Wobble-plate axial engine design project.

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
I split these posts up a bit, otherwise they get too long with my re-edits and addons.

Topic: exhaust release temperature:

1) Normal exhaust header pipe temperature is actually cooler than the muffler core temperature, with my 2-stroke engines.
2) high performance 4-stroke engines tend to get a bit of a red glow going on the exhaust manifold.
That would be over 530 degrees C .
Turbo exhausts definitely glow, as do formula one high-revving engines- the old V10s.

Possibly there is still some low level combustion going on with Carbon Monoxide and unburned hydrocarbons.

This clashes with the 200 degrees C predicted.

3) if a car engine is working hard, there is a lot of radiant heat coming off the exhaust manifold, even if it may not glow.
Enough to melt plastics several inches away- I found that out with a mk1 ford Cortina 1400- it melted the plastic clutch
pressure pipe.

4) if you rev up a 4-stroke motorcycle off-idle with no exhaust header pipe, blue flame comes out.
This shows that gas is still ionised, and probably over 400 degrees C.
Possibly delayed combustion due to low chamber pressures?
4-stroke Motorcycles tend to biff a lot of unburned fuel out the exhaust at low speeds, due to long overlap.

5) Anecdotally, I have heard of slightly glowing exhaust manifolds which you can see at night, if it is a sort of stripped down beach buggy front-engine car- definitely not road legal! You must have at least a firewall, and I think that exposed-engine hot-rods may not be that legal anymore. - fans and belts can be a hazard-

Possibly a firewall and a windscreen is enough for accidental fire protection.

With bad exhaust and inlet valves, you can set a nice bonfire going in the air cleaner.
Some aged filter material is not that fire resistant.

No air cleaner is an invitation to setting your motorcycle or car on fire- the old Triumph motorbikes used to do that frequently.
With some kind of backfire through the carburettor.

The carb tickler system used to burn quite well, set the carbs alight, then spread to the fuel pipes and fuel tank.
I have seen the result, but have not had it happen to me.
Barrel engine cross-section.
Features shown:
1) sparkplug
2) hold-down bolt locations.
3) mixture channels
4) two ports per cylinder, 10mm bore each, 30mm cylinder bore.
5) cylinder port seals.
6) centre well seal for inlet.
7) two rotary valve ports shown.
8) proposed finning. , with a possible trim line.

Details are not fully shown for each sector.
These ports should give 10 degrees-50 degrees timing.
This needs several cross-sections to show:
1) front gears
2) dividers in the rotary valve. This is to be all inlet, except for the exhaust channel.
3) piston cross-section.
4) bolting pattern for the rotary valve cover.
5) bolting pattern at the centre cross-sectional split.
6) far end cover and inlet manifold.

There needs to be a seal between the front gearbox and the exhaust area.
This will allow oil fill of the gear chamber.
A controlled amount of oil can leak through to the inlet side, so it needs topping up after each run.
All the main bearings will be in the inlet premix area.
another seal between the lower outer edge of the rotary valve and the exhaust area could be a good idea.
The bolts in the rotary valve surface could be plugged to reduce pocket of gas.- cover plugs need to be well retained, but removable.

I am not planning a separate slipper plate for now, but I will test whether the aly -to-nicosil mating face is a good match.

Re: obtaining the materials:

Probably casting blocks from ingots and then machining would be good.
I can get this done fairly cheaply by a local foundry.
I will draw up the block shapes and check it out.
The material is only available in ingot form.

It would be a major exercise making casting patterns and getting them cast.
Easier to just machine from solid.
The castings can be roughly the correct dimensions.

What do you think?
Getting the sunken finning detail between the main casing flanges could be tricky.
The spacing between the barrel tops and the valve face is quite small.
The cylinder barrels could be recessed into the main head plate, to give room to recess the bolt heads.


  • barrel engine cross-section   25-02-22.jpg
    barrel engine cross-section 25-02-22.jpg
    225.2 KB · Views: 75
Optimisation of chamber shape and sparkplug location.
1) the pop-top on the piston can be made taller.
This allows the sparkplug to lay down totally flat, and lets me use a bigger rotary valve disc and larger head ports.
The head ports could go up to 12mm from 10mm.
2) this change could allow me to adjust the exhaust collector ring shape, and shorten the height of the rotary valve.
3) the head split line can be just above the compression ring. This allows more room to recess the retaining bolt heads.
4) provision should be made to locate the head on the barrel section.
A slight barrel recess will do this without having to add extra dowels.

drive flange "nose":
This needs to be longer, and an area added for the magnetic pickup.
Is a single pickup still OK, or do I need a sawtooth pickup?

