Offenhauser Mighty Midget Racing engine

[Eccentric]

I am going to try my hand at developing plans for an Offenhauser Mighty Midget Racing Engine. There were many variations manufactured through the decades and my version will likely end up being an amalgamation of many of those. I will also be taking liberties for ease of manufacture/assembly and to increase the chance of it being "a good runner"--or at least a runner. Inspiration came from Terry Mayhugh's build of Ron Colona's Offy and I will be taking advantage of his engineering insights shared in his (Terry's) build log. Why don't I just build that engine? Well, it is a very complex model and I don't feel ready to tackle that level of sophisticated craftsmanship. Second, I want to have control of the plans and have the ability to freely distribute them if they ever get to that point. The variant of the 97 Cubic Inch Midget Offy I will build will have two valves per cylinder instead of the large Offy's four valves per cylinder, the crank will be supported in three places instead of the large Offy's five and the 97 cu in Mighty Midget Offy is smaller overall. I will design in 1/4 scale so the model engine will be about 5.375" long, 5.5" tall and 4 inches wide.

Source: Fred Offenhauser Photos and Premium High Res Pictures - Getty Images

Source: https://www.amazon.com/Offenhauser-Legendary-Racing-Engine-Built/dp/1626540411

Source: Amazon.com

Source: Amazon.com



I will use the line drawings available in the book noted as a source for the attached pictures, then create 3d solid models of the major components, and I literary mean solid as there will be no internals initially. I will layout the moving components, timing gear train, cam shaft, crankshaft etc as simple stick models to define their geometry. Then I will add increasing detail to the engine, realizing there is a high degree of interdependence between all the components; a small tweak in one place will ripple through the whole engine. Once I am happy with the 3D CAD model I will begin developing the plans themselves. Most plans I have seen are manufacturing method agnostic, that is, they can be used for manual machining methods as well as CNC machining. My plans will be developed specifically with some limited CNC machining in mind. Home grown CNC routers are more common now and mine has given me greater flexibility in producing a complex part in a reasonable amount of time. What this really means is that I will not only produce a set of dimensioned prints, but some IGES files designed for specific machining operations. If someone chooses to build to my plans, they will not need to create their own 3D CAD model from a set of dimensioned prints, they can use the supplied IGES files.

I have scoured the internet and downloaded lots of pictures, enough I think to guide me in producing a realistic replica.

I scale a 2D drawing and import it into my CAD program.

Then I extrude the major components.

And create an assembly.

Then begin dimensioning the major components.

I will start nailing down the models specification's, for example I think it will have a bore of .75"

 

[stevehuckss396]

Great project. I'll be keeping an eye on this one.

 

[mayhugh1]

Very ambitious. Am looking forward to following your progress. - Terry

 

[kuhncw]

Which CAD software are you using?

Chuck

 

[a41capt]

Looking forward to following your build. Ambitious project, but one that’ll be rewarding!

John W

 

[Majormallock]

Count me in please for a set of plans, looks a great engine and your approach works for me

 

[propclock]

Fantastic I will be following along. I have a 3/4 axis cnc mill
and also would like to know what software you are using.
Beautiful engine there.

 

[dnalot]

I will be following along as well

Mark T

 

[Eccentric]

Gear Train

I need to define the gear train that drives the camshafts and magneto (distributor) from the crankshaft. My goal is to define the pitch and tooth count of all the gears.

Below is a picture of the Midget Offy's gear train.

Source:

The constraints for the gear train include:

• I would like to use a "standard" gear pitch. The smallest that can be considered standard is either a 48DP or a .5 Module (which is about 50DP).
• The camshaft has to turn 1/2 the speed of the crankshaft.
• Would like to have a single plane gear train for simplicity. The Mighty Midget has a dual plane gear train as can be seen in the photo above. The dual gear in the gentleman's hand is the idler between the crankshaft pinion and the gear train in the gear tower.
• The camshaft gear, as well as all of the others, has to fit in the gear tower housing
• The crankshaft should have as many teeth as possible.
• There should be as few bearing sets as possible. This is an advantage of a two plane gear arrangement, multiple gears can use a single bearing set. In a single plane gear train this means the number of gears between the crankshaft gear and the camshaft gear should be minimized.
• A plus would be to have as few gear types as possible, also "standard" tooth counts would be preferred.
• There are other mechanical constraints such as the 88 degree inclusive angle between the two valve banks and the distances between the camshaft and crankshaft.
• Also the magneto needs to turn at the same speed as the crankshaft.

