level track, both hornblocks on this axle will be at the top of their slots and sharing the load on this axle equally. But, if for example there is a dip in the rail under the left hand wheel of this axle, this wheel can drop to follow the rail head, its hornblock must drop with it and thus part contact with the top of its frame slot, and consequently this wheel carries no weight at all (save the negligibly small amount applied by the phosphor bronze wire spring). So the loco is now balancing on 3 points of support, namely the 2 wheels of axle 2 and the right-hand wheel of axle 1. Consider this on plan in figure 3 (wheels and hornblocks are denoted L1, R1 etc for ease of reference). Fundamentally, the loco can indeed balance on these 3 points because the centre of gravity of the loco lies within the shaded triangle defined by these 3 supports R1, L2, R2. Had the centre of gravity been nearer the axle 1 than to axle 2, it would lie outside the triangle, and the loco could not balance on R1, L2, and R2. Instead, it would tip over until hornblock L1 once again comes in contact with the top of its frame slot, lifting wheel R2 off the rail in the process (remember - axle 2 is the rigid axle and therefore this axle cannot move from its datum level in the frames). The loco would still operate, but wheels lifting off the rails are totally contrary to the concept of compensation!

Showing which axles carry weight
Figure 3

This process of considering how the loco behaves when it encounters imperfections in the rail level can be done for each wheel in turn and also for humps in the rail as well as dips. It is only necessary to consider one hump or dip at time - if the track contains more than one imperfection then the resultant behaviour of the loco is a logical combination of its behaviours under the individual imperfections, added together, if you see what I mean! In fact, a hump in the rail under one wheel is equivalent to s simultaneous dip in the rail under all five other wheels, so it is strictly not necessary to consider humps as well as dips. However for ease of imagining how the loco performs, I think it helps to do so. In summary, the causes (track imperfections) and effects (wheel movement/loco body movement) can be tabulated as follows:

Table A. In this table the following definitions apply:

Cause Effect on Wheels Effect on Body Supports
Dip in rail under wheel L1 L1 drops into dip and thus drops relative to frame datum

L2 R2

Dip in rail under wheel L2 L2 drops into dip. Relative to frame datum level, L1 & L2 doesn't move, L3 & R3 rise, R1 drops. Tilts ACW

Rocks CW


L2 R2

Dip in rail under wheel L3 L3 drops into dip and thus drops relative to frame datum level.

All other wheels; no effect.

L1 R1

L2 R2

Hump in rail under wheel L1 L1 rides over hump.

Relative to frame datum level R1 drops, L3 & R3 rise,

all other wheels; no effect.

Rocks CW.

No tilt.


L2 R2

Hump in rail under wheel L2 L2 rides over hump. Relative to frame datum level L1 & L3 drops, all other wheels; no effect. No rock

Tilts CW


L2 R2

Hump in rail under wheel L3  
L1 R1

L2 R2

Row A in the table reflects the situation considered above and shown in figure 3. Rows D, E and F constitute the great screwdriver-tip-on-railhead test so beloved of loco builders! Yes, you really can watch the wheels climb over the screwdriver tip one at a time whilst all others remain in contact with the rail, and the body behaviour is as predicted - not totally as the prototype would do as it would have full springing on all wheels of all axles, which is not the case in this model. That the wheels do remain in contact with the rails at all times is amply demonstrated by the improved electrical pick-up reliability compared with the rigid chassis versions.

Further experience

More recently, I embarked upon the building of two more diesel shunters of different prototypes. Both were very short but identical wheelbase 0-6-0's with jackshaft drives, but with totally differently shaped bodies. For ease of construction, I decided to make the two chassis identical with the exception of the jackshaft, which would be at opposite ends. They are shown in figures 4a and 4b.

Figure 4a

Figure 4b

Now, on the class 08 I had had the benefit of a finished and working model, albeit with a rigid chassis, from which I was able to determine quite accurately the position of the centre of gravity of the complete loco. From this information the arrangement of the compensation had been determined. But for these two new locos I clearly couldn't do this - the classic chicken and egg quandary! By studying the positions of the proposed chassis within the two different bodies and the consequent disposition of lead ballast weights which could be fitted into the body voids, I expected the centre of gravity to be just in front of the middle axle in both cases, one by more so than the other.