Cooling fin form:
another fin can be added to the cylinder at right angles to the trim line.
This makes a block of 5 parallel fins on each facet.
possibly the centre fin could be deleted.
The other 2 fins on the inter-cylinder case divider still help cool the cylinders.

Accurate matching of the barrel halves.
Is a slight step acceptable here, or do the bores need to be done on a line borer?
when the 2 case halves are joined, the bores are quite long.
This could be contracted out. I will check availability of boring services.
New images added.
Currently, the ports are up to 11 mm diameter.
The new head thickness is shown.

Question: could the exhaust collector have an inner ceramic lining?
There are exhaust people in NZ who do such things.
Ceramic top coatings have been put on pistons.
This will cut heat transfer into the end gear case, and the head.

The head is now wide enough to have its own fins.
Most of the heat transfer comes via the head. The barrel fins are not so important.
Possibly the head fins could extend out a full inch? Ten mm is shown at present.

I need to draw a layer of just the head and the rest of the head bolts and gear case bolts, plus a side view.

Should the top section gear case bolts be fully outboard of the exhaust collector ring?
They can't really go anywhere else.
The actual flange can be further up, so the bolt shanks are close to the collector wall.


  • barrel engine cross-section V2  26-02-22.jpg
    barrel engine cross-section V2 26-02-22.jpg
    212.1 KB · Views: 67
  • barrel engine lengthways-section V5  26-02-22.jpg
    barrel engine lengthways-section V5 26-02-22.jpg
    400.4 KB · Views: 61
Last edited:

1) finning for head section
This should project beyond the head profile

2) stirring fan for inlet and oil distribution- crank speed.
This could add a slight overpressure as well.

3) single carb adapter.
Initial carb- scooter float-slide type, 16mm bore.
This is rather heavy for a model aircraft engine, but will work better than a needle-spraybar-barrel throttle type.

3) Tiedown bolting for the gear case and rotary valve upper surrounds.

This could be getting closer to actual individual part drawings.


  • barrel engine cross-section for V6  27-02-22.jpg
    barrel engine cross-section for V6 27-02-22.jpg
    254 KB · Views: 55
  • barrel engine lengthways-section V6  27-02-22.jpg
    barrel engine lengthways-section V6 27-02-22.jpg
    376.8 KB · Views: 54

This design is looking quite heavy. probably 30% of the overall weight is in the head area.

This layout would make quite a good standard cycle 2-stroke though.
This would also drop its cost of build quite a bit.

With main transfers of 13mm diameter I could get a transfer pressure ratio of around 3:1, which is quite high.

This would allow less blowdown time, and better use of available volume.

This would also allow an overall weight target of 2 kgs to be achieved, at the expense of high fuel consumption at full output.

I could keep the central stirrer fan, use one carburettor, and 3 sets of reed valves.

As a 2-stroke, I can get pressure pulses and use a Walbro diaphragm carburettor, which is a weight saving, and a fairly good carb.
This also removes 4 largish gears and two big ballraces.

The skirt at the combustion end needs to be longer, so the piston needs a bolt-together construction.
This is not hard to arrange, and is used in the Axial energy designs.

I will do a drawing.

This should really be called a "swash-plate engine" now.

Here is the drawing of the two-stroke version.

I am currently working on single cylinder two-stroke mods, so this would be a good continuation.

The advantage over single and twin two-strokes is better balance, plus they sound better.
Probably quite high pitched, though. You want good muffling.

I like the idea of 4-strokes, as they are smoother at idle and off-idle, unless they are full-out 2-valve speed engines.

They sound just as rough as two-strokes.

Check out a startup of Burt Munro's Indian, on youtube!

A problem is removal of excess material from within the piston structure.
A possibility is fine thread holding the piston crown and upper ring land on.
The "2-Stroke Stuffing" guy has pistons with a threaded skirt.
This would be similar.
The pumping ends is easier, as it doesn't need long skirts.

The need for the long skirt is to blank off the transfers from the main case.
The transfer leakage through the piston gap shouldn't be significant.
This is similar to more standard engines stopping transfer blowby out the exhaust.
The second drawing is and end view, with some sectional details.
Volume in the transfers looks good for now.
I made the nose section partly conical to take it out to the 45mm prop drive flange,
but also give more room for head finning. It could just step out rather than be a cone.
It uses 4 short head bolts, and three separate cylinder heads.
The upper barrel finning can just follow the lines of the head finning.

