Saturday 26 June 2021

The Lubrication System

Motorcycle lubrication systems underwent a major transformation in the 1920s from total loss, with either a hand or mechanical pump (or both!), to recirculating dry-sump systems in which oil was contained in an oil tank and pumped into and out of the engine.

As far as I am aware the OHC AJSs of 1928 – the 350cc K7 and 500cc K10 were the first production motorcycles from AJS that had dry sump lubrication and this was a huge technical step compared to the previous total loss systems. Admittedly, Velocette had a recirculating system in production from 1925 in their OHC Model K.  The most significant difference with a dry sump system is that oil is delivered strategically to critical parts of the engine rather than just having fresh oil dribbled into the engine and hoping that it reaches the right places.

One aspect of the relatively (by previous standards) sophisticated lubrication system on early AJSs is that there are pretty well no internal passages for moving the oil from one part of the engine to another and this results in a lot of external oil pipes around the engine. My recreation is no exception and this does endear it with the appearance of a Victorian steam engine.

In this V-Twin engine I have incorporated what I believe to be the better features of the original AJS engine whilst learning from the previous experience of the AJcette single cylinder engine, particularly with regard to cambox lubrication. One point to bear in mind is that the reciprocating oil pump, of AJS manufacture, fitted to the original OHC AJSs and which looks like this:

(apologies for the blurry picture)

has a very low flow rate and delivers oil at a much lower pressure than the Velocette gear pump that I have fitted.

As discussed in a previous blog this engine has been built with 3 oil pumps, one to deliver oil to the engine and return it to the tank and 2 additional pumps to scavenge the cam boxes. It is interesting that the first incarnation of the original AJS V-Twin used the AJS reciprocating pump (as above and the same as the single cylinder engines) and there were no cambox scavenge pumps but this evolved into the same configuration that I am using here with 3 gear pumps, seen in the picture of the bike as it is today (in supercharged form) in the National Motorcycle Museum.

 The oil circuit that I have adopted is as follows:

1)  Oil is fed by gravity from the oil tank to the upstream side of the oil pump. A brass petrol tap is incorporated into the oil line to avoid wet sumping. The tap has been bored to ¼” ID to reduce the flow restriction. The main oil pump, which contains 2 pumps - the feed from the tank into the engine and the return from the crankcase to the tank, is driven at half engine speed from the camshaft drive and has inlets/outlets as illustrated below:


A suction filter has been incorporated into the oil return pipe from the crankcase – more later.

 2)  Oil is fed from the downstream side of the feed pump to the left side of the engine crankcase where it is then distributed to other parts of the engine.

A pressure gauge has been incorporated so that some quantitative measure is available of the oil pressure being delivered to the camboxes and this is controlled by a valve that diverts excess oil to the centre of the crankcases, immediately above the connecting rods. There is also an oil drilling in the casting below the main connections that delivers oil to an annular phosphor bronze distributor ring sandwiched between the drive-side main bearings, shown below

and then via a drilling in the mainshaft to the big end. This is the only internal oil passage in the entire engine.

Incidentally, the copper tubes need repeated annealing to allow bending and I always make a pattern using galvanised fencing wire before bending each copper pipe.

 All connections are silver soldered – there are no soft soldered joints.

3)  The oil pipe to the camboxes is split at a T-junction and the flow rate is then controlled by forcing the oil through an orifice contained in a brass housing. There are 2 of these, one for each cambox, as shown below in assembled and disassembled forms.

Inside this fairly compact oil restrictor is an orifice with a 0.6mm diameter hole and 2 filters, a 100 mesh filter immediately upstream of the orifice and then a coarser filter upstream of that. This arrangement worked well on the AJcette engine but will almost certainly require some calibration when the V-Twin is eventually fired up.

The restrictors are located on the sides of feed pipes that are screwed directly into each cam chamber. On the top of each feed is a drilling, blanked off with a small knurled screw, to enable the cam chamber to be primed with oil if the engine has not been started for some time, for example, over the winter months.