Spatial considerations within the frames and chassis dictated that the front wheels, axle 1 had to be the driven ones and therefore this perforce would be the rigid axle. Also, the bonnet profile of one of the bodies forced the very low position of the motor, the bottom of which was so close to axle 3 that it would not be possible for this axle to be the floating one - no upward movement of the axle from its datum level could be achieved. Axle 3 could, however, be the partially rigid one, i.e. allow downward compensation only as was axle 1 in the class 08. Axle 2 thus had to be the floating axle ad space did indeed permit this. The jackshaft in both cases would be rigid within the chassis. The jack shafts can be ignored, as they do not provide any support to the locos. To convince myself of the validity of this arrangement I constructed the "causes and effects" table similar to that for the class 08. (Not reproduced here). Most importantly, I produced the equivalent to figure 3, which is shown in figure 5, but considering the case of a dip in the rails under wheel L3 which this time was to be the partially rigid axle. The estimated position of the centre of gravity was again within the shaded triangle, so the completed locos should balance and compensate correctly.

Figure 5

One other feature worthy of note is the lack of bearing and locking wire for the driven axle, axle 1. I had provided these in the class 08 for the following reason: - (see figure 2, axle 2) in the forward direction of motion, the small pinion which drives the drivegear on axle 2 is rotating clockwise, and hence its teeth are pushing down on the driveage teeth. As the pinion is positioned very nearly horizontally from the axle 2 location, there could be a tendency for this pushing down action to force the axle 2 down towards the bottom of the frames, which would cause the loco to tilt even on level track. (Alternatively you might like to consider this as the loco trying to climb up off the side of the drive gear). If all is running freely and the loco isn't suffering wheelspin, it shouldn't occur, but nude certain conditions (e.g. and extra pulse of volts from a Pentroller to give the motor a good "kick") it does. Hence the provision of the locking wires to hold the bearings firmly in place. On the two new locos, I had very deliberately positioned the pinion which turns the drivegear directly above axle 1. This means the pinion can only impart horizontal forces on the drivegear teeth, and therefore there is no tendency for axle 1 to be forced towards the bottom of its slot. It thus need not be locked in place and this also obviates the need for separate bearings.

Vast numbers of hours of work later, I was able to unite both chassis with their respective lead-filled bodies - the moment of truth! The first one (figure 4a) was fine, the centre of gravity was indeed just in front of the middle axle when tested by the method in figure 1. And in the screwdriver-tip-on-railhead test, it behaved exactly as predicted. Result - happiness! However the centre of gravity test on the second diesel (figure 4b) revealed it to be just behind the middle axle, but a millimetre of at most two. Alarm bells rang! The amended figure 5 thus becomes as figure 6. The centre of gravity was now outside the shaded triangle, hence predicting that, in the case of dip in the rail under wheel L3 the loco would not balance and would tilt and rock until hornblock L3 reached the top of its slot, consequently lifting wheel R1 off the rail, which was unacceptable. So I tried it - it was true, R1 did indeed part company with the railhead. I needed an escape route!

Figure 6

One option would have been to remove some of the lead filling from inside the cab (rear) end of this loco. This would tend to move the centre of gravity in a forward direction, although I was dubious as to whether the removal of all the cab's lead would be sufficient to push the centre of gravity to the other side of the centre axle. Doing this would also reduce the total weight, and hence tractive effort, and as this was already the lighter of the two locos I was unwilling to do this. Further head scratching ensued, culminating in the following alternative solution, which I adopted: --

I "tightened up" the phosphor bronze wire springs which were keeping wheels L3 and R3 on the rails until, instead of imparting only a nominally small force on their respective axle ends, they were applying a larger and significant force. This force was in fact sufficient to cause the chassis to lift up its rear end when the body was removed, but the weight of the body when replaced would return the hornblocks L3 and R3 to the top of their slots. This time, when reconsidering figure 6, wheel L3 cannot be ignored as it will always be carrying a significant proportion of the total weight on axle 3, even if hornblock L3 is not at the top of its slot. The effect of this is to move the point of support to the loco from being wholly at wheel R3 (as figure 6 to a small distance along the axle 3 from R3. This distorts the shaded triangle to the shape shown in figure 7, and in doing so the triangle now encompasses the point of centre of gravity of the loco. Consequently the compensation should work as originally envisaged, and tests supported this to be the case. I do not know exactly how much along axle 3 the apex of the triangle moved, it is not important to know, but it was enough to solve the problem.

Figure 7

On the left, the class 08 chassis after conversion to compensated, as described in the text. The bare dummy outside frames has been fitted - these also have to be slotted to allow the extended axles freedom of movement. The top connection to the motor has yet to be lowered a little. On the right is 08575, one of the completed rigid chassis locos. The rather poor electrical pick-up reliability of these locos had been overcome by sprung railskids rubbing on the railhead. One is just visible in front of the front wheel.