There were some doubtful areas with the 4-stroke head, in whether it would cool properly with air.
A two-stroke is easier to air cool.

I think this design shows promise at coming in at around 2 kgs, and should make slightly more power than a
two-valve four stroke of the same capacity.

It also is likely to use 30% more fuel at full power than a four stroke.

A highly economical 2-stroke is usually vastly detuned.

I used to have a RD350 motorbike, and that would drop down to 15 mpg if I thrashed it.

That would "disappear" a tank of gas fairly quickly! :)

An equivalent 350 Honda 4-stroke likely wouldn't get much below 30 mpg if treated the same way.

The Honda made most of its power around 9000 rpm, and didn't have the same "get up and go" feeling.

A 450 Honda 4-stroke would easily beat the 350 Yamaha 2-stroke on a race track.

However, I have heard that the 750 Kawasaki triple two-stroke was good for 40 mpg (US) cruising.
That was likely the later detuned model.

That is quite a powerful bike- king of the 750s. Only the later 900 4 Kawa could beat it at that time.

I see the big Kawa 750 triples are now selling for 50,000 USD, fully restored. They are getting a bit scarce.


  • barrel engine cross-section for 2-stroke  28-02-22.jpg
    barrel engine cross-section for 2-stroke 28-02-22.jpg
    267.8 KB · Views: 73
  • barrel engine section for 2-stroke  28-02-22.jpg
    barrel engine section for 2-stroke 28-02-22.jpg
    202.1 KB · Views: 66
Last edited:
Hello Owen.

Is this engine actually going to be built or are you still in a design phase to see if it is possible? This is interesting. Thanks for posting.
It is a design phase for now.
I have another engine project going now, just waiting for materials.
The 2-stroke version looks promising.
I am a bit doubtful whether the 4-stroke version would be properly cooled.

Additional air cooling channels for the 2-stroke version:
This allows better flow over the head, and extracts hot air out of the engine core.

The central bearings rely on blowby oil mist for lubrication,
and the nose bearing can be a grease-sealed option.

I don't think offsetting the piston rollers is good for the needle bearing.

The lower boss can be narrowed a bit. At present, I have left room for a circlip.
I am planning 14mm wide x 10 bore for the bearing at present, with an 8 mm roller edge.
Additional oil channels will be needed.
There needs to be a thrust washer each side as well.

The high speed needle rollers will be a bit experimental.

The rest of the design is fairly conventional.

Care is needed on assembly fitting to get the "nip" right on the angular contact bearings.
The centre bolt on the stirrer fan will take the propeller thrust load-
One x grade 8 m5 bolt- Loctite?

The reed valves off one of my present engines should fit well. I can order more.
I need to find a source of 30mm piston rings.

Also, I need to investigate accurate track cutting for the swash plate.

Does the surface of the swash plate track need to be hardened and ground?
Do the roller outers need to be hardened and ground?
Some form of surface hardening may be best. I will look into this.

Can I get add-on features for a standard mill that will do the job?- profiling to shape, and grinding?
I will start googling this.


  • barrel engine vent-end 01-03-22.jpg
    barrel engine vent-end 01-03-22.jpg
    207.9 KB · Views: 58
  • barrel engine vent-side 01-03-22.jpg
    barrel engine vent-side 01-03-22.jpg
    297.8 KB · Views: 59

1) Lubricating the piston roller needle bearings is a real problem.
I can't think of any easy way of scraping up oil and channelling it into the bearings.
simple oil mist is insufficient.

2) possibly I can have a one-directional scraper on the cylinder wall, and pressurise the roller wrist pins a bit, with capped ends.
The pin could have an oil hole in it.
Would this work in all the piston orientations?

3) A pressurised oil jet directed at the roller axle oil scoop would be better, but this then runs the risk of over-oiling the cylinder bores.
I would need a skirt-level oil scraper ring, and a separate oil scraper ring at the pumping end, and a oil circulation system with
a dry sump, and a bladder-type oil reservoir.

4) Fitting a gear drive for a double oil pump is complex, plus oil pumps are complex,
and hard to find in small sizes.

Could I use air pulsations to drive a circulation system?
Any ideas on pulsation-driven oil pumps with a respectable delivery?

5) Could I use premix as a lubricant, with a separate feed from and to the fuel tank?