 4)  Oil enters each cam chamber from the downstream side of the restrictor, lubricates the cam and rocker skid and can then “escape” via 2 possible routes. The first is via the large bearing at the drive end of the camshaft and then into the top of the timing case; the second is via the 2 holes in the casting through which the rockers emerge immediately above the cams. The oil can then find its way into either the primary or secondary gutters on both inlet and exhaust sides or it escapes from where the rockers exit the cambox and then covers the cylinder head and everything else in oil!

 Oil from the primary gutters is collected at the end of the camshaft and fed into one outlet whilst oil from the secondary gutter on the exhaust side is fed into another outlet. Oil from the secondary gutter on the inlet side is fed via a 2.5mm diameter pipe directly into the inlet valve guide where (maybe?) it lubricates the inlet valve and guide. The picture below shows the arrangement on the front cylinder.


The oil drains from the gutters on each cambox are combined using a Y-junction (see above picture) and then fed into the upstream side of the cambox scavenge pump at the top of each timing case. An oil pipe connected to the downstream side of each scavenge pump deposits the oil into the right (slack chain) side of each timing case from where it can dribble down and lubricate the chain and gears.

Oil is transferred, in either direction, between the crankcase and the timing case through the timing side main bearing.

 7)   Eventually all the oil finds its way to the bottom of the crankcases (at least, the oil that hasn’t leaked out or escaped by some other means) and is collected through a pipe connected to a small oil collection chamber. The picture below shows this just after the crankcases were machined.

It is beneficial to include some kind of filter before the oil is sucked into the scavenge pump and returned to the tank to try and prevent any debris causing damage to the gears in the pump and, indeed, to prevent any debris from circulating further in the engine. There was not sufficient space to incorporate an adequately sized filter into the crankcase itself and an external filter was therefore made, housed in a cylindrical brass casing, to have an in-line filter between the crankcase and the oil pump.

The picture below shows this plumbed into the oil return line and contained in an aluminium housing. 


This will be secured to the bottom of the engine plate using 2x 3/16 studs the next time the engine is removed from the frame.

As I mentioned at the beginning, the combination of a fairly sophisticated lubrication system and a lack of internal oil passages does result in a lot of oil pipes around the engine.


Monday 14 June 2021

Footrests and Rear Brake

The footrests and rear brake arrangement on the donor chassis, the 1931 AJS SB8, were of no use for the V-Twin for 2 reasons. Firstly, the footrests were fixed to the engine plates between the engine and gearbox and there is absolutely no space for such an arrangement here and, secondly, the Royal Enfield shock-absorber hub that I am using in the rear wheel has the brake on the right side and this requires a cross-over shaft for the footbrake pedal for the pedal to be located in the usual place for a vintage British bike on the left side.

So, the question arises as to where to mount the footrests and how to arrange the brake….?

The solution that I adopted was to mount 6mm thick steel plates on both sides of the frame and attached to two of the three 7/16’’ diameter high tensile studs that fix the front and rear parts of the frame together. I was also able to support the plates at the rear using a conveniently positioned frame lug that was present on both sides and in exactly the right place to give the plates a 3-point fixing.

I had, for one of my previous Norton Model 18 projects, purchased a pair of original WD (War Department) Norton footrest hangers on ebay from a guy located in Australia. If you want to get any then the company name is Trojan Classics ….but be quick because there probably aren’t too many left. Norton used longer footrest hangers on their vintage bikes but the WD Norton hangers are shorter.

I decided to use these here because they can be conveniently rotated in 30 degree intervals to optimise the position. They arrived, still wrapped in the grease and paper from when they left the factory, and probably haven’t seen the light of day since 1943. It is a mystery to me how bucket-loads of WD Norton footrests ended up in Australia but, as of the week of writing this blog (15th June 2021) the UK and Australia have just signed a free-trade agreement, so maybe we’ll be able to make these a core export in the future….

and after a bit of degreasing, they emerged as good as the day they were made.