An overview

If you now stand back a little and take a broad look at this approach to compensation, it is in reality a halfway house between fully rigid chassis construction and fully sprung. For that reason it probably isn't new at all, but seems ideal for this scale. I would summarise the pros and cons as follows:



I have used this approach on 0-6-0 locos only so far, but I see no reason why it should not be adaptable to other wheel arrangements. With due consideration given to the support (if any) provided to the loco by pony trucks or tenders, it should also be possible to extend the logic to cover larger main-line type of locos which involve coupling rods. The fundamental requirement is to provide exactly three points of firm support to the locos at all times (although not necessarily the same three points at all times) wit the centre of gravity of the loco always falling within the shaded triangle defined by these three points. (Analogy: - provided that you don't lean over and move your centre of gravity too far, if you sit on a 3 legged stool, all three legs will remain in contact with the floor no matter how uneven the floorboards! Not so a rigid 4-legged stool). This approach to conpensation is not applicable to mainline diesels with gear trains linking all the driven axles.

Construction details

Figures 8a, 8b, 8c show a typical hornblock, and two axle bearing arrangements. The hornblock is simply a rectangle of brass, the same thickness as the frames, and drilled to fit onto the axle. The slot in the frames has to be only just wide enough to allow the hornblock to slide up and down freely within it, but with almost no lateral play. Note that in both figures 8b and 8c the washers at the ends of the muffs are of larger diameter than the width of the hornblocks. The hornblocks are thus trapped between the axle muff and the back of the wheel boss (figure 8c) or between the axle muff and the phosphor bronze spring wire (figure 8b) and hence hornblock guides are not required. A simple keeper plate along the underside of the chassis is required to prevent all the wheels from falling out of the frames every time the loco is lifted off the track.

Figure 8a

Figure 8b

Figure 8c

Those of you with inquisitive minds will be asking "why hornblocks? Why not simple slots in the frames for the axles alone?" Well, I tried it and it definitely doesn't work! The horizontal action of the coupling rods tens to push and pull the axles firmly against one side of the slot or the other, this, combined with the rotation of the axle caused also by the coupling rods (that's what they are there for!) results in some axles "climbing" up the side of the slots and hence lifting wheels off the rails - the ultimate sin!


I am satisfied that the logic of the method described is correct - not only as a consequence of the "tests" but also because a structural engineer should have a good grasp on such matters! I would, however, say that the adoption of the centre axle in an 0-6-0 as the rigid axle (as per class 08) is best avoided for the following reasons: -

  1. When one of the rigid axle wheels, say L2 drops into a dip in the rail, the loco will "rock" and the vertical movement at the extreme ends of the loco (buffer heads) is several times larger than the depth of the dip. There is thus a visual "magnification" effect on the track defects. (This magnification effect still occurs on the two subsequent locos figures4a and 4b, but the adoption of one of the outer axles as the rigid one means the magnification factor is smaller).
  2. Allied to the case described above (1), the floating axle hornblocks would need to have twice the vertical travel of the maximum rail defect encountered. Thus, where I had chosen to work on a 0.5mm track level defect, axle 3 on the 08 should really have hornblock slots for + or - 1.0mm vertical travel of the axle from frame datum level - further weakening of the frames.
  3. In general, the centre of gravity of the loco is going to be somewhere close to the middle axle. If, even after your best initial estimates, the centre of gravity of the finished loco lies on the other side of the middle axle from that which you assumed, you have some serious rebuilding to do!
  4. Allied to 3 above, if the centre of gravity is very close to the middle axle, the balance of the loco will be very sensitive to tractive forces (pushing and pulling of rolling stock); these tractive forces, when running "heavy" trains, can have the same effect as temporarily moving the centre of gravity of a loco a millimetre or two in one direction or the other, and if this is enough to temporarily move the centre of gravity to the other side of the middle axle, the compensation will not work properly. The effect of these tractive forces depends on the difference in height above rail level between the couplings/buffers and the loco axles. This has not been considered up to now in this article, and in general need not be considered provided the rigid axle is kept as far away as is possible from the centre of gravity.

Figures 4a and 4b show much better arrangements which are far less susceptible to these pitfalls, and I would say it is best to choose the rigid axle to be as far away (on one side or the other) from the centre of the locos as possible. That the class 08 infringes this rule of thumb was forced upon me by its previous existence as a rigid chassis.

08669, the other non-compensated loco, the angle of the photo is just perfect to hide the railskid behind the front sandpipe. The other clue to its rigid form is the lack of hinge in the coupling rod. All class 08 bodies have been scratchbuilt.