6) Could I use low temperature plastic fuel pumps as lubricant pumps?
these are small plastic gear pumps, about 5mm gear size.

I can run these off a brushless DC system, run off the main system 6V , or 2-cell lipo?

I have some pumps here I can modify to brushless drive.- or just use the fully enclosed brush motors at 12V??

This could be a bit of a fire hazard, plus the pumps don't like running dry.
They are rated for petrol.

- Maybe a reserve holding tank, so the engine stops first.

The motor units on the pumps are pretty massive. they need a 2mm "D" shaft input, but only need 10 watts or so power unit.
They look like about a 40w brush motor at present, intended for refuelling operations.
I would need to regulate fluid pressure to avoid overloading the motor.

In this case, (lube with premix) excess oiling is not such a problem.
The petrol-diluted oil shouldn't over-oil the bores, I think.

7) how should I vent the cases? If there is a lot of petrol vapour in there, I can't vent into the exhaust- It will flash back.

I don't think flashback is a major risk with normal 2-strokes, unless the ring gap is too large, then it will not run properly anyway.

Mr "2stroke Stuffing" had a backfire which took out the reed valves, but that engine was rather sick.

Possibly I can vent to the exhaust after passing it through a flash arrestor.

Can you buy small ones of these?
How fine a mesh should I use for one.- 0.5mm holes, woven mash?

Any other suggestions?

8) The needle bearings need a reserve supply of oil if they are going to spin at 20,000 rpm.

The main ballraces at 6,000 rpm can survive on oil mist, plus residual a oil coating.

9) **** I could possibly have a small oil trap in the piston, which does not drain,
which can be primed before first start, then holds reserve oil between runs, or during storage.
The reciprocating action of the piston could assist oil delivery.

(different topic)

10) piston backbone structure is also a concern.

Less than half the bore can be used for stiffening the piston.
Level of end loading is in the region of 2 kN mechanical, but maybe 8 kN combustion.
1 kN = 220 lbf.
Will this cause any bending stress-raisers? (bore = 30mm or 1.2 inches)
Does the piston need a high-tensile spine, or would an built-in piston alloy beam along one side be sufficient?
This also had the roller pin bores through it.

This beam section also acts as an anti-spin bearing on the edge of the swashplate.
Possibly this area should be thickened a bit.

Alternate wobble-plate system.

This deals with the high-revving bearing problem, but may be a little undersized for the loads.

It allows twisting, rocking, and transverse motions.

This could be made a little larger by increasing the piston length, and having a larger centre section.
either square, or larger diameter.
I have left out the needle roller, and instead use plain bearings.
The centre spider could be bronze to avoid having bearing insert sleeves.

This design has an added advantage that fancy cam-cutting gear is not needed to build it,
at the expense of having a few more parts.
6 pin and bearing sets with circlips;
6 rollers ;
1 swash-cam plate.

centre spider;

2 more angular contact bearings;

2 complex crank webs with nip bolts;
3 x cross-slide "pistons".
3x 10mm pin sets with retainers

Assembly of the crank:

Press-fitting could be possible, but the shapes look like nip bolts and slots would work.

Over-cuts may be needed to make sure that about half the nip bore is free to deflect.

question- would an interlocking nip like this actually work? would a full split like a bearing cap work better?
The two halves of the counterweights could be bolted together, as they are quite small.
Hollow dowel tubes can be used for alignment, and standard grade 8 bolts.
Getting all the bores right for total runout could be problematic.

A standard pressed-up crank is not so tricky, as flywheels can be bored in pairs,
at the same time.
Possibly these can be bored as a unit, then cut apart later, discarding the centre section.
It can also be made in 2 halves, and temporarily fastened together.
Choice of material?
" Medium-carbon steels are similar to low-carbon steels except that they contain carbon from 0.30% to 0.60% and manganese from 0.60% to 1.65%. Increasing the carbon content to approximately 0.5% with an accompanying increase in manganese allows medium-carbon steels to be used in the quenched and tempered condition. "
This stuff?

Similar mechanisms have been made in the past.
an actual arc of a circular toothed gear would do a similar job, but not as smoothly.

You would get sliding line contact.
This layout has actual bearing-type contact.

The centre element could also be a ball, and the cross-slide could screw or bolt together.

What do you think?
Here is the Almen engine.

I don't know how he arranged to deal with side motion of the ball-head.
I will look at this more closely
Thinking about my mechanism idea:
I don't think it will replace a ball centre, as the fore and aft rocking motion always has to be square on to the bore axis.
The twist has to be aligned with the wobble spider arm.
My method does not do that.
I will look at this again.