The hangers need fixing to a similar 30 degree indexed mounting and these were made in the milling machine.

It is interesting that such a complicated setup is needed for such a seemingly simple component.

Two of these footrest hanger attachments were made with a shoulder to positively locate them into the 6mm steel plates, which had been cut to shape by making a cardboard template, and were then brazed to the plate.

The next step was to make the crossover for the brake. A consequence of having the brake pedal on the left side whilst the brake itself is contained in the hub on the right side is that the brake rod connecting the brake arm on the brake pedal shaft to the brake operating lever on the brake plate would need to be positioned on the top of the brake arm to have the rod in tension (pulling). The picture below illustrates this on my 1928 Model 18 Norton.

There was not sufficient space to use this arrangement for the V-Twin and the lever needed to be positioned below the spindle. A result of this would be that the brake rod would operate in compression (pushing) rather than in tension if the brake arm was fixed onto the end of the brake pedal shaft. It is not really a good design to have a control rod such as this operating in compression as it would be subject to buckling loads and there is a good chance that it would bend in the middle if any substantial pressure was applied to the brake. There are 2 possible ways to solve this: either use a thick brake rod (typically 3/8”) if it is to be in compression or reverse the direction of “pull” and use a ¼” rod in tension.

I chose the latter route and this necessitated using an additional shaft and a pair of gears to change the direction of rotation. The picture below shows the collection of bits necessary to achieve this.

Here, there is a brake pedal mounted onto a 11/16” diameter shaft as a tight interference fit and also held on with a ¼” high tensile bolt and is silver soldered. The other end of the shaft has a ½” square machined onto the end onto which is fitted (a light interference fit so it can be assembled) a 20 tooth gear. There is another gear mounted onto an adjacent shaft (this was subsequently silver soldered to the shaft) which fits into a short housing containing a grease nipple. The gears are off-the-shelf items and come with a 1/2'' hole; this needed spark-eroding into a 1/2'' square. On the end of this shaft are 12 splines onto which is mounted a brake arm (ebay find from a BSA C15).

The picture below shows the assembled collection of bits. The brake return spring is yet to be added.

One issue that is yet to be addressed is the footrest on the right side. I would prefer to have a kickstart on the bike rather than having to run and jump on ….I’m getting a bit old for that, but with a footrest in a fixed position there is no chance of having a kickstart.

I found a pair of folding pillion footrests and modified one of these to screw into the 7/16’’ BSCY thread on the Norton footrest hanger.

The picture below shows the modified footrest (upper) together with the unmodified one (lower)

The complete collection of bits is shown below. There is now the addition of a brake return spring (Triumph T100) and the spindle support housing has 16 holes to enable the spring to be positioned optimally.

The plates have been trimmed to eliminate unnecessary material.

The assembly, ready to fit on the bike, is shown below.

With the addition of a brake rod and a knurled adjuster knob the complete kit is now mounted on the bike and works as intended

….but it turned out to be quite a lot of work for a couple of footrests and a rear brake.


Tuesday 1 June 2021

The AJS V-Twin Crankshaft: Part 6 Assembly and Assessment

Before assembling the crankshaft there were a couple of final machining operations.

The first was to machine the final drive sprocket and, in particular, to ensure that the 40 taper positioned the sprocket in the correct place on the mainshaft. I had previously set up a dummy shaft through the main bearings to determine the axial position of the sprocket to ensure that there was clearance between the chain and the crankcase casting and to position the clutch chainwheel.

The taper was then machined on the actual mainshaft to position the sprocket correctly.

It is much easier to measure and check the machining of the sprocket taper in the lathe with the mainshaft alone rather than when inserted into the flywheel. The change in axial position for a change in the radial cut is determined by 1/(tan(taper angle)) which is 14:1  -  in other words, a 0.001” increase in radial cut will move the sprocket 0.014” along the shaft and so care needs to be taken to avoid taking off too much material and positioning the sprocket too far inboard.