If a single plane gear train is used, the camshaft gear has to have twice the number of teeth as the crankshaft gear. It does not matter how many gears are in between or how many teeth they have. I made an excel spreadsheet to prove this to myself. I think I will settle on 48DP, .5 Module does not give me a measurable advantage in tooth count. The largest camshaft gear that fits has 32 teeth; this gives me a 16 tooth crankshaft pinion. I would prefer to have more teeth, but this is acceptable. I might design a dual plane gear train just to see if it would be worth it.

I will be using the gear pitch circle in my diagrams, this diameter is calculated as the number of teeth divided by the diametrical pitch. If I am using 48 DP then it is easy to translate back and forth between pitch diameter and number of teeth. My go to book for info on gears is Ivan Law's "Gears and Gear Cutting".

The above example uses a 16 tooth crank pinion, 32 tooth camshaft gear used in three places and two 50 teeth gears. Below is another option, but it does not fit in the gear tower as the idler gear interferes with the highlighted bolts.

Would like to see if I can fit an 18 tooth pinion and 36 tooth camshaft gear. Below I try this with the gear tower housing.

Opps, the gear driving the magneto needs to have the same number of teeth as the crankshaft; the magneto needs to turn at the same speed. This is because the magneto will be used as our distributor and this is a four stroke. I add this to the list of constraints.

Also I have been using the gear's pitch circle, when we look at the gear outside diameter, as below, we find that the 36 tooth gear does not really fit. The distance between the gear and the outside wall is only 1/16" of an inch, too small. Back to the 32 tooth cam gear and 16 tooth crank pinion.

With this configuration I have 4 gears between the crank pinion and the camshaft spur gear, I think this is the minimum. I have four different tooth counts, 16, 32, 40, and 56 with a diametrical pitch of 48. I will move forward with this configuration for now, I believe it meets all of the criteria I laid out. I may find more issues as I make more progress elsewhere.

This is where I ended up on this Christmas Eve.

 

[aonemarine]

Following...

 

[Jan Dressler]

Hello... Nice, and very interesting project!

It kind of reminds me very much of my own, the type 35 Bugatti engine... Only that I work on mine since 2013 (not continuously, of course, but there are a couple of 1000 hours into research (most of them) and CAD work).

Based on my experiences, a few comments:

I will use the line drawings available in the book noted as a source for the attached pictures,

That is basically exactly where I started. The CAD pictures remind me very much of my own.

then create 3d solid models of the major components, and I literary mean solid as there will be no internals initially. I will layout the moving components, timing gear train, cam shaft, crankshaft etc as simple stick models to define their geometry. Then I will add increasing detail to the engine, realizing there is a high degree of interdependence between all the components; a small tweak in one place will ripple through the whole engine.

Exactly. With this approach, you need to get all the confinements that define the parametrics of the CAD model really, really right.

"Add increasing detail" means doing more research finding more drawings, other books, pictures, doing dozends of screenshots of youtube videos, and so on... And you will, on one point, stumble upon a thing that basically throws everything down again. One major dimension that needs to be changed a tiny little bit to be "just right", one moment of "Oh... THAT is for what this feature is for" that means that a whole lot of things have to change to be "accurate"...
You will forever know if something is just not "right"

And this means the parametrics of your model have to be very well thought out, and "robust". Ask me from where I know that...

I have scoured the internet and downloaded lots of pictures, enough I think to guide me in producing a realistic replica.

Trust me, there will never be "enough" If you want it to be realistic, you will end like me, visiting museums, meetings where cars with this engine are supposed to turn up, and so on and so on...

I will start nailing down the models specification's, for example I think it will have a bore of .75"

My scale was actually dictated by the crankshaft roller bearings... The exact same type is available in 1:3 scale, so 1:3 it was.

Good luck, I will follow this with interest!

 

[Eccentric]

Inside Timing Gear Tower

I import pictures and drawings into my CAD program (SolidWorks) and scale them to my working scale, this includes rotating them to establish some horizontal and vertical reference points. Most photos and drawings are not very good representations of a part as there is parallax and vanishing points and focal distances that distort the resulting 2D image. But I am not interested so much in just capturing an outline, but understanding the relationship of surfaces. For example the outside edges of the gear tower that contains the gears to the cam shaft have a negative inclusive angle, that is the sides are not parallel and get closer together towards the top; I have this angle shown as 3 degrees below. I'll take many photos and drawings and attempt to ascertain this angle. There were variations between engines and drawings so I am a able to take a little liberty in establishing an angle that works for my design, looks representative to the eye and is true to the information I have at hand. But I don't agonize over getting it "perfect".