Extending and twisting the whole spider arm may not work so well.

there is quite a high bending moment that must be carried over somewhere.
this causes a lot of friction in sliding contact.


  • alternate wobble-plate method 01-03-22.jpg
    alternate wobble-plate method 01-03-22.jpg
    263.7 KB · Views: 67
  • almen internal.gif
    almen internal.gif
    63.8 KB · Views: 70
Last edited:

I have come up with a compact UJ spider layout that should fit.
1) plain bearings.
2) main wobble spider in bronze.
3) piston in 4 pieces bolted together, including bearing side plates for the UJ
4) the rod from the wobble spider is not located in the cross-bore of the UJ, but is free to slide.
5) wobble-plate rotational torque is taken by the UJ and the piston.
6) The UJ spider material can be medium carbon steel, surface hardened and ground.
Possibly the bore can be left as a reamed finish? at a machinable temper.

I will draw this to scale tomorrow, but it looks like it will fit.

I will look up relevant bronze properties and check proposed section stresses.
it looks like the wobble bearings can be set closer together.

The Almen gif looks like a lot of detail is missing. The artist may not have understood how it worked.
The wobble assembly looks impossible to assemble as it stands.
There is no wobble spider extension detail shown. The engine would not work as shown.
Plain bearing pressure is around 5-6 Mpa for the sizes I have selected.
This seems to be acceptable for sintered bushings.

I am trying to find tables of various materials with bearing properties and values, with Google.
I have had one of these before- I may have more luck searching my files???

A hardened surface is suggested for the steel part.

Possibly a thin bush could be used, if the wall thickness is reduced to 2mm, and the side pins are shortened to 7mm wide by 10mm diameter.
I have used 85 % of the projected bearing diameter as my working area for the bearing, boundary lubrication, similar
to a wrist pin plain bearing.
This should work with mist lubrication, if there is side clearance, or if there are lubricant channels provided.
Pre-lube on assembly.
the existence of side-thrust complicates lubrication a bit.
grooves in the side-thrust faces may improve lubrication.

4-stroke engine usually squirt oil under the piston, and have a small cup and hole at the top of the piston-rod eye,
and a matching hole through the bush.

the orientations of the bearing may mean that extra efforts need to be made with oil scrapers and passive slingers to get enough oil to the right place.

Sacrificial components would be a good idea, and frequent strip and inspection, along with alterations, to make sure the bushings will work reliably.
Very small 2-stroke engine work fine with plain bearings, but the surface speeds are lower, even if they spin at 15,000 rpm.
The Ali 2-stroke axial 2-stroke outboard motor engine 1922
This looks similar to what I am working on.


  • alternate wobble-plate method V2 01-03-22.jpg
    alternate wobble-plate method V2 01-03-22.jpg
    100.3 KB · Views: 66
  • The Ali 2-stroke outboard engine 1922.jpg
    The Ali 2-stroke outboard engine 1922.jpg
    24.5 KB · Views: 59
Last edited:
I did a thought experiment:
If the wobble spider is in a position where two of the three pistons are offset towards the top of the bore,
the arc of the tip stays on the line between the piston axis and the engine centre line?

This does not sound correct, as in this position, looking down square on the two pistons the projected lines are converging.
If I use my extending device, then the effective two tips of the extending arms will spread further apart, which causes a side motion problem.
How far do they spread in this case if motion along the arms is 3mm?
The arms are 120 degrees apart, so an outward motion of each tip of 3mm along the arm will spread the arms by 2x 3xsin60 mm. or 5.2mm,
2.6mm offset per piston. that spreads the pistons apart
what is the relative motion if I was directly above one of the pistons?
This spins the whole array around by 60 degrees, foreshortening the displacement.
the angle again is 60 degrees, but displacement is the cos this time, or 2.6 cos 60 = 1.3mm sideways.
This direction is important as I want to take side-thrust in this direction.

Maybe there is a way of managing this displacement so that it results in a twisting motion rather than a linear displacement.
The previous mechanism with the little cross-piston possibly did that.
That would need another axis along the piston??

This means that the spider cannot be restrained fully against turning using this system.

Things seem to remain in the correct relative positions that a simple wrist pin and universal joint can be used with a piston rod,
as per the Duke mechanism, but this cannot be a simple UJ layout.