The collection of bits and pieces that is now just about ready for assembly is shown below.

Before starting to assemble the various parts, there was one final check to be made and this was to carefully measure the 3 main dimensions that would determine the crankshaft end-float; these are shown in the sketch below.

Here, A and C are the widths of each of the flywheels between the mainshaft boss and the big-end boss and B is the width of the crankpin between its contact surfaces with the big-end boss. Clearly, A and C should be the same and the sum A + B + C should be equal to the distance between the main bearings minus the desired end float. A few thou were removed from each mainshaft boss to satisfy A = C and to simultaneously provide 0.010” end float.

The mainshafts were then pressed into the flywheels.

The pressure on the hydraulic press was monitored during pressing and rose progressively to a value of around 12 tons - very close to the theoretical calculated value. The mainshaft oil holes were lined up carefully prior to pressing and checked immediately to ensure that there was a flow connection between the oil hole and the big-end drilling. A small amount of lubricant (Loctite bearing fit) was applied prior to pressing. I believe that a certain amount of broaching occurred when pressing in the drive-side mainshaft and that this is the result of the high interference fit; more will be said of this later.

The next step was to press the crankpin into the drive-side flywheel, again ensuring that the oil holes lined up. Again, the maximum pressure required was very similar, around 7 tons, to that calculated.

The connecting rods and bearings were then assembled onto the crankpin

making sure that the connecting rods were the right way round (!!) and the pin and timing-side flywheel were pressed together.

To help with alignment, a silver steel (drill rod in the US) ground 12mm rod was inserted through both flywheels prior to pressing.

 The crankshaft was then immediately set up in the crankshaft measurement jig

to check the runout.

I couldn’t have wished for a better result: both the drive side and timing side mainshafts showed a deviation of maximum 0.001” adjacent to the flywheels

and no greater than 0.002” at the far end of the drive-side mainshaft.

I was extremely happy with this result.

The dodecagonal big-end nuts were then tightened to 150 ft-lbs torque and the runout was rechecked to ensure there was no change.

Some final measurements were made:

Total width between mainshaft bosses = 4.118” – 4.120”

The measured distance between the main bearings is 4.128” and this gives the desired 0.008” - 0.010” end-float.

Total mass = 26 lbs

Apparently the crankshaft in the original 996cc AJS V-Twin weighed 42lbs according to this article so maybe my crankshaft can be considered a mere lightweight…In fairness, the OHC R10 engine, on which the original V-Twin was based, had a stroke of 101 mm and this would have invariably resulted in larger diameter flywheels.

The actual balance factor can be calculated by suspending a mass from the small end of one of the connecting rods so that the rotating crankshaft is in perfect balance, that is, it will rotate without stopping in any prefered position.

To achieve the design balance factor of 59.2% the mass to be suspended from the connecting rod is calculated as follows:

Total Reciprocating Mass = (Mass of Con-Rod Small Ends) + (Mass of Pistons)

= 418 + (373.2 x 2) = 1164.4g

Mass to be added for perfect balance should be

= Reciprocating Mass x Balance Factor – Mass of Connecting Rods Small Ends

= 1164.4 x 0.592 – 418

= 271g

However, the actual mass that was added to achieve perfect balance = 338g

It would therefore seem that an additional 67g needed to be added compared to that required for a 59.2% balance factor.

This means that the flywheel has been “underbalanced” compared to the design value.

The actual balance factor can easily be calculated to be 56%, which is 3% less than the target value.

Does this matter? Not really. From the discussion in a previous blog, V-Twin balance factors can be anywhere in the range 50 – 60% and so 56% should be quite OK. Why is the value different from that calculated? I have added that to one of the points under Lessons Learned.