Does this sound correct?
Possibly the Duke engine has a matching offset in the UJ at the spider end that keeps things lined up?
The 2 axes of the joint they use do not seem to intersect.

Can anyone graphically transform this problem to make it easier to visualise?

I wonder if there is a mod I can make to my sliding idea that would work better, and can cancel out these oscillations.

I will try some different layouts , and see what I can find out.
I am not sure whether this is a real problem or one of my perception.

I drew a few drawings, and the extension happens at the top and bottom of the stroke, and not at right angles to the pistons,
if I look at it when 2 are in the plane of the page.
When one is at full stroke, the other two are at half the distance from the centre position to the top, but in the opposite direction.
There is a possible crossover position in extension, but this does not matter, as it doesn't affect the geometry.

Where are the other two pistons, if one is at the centre position in the stroke?
I will work that out, and try to draw it.
Last edited:
Views to scale of the latest proposal, with some "true length" geometry.
some values seem to contradict each other, so more work is required here.

correct figures:
1) true spacing of cylinders = 40mm from centre.
2) two cylinders are spaced 69.3mm apart.
3) the apparent swash angle is 23 degrees.

4) find the actual swash angle:
16.5 and 40 are given; by squares, we have 1600 + 272.25 = 43.27 *OK.
actual angle = (16.56) = 22.5
at 22.4, (16.49)
say 22.45 degrees.
5) find the projected length of the spider arm at 120 degrees rotation.
Both pistons are at the same height at this angle.

projected triangle c = 20mm, a = 22.45.

then 120 degree displacement (s) , r = c/ cos(22.45) = 21.64mm, s = 8.26mm

6) Now apply these value to work out the true length of the spider leg.

the true angle of the leg plane is 60 degrees, c = 21.64mm, r= c/cos60 = 43.28mm

this is spacing the two pistons apart excessively, as s = r sin 60 = 37.48mm

spacing = 2s = 74.96mm, vs correct = 69.3 mm, which is 5.66 mm too much.

I think this is correct reasoning, which means simply extending the spider arms is not a solution.
possibly a combination of a ball and an eccentric sliding piston may work better.
5.66mm is quite a bit to absorb, however.

I will consider some more mechanisms.


  • end view for revision 7 with angle diagrams 03-03-22.jpg
    end view for revision 7 with angle diagrams 03-03-22.jpg
    238.3 KB · Views: 60
  • side revision 7 with angle diagrams 03-03-22.jpg
    side revision 7 with angle diagrams 03-03-22.jpg
    240.8 KB · Views: 57
  • piston top for revision 7 with angle diagrams 03-03-22.jpg
    piston top for revision 7 with angle diagrams 03-03-22.jpg
    177.3 KB · Views: 57
Last edited:

Radial double-ball links with a master link.
This one should work.
It possibly doesn't need all the balls, but the sockets are easy enough to make.

I am not sure of the source of all the operating misalignments.

Alignment is "true" at top and bottom dead centre, but seems to have a mismatch at other positions.
The behaviour of the 3 pistons should be similar. There may be a rotating torque applied to the non-master cylinders,
but it should average out.

One piston set takes all the torque loading for the whole engine, but it should be capable.

What is the loading effect on the master link? It has a 6mm shaft, but is fairly short.
I will work it out.

Piston structure is similar to revision 7.

The ball sockets are split in both cases.

The rectangular slide is retained by the shape of the ball, plus the housing.
Ball clearances are slightly loose in that case.
Possibly provision should be made to nip up the crank assembly more positively.
This has to be hand-fitted to both preserve bearing nip and shaft runout.
Various thin shims may be handy.

The crank could be made in one piece and split afterward by removal of the centre section.
It is also split like a big end cap.
Possibly machining areas around the bolts and dowel areas will allow final boring as a bolted unit?
The external shape is not so critical.

What do you think of the prospects for getting the balls on the links?
I was thinking: soften the balls, drill and press in the rods, heat treat, hand sand the balls a bit smoother.

Brazing the balls on could be done as part of an oil quench and temper. That is about the right heat??
Temper back to blue oxide? It would need to sanded a bit first.


  • mechanism  for revision 8  03-03-22.jpg
    mechanism for revision 8 03-03-22.jpg
    164.9 KB · Views: 54
Last edited:
Mechanism for revision 9:

1) The ball swivels have been abandoned.
2) side rocking is now taken up by piston twist.
3) the central sliding pin has been put back.
4) arm misalignment is now managed by bottom pins, in flat 6mm arms.
5) the "wrist" action is now taken by a 20mm dia. x 15mm wide bronze pin.
6) the bronze pin runs in steel slipper plates, hardened and ground cylindrical surfaces.

Problems now dealt with:
1) wrist surface pressure is now at an acceptable level.
2) Twist and slide is removed from the wrist action, reducing surface speed at the wrist surface.

the "PV" calculation as a measure of bearing load, is now comparable to a normal wrist pin bearing.
Notably, this is about a factor of 100 above standard values for a journal bearing.

An engine has a strong dynamic component for wrist pin lubrication, so that it is a lot better than a standard
grease-lubricated, constant load bearing.

There should be enough oil slinging about from mist lubrication and component movement, to
supply sufficient oil to the sliding joint areas.

It is somewhat experimental, but looks promising.

colours on the drawing are a bit off due to the white-out, but it should be legible.

Please ask is some details are not clear to you.

The fairly high maximum "V" value at the surface could lead to heating problems due to friction,
but the oscillating surface speed, passing through zero, may help with heat dissipation.

Notably, standard big-end needle bearings seem to run close to their heat dissipation limit under mist lubrication.
I spotted a you-tube article where a small engine maker reduced the big end pin size to fix an overheating problem.
I think this was a pressed-up flywheel engine, but quite small.
The difference to a conventional wrist pin is that maximum load is at zero rotational speed, and almost no load at maximum surface speed.
This allows a larger diameter to rpm/load ratio with no increase in heating from surface friction.

Next problem: angular contact ballraces have to low a rating, only 3.35 kN static.
I am sure we need at least 6kN static.
the mechanical dynamic load opposes combustion, so it is not a problem
I will also look into leverage effects.
There is a lever multiplying effect up to 20.6 kN.
I will look into taper roller bearings., and axial loading.


  • mechanism  for revision 9  05-03-22.jpg
    mechanism for revision 9 05-03-22.jpg
    274.6 KB · Views: 57
Last edited:
according to the loading fomulas, the nearest taper roller bearing is
This is too enormous to fit, so I will next check roller and thrust bearing sets.
This will make the crank assembly wider.
I will try to stick within the 32mm outer diameter but possibly up another 10mm?
Needle roller sets with inner and outer races may be more compact.
The single load rating is 10 kN radial and 20.6 kN axial.
Axial rating jacks right on one edge, so this is not ideal for thrust bearings.
They generally like a nice push along the shaft direction.
incorrect assimptions.
I can treat any axial load as if appried centally, and moments only affect radial load.
In this case, a tapered toller bearing of will do the job.
I wiil draw this up in all 4 bearing positions.
The radial load at the main shaft bearings is lower, but this is still a suitable bearing.
Last edited:
Lengthways cross-section with taper roller bearings.
This engine layout seems to need more substantial bearings than a standard crank and rod system.
A taper roller bearing that was exactly half-way between axial and radial would be nice, but SKF don't make them.
axial load = 6.6 kNmax, radial load = 10 kN max , by leverage.
Skf bearings are about 1:4 ratio.

static 0.5Fr + 0.11Fa

dynamic 0.4Fr + 2.2 Fa
Fr = radial force, Fa = axial force.
L10= ((C/P)^3.33)
where L10 = 90% reliability life in millions of cycles,
P = applied load,
C = rated dynamic load.

and L10h, which is the same thing in running hours,
L10h = 10^6/6x rpm)x L10
usually, P should be less than C
if P = C then L10 = 1 million cycles, or at 6000 rpm, hours = 10^6/6 x 6000)x 1= 27.8 hours.
if c/p = 1.2 then L10 = 1.06
Static load rating is based on peak loads.

However, there are shock loading, peak load, cyclic load, and average load factors to consider, which would extend life into the thousands of hours, potentially.

I can now draw up the wobble hub, the arm links, and the crank throw pieces.

I have specified the entire wobble hub as bronze. Stresses are not very high.
Tin-lead-phosphorous bronze would be good.
This is fairly hard, so the steel parts need to be hardened and finely finished.
The lead is there for free machining.
Very high lead bronze is a lot softer, and not suitable for this job.

The cross pins can be pushed in separately, as they are well retained.

The radial 10mm round could be a cut-down wrist pin, and screwed in place.

Adding a screw boss to the radial arm is a problem, as the threading needs to be done before hardening.
A 5mm grade 8 socket-head screw should be strong enough.