Lessons Learned

Designing and making the crankshaft has probably been the largest and most challenging sub-project in the overall V-Twin project and has taken many weeks in design and in the workshop. There were no diasters during the manufacture and no aspect really went badly and I have ended up with a crankshaft that is strong and fulfills the accuracy and balance requirements.

Nevertheless, it is useful to reflect on what has been learned – what went well and things that I would change if were making another crankshaft. I believe it is the Spanish philosopher George Santayana that is credited with the aphorism “Those who cannot remember the past are condemned to repeat it,” …variants of which are also attributed to Edmund Burke and Winston Churchill.

Things that went well…

1)  The use of the Harley Davidson EVO connecting rod and big-end assembly was a good choice. This was bought for a remarkably good price, it is of excellent quality and fits the bill perfectly.

 2)  Design of the flywheels, in particular the balance calculations, worked out well and with a good engineering solution. If I was starting this project again I would automate this process either using a spreadsheet or, better still, use CAD rather than making laborious repetitive calculations as quite a few design iterations were necessary.

 3)  The introduction of a 12mm hole in both flywheels diametrically opposite the crankpin proved invaluable in the accuracy to which the second flywheel and crankpin could be pressed together using a piece of ground silver steel as a guide dowel. No changes or adjustments were needed to the alignment after pressing.

 4)  The use of electroless nickel plating on the drive-side mainshaft worked well. It is straightforward to use, the deposition rate can be checked easily with a test piece and is sufficiently slow that a known thickness can be deposited in a given time period and it provides a thin hard protective layer.

 5)  In the absence of accurate internal grinding, the use of parallel reamers to finish the mainshaft and big-end holes worked well. The only disadvantage is the cost; the 2 reamers that I used were purchased especially for this project and cost 230 GBP for the pair; quite a lot of money to make a couple of holes.  

 Things that I would change….

1)   The interference fit of 0.003’’ – 0.0035’’ for the mainshafts in the flywheels is too great and, as mentioned previously, has probably resulted in some broaching on the drive-side. I would reduce the fit to 0.002’’ -  0.0025” in future.

 2) The initial decision to use EN40B and nitriding as a hardening process was taken on the basis that the shafts could be machined-to-size prior to a relatively low temperature heat treatment process. Unfortunately this did not account for the white layer formed during the nitriding process, which I became aware of later, and which would need to have been removed with a consequential change to critical dimensions. If I were making these shafts again, I would use EN24T as it has similar mechanical properties and is readily available and to apply electroless nickel plating.

 3) The balance factor turned out to be 3% lower than the design value. Having rechecked my calculations I am almost certain that the discrepancy stems from not accounting accurately for the radii in the annular segment – shown below

This reinforces the desirability of automating the balance calculations where all details are accounted for properly.  

In summary, I am pleased with the crankshaft and, although there is one final check to make, namely the piston height relative to the top of the cylinder, I am confident that it will perform well in the engine.

....and finally.... In one of my previous blogs on the crankshaft I included a picture of the "as received" crankpin with a taper for entry into the flywheels and a reground crankpin with a parallel fit. As they were both in the same picture the curious might have asked "why do you have 2 crankpins?". Well, when I was checking the rod assembly prior to pressing in the crankpin I found that one roller from the bearing on the forked rod was missing. I searched everywhere for the roller but couldn't find it ....the workshop fairies had hidden it somewhere.

As it is possible to buy standard sized rollers from bearing supply companies I checked the size to replace the missing roller but it is a non-standard size and so this was not an option. There was no alternative but to order a second complete connecting-rod and big-end assembly ...just for one roller!

A day or so later I noticed what I thought was a shiny piece of swarf of the floor. No, the workshop fairies, who obviously have a sense of humour, had returned my single roller. And so I now have a complete spare rod/bearing assembly for a Harley Davidson EVO engine. I had never planned to build a second engine but I'm certainly acquiring enough parts for one.