Possibly a higher grade is needed, and very hard Allen drive head.- torque wrench?
Usual Allen keys are rather soft.

I will check.

This needs quite high fastening torque as per big-end nuts.

They generally are not locking nuts??

This would need a special hardened stepped washer or cap, as the wrist pin hole is over 5.5 mm in diameter.

Possibly the 6mm section could be machined out of a 10mm section of medium carbon-manganese steel.
Getting small quantities of this stuff could be a problem.
It may need to be imported internationally.

Similarly, the blanks for the barrels and pistons need to be cast from ingots. The ingots are available.
A barrel blank 90 long x 180 diameter is a fair bit of aluminium.
.den 2.7, vol= 2.29 x 10^3 cu mm or 2.29L
at 1000l per cu m, = and 2.700 kg per cu mt.
1l = 2.7 kg, x 2.29 = 6.2 kg.
Fully machined, this will go down to around 200g or so, so a fair bit to remove.
remainder = 3.2%? does this look possible?

This bearing arrangement would not be a good mass production proposition, as a lot of measuring and swapping shims is needed.
A better system would have sliding half-sleeves that can be clamped in place, or threaded sleeves and huge ring nuts.


  • barrel engine  bearings- revision 10  06-03-22.jpg
    barrel engine bearings- revision 10 06-03-22.jpg
    310.2 KB · Views: 47
Last edited:

Start of total mass estimate:
1) core of crank.,
If it were solid steel, dia 40 x 110, it would be about 1.1 kg.
Level of voids is pretty high though.
Free space inside bearings. - 20%
free space in a counterweight crank - 30%
Free space in the centre spider assembly with 3 arms in place- 20%
overall free maybe 25%, residue = 750g.

2) Basic cylinders: 160 x 30 dia, 3mm walls, D = 2.7 x 3 = 149.3 x 10^-6 x 2.7 x 10^3 = 400g

3) Finning would probably be more than this, at 35% material. discs cover 80mm, dia 36, 60
Area = 324, 900, 1809, 434 x 10^-6 x .35 x 2.7 x 10^3 = 410grams
This may be a little lower due to longitudinal fins, centre casing interference- but that could count as part of the finning.

4) cylinder heads : dia 80, 20 sliced off one side, 40mm tall, maybe 50 percent material?
some at 100 % , some at 35%.

226 x 10^-6 cu mm, x 2.7 x 10^-3 = 610 grams.
Total so far = 750 + 400 + 410 + 610 = 2170g.

5) Then we have the pistons, the nose assembly, the rear plate assembly, some ducting, reed valves.

6)The pistons are maybe 30% solid, 130 x 30 dia, vol = 29.25 x 10^-6
mass = 29.25 x 2.7 x 10^-3 = 79g

7) Total mass at this stage = 2249g.

We look like we are heading close to 2.5 kg overall.

I will revise this as I do actual drawings.
I have been using the wrong formula for second moment of area.
it is I = pi/4 (r)^4 and s = My/I
Thus for round sections in the link rod, at 10mm must be 12mm diameter, at 22mm must be 14mm in diameter.
I will work it out for a few sections at r = 22mm (the moment arm), and redo the hub of the wobble plate.
The bearings should still be OK.
I will find the correct square section, then adjust for width and depth.
This means to top spinner needs to be changed as well, if its diameter must be 12 mm.
Length gain of 21mm.
new dimensions shown.
12x12mm is ok to 225 mPa.
The 10mm section of bronze can be much wider- a full disc if needed.
moments about a point: 6.5 x 27 = 12 x 15 approx.
usable stress is around 150 MPa.
peak force by moments is12kN, but this can be spread over about at a lower average force.
at 12 kN per sq cm, this is x 10^4 to get to a sq m, or 12 x 10^8, = 1200 MPa - far too large as a bearing pressure which should be
around 30 MPa.
Some more reconsideration of this design is needed.
Possibly it can be made wider, and the loads can be taken partly by horizontal pins.
Maybe the pins can be raised above the bearings,
and the outward link can be made much shorter- like about 18mm.
This means it and the pistons need to swing more to take up the possible 3mm misalignment.
3mm over 18mm = around 10 degrees.
but the piston is also rotating over 15 mm radius, so the final angle will be around half this.
I will look into this and post again.
- this picture is not practical. I will post another soon


  • barrel engine  bearings- revision 11  06-03-22.jpg
    barrel engine bearings- revision 11 06-03-22.jpg
    151.4 KB · Views: 57
Last edited:

Latest posts