Source: Offenhauser 110 Engine


Serial number 214 Built early 1947

I have a spline estimating the curve shown in the highlighted red circle above. I don't like splines, I try to reduce complex curves to a series of tangent arcs, but sometimes splines provide the best fit. When using splines I try to reduce the number of points to the minimum, maintain tangent entry and exit points and have some mathematical relationship between the points. Not that engineers design this way, but the eye is very sensitive to minor visual imperfections. Below is another try at estimating this curve using just two tangent arcs (highlighted in red). This is the simplest and I think the best.

In the picture below the included angle of the gear tower is about 6 degrees. Also the two arcs represent the side curve pretty well.

Dated 1935 - Source Gordon Eliot White's "Offenhauser the Legendary Racing Engine and the Men who Built it" (as all following diagrams)

In the real Midget Offy, the shape of the gear tower was dictated by the size of the gears it was housing and these were as varied as were customers for the Offy.

Drawing from 1953

1965 Drake Engineering Drawing

So what do you think? I am leaning toward going with 6 degrees. OK, that is one dimension, now on to the next. I hope this provides a little insight into my process for developing a CAD model without having the factory drawing set (wouldn't that be great).

I smell turkey cooking in the kitchen and my wife is called for me to get the extra leaves for the dining room table as company is due soon. So this is going to be it for today, Merry Christmas/Happy Holidays everyone.

 

[stewart drummond]

cam gears vary according to the customer ? cam rotation is 50 % crankshaft rotation no matter the cam/s or the customer for that matter

 

[Eccentric]

You are correct, the gear ratios did not change, but the displacement/bore/stroke frequently were custom, so the height of the gear tower, thus the size of the gears was changed. That was a lot of work. The ease of changing the bolck and gear tower height is one advantage of using a chain drive as opposed to a geared drive.

 

[Eccentric]

Front Timing Gear Tower and Bearings

The front timing cover at first glance looks like a mirror copy of the rear timing cover, but on closer inspection one can identify some features we would like to include in our model. The front timing cover is thinner than the rear timing cover as well.

Features of note on the front Timing Gear Tower:

• A - a raised lip around the edge of the front cover that extends out past the edge of the rear cover.
• B - Sizable radius on edges of the front cover, much smaller on the rear cover.
• C - Countersunk mounting hardware joining the two timing cover halves. Also there is a boss that extends above the edge radius noted by B.
• D - raised round boss and support web to house main idler bearing. This is a double gear so the bearing housing extends further forward.
• E - Raised Boss for larger mounting hardware holding timing covers to the block
• F - There is a raised boss that the camshaft gear cover mounts to.

This is also a good time to define the stack up of the timing gears, their bearings and the bearing supports machined into the timing covers. I am in the US so I am most familiar with imperial dimensions and parts, but metric components are just as easy to obtain here. In some instances metric parts are cheaper and more readily available, bearings are a good example. I am thinking of using a 1/4", 3/16" or 5mm shaft size for my timing gears. Since the gears are 48DP (imperial) I will start with imperial dimensioned shaft and bearings. A quick search on the internet and 3/16" X 3/8" X 1/8" bearings and 3/16" shafts should be a good place to start. Gears 7/32" wide with small shoulders should fit well.

I know what cosmetic features I want to add to the Timing Gear Tower, but this is good enough for now. No sense putting in that detailed work if I need to reconfigure everything later. I am also missing the bevel gear housing for the magneto, but that will not impact the general layout, so that will have to wait.

 

[Eccentric]

Front Cover

Today I will be working on the front cover, this is indicated by the arrow in the picture below:

The early Midget Offy's like the one above hard mounted the front engine mounting plate to the engine as can be seen above indicated by the red circles. The flat plat with the wings on it mounted the engine to the frame. A similar plate is on the rear of the engine. This was a problem because the race cars had flexible frames and the twisting of the frame imparted a twisting load to the engine block and caused them to crack.

The picture below represents a later version of the Midget Offy with the front engine mount removed. It it can be seen that it is no longer bolted to the front of the engine, but is allowed to pivot on the crankshaft cover. This allowed the race car chassis to twist without imparting that twisting load to the engine.

Source: The Miller/Offenhauser For Sale Page

However, we are modeling an earlier vintage Midget Offy so we will have the larger boss for the motor mount rigidly mounted to the front of the engine.

The crankcase needs a little work to blend into the front cover. Below is a wonderful picture of the crankcase without the block and we can get a good view of the shape of the crankcase where it meets the front cover.

Source: In the Shop: 1934 Offenhauser Engine - Update



I create a cocktail napkin sketch of the front cover, more to capture the features I want to create than to get the dimension correct.

Then I start working on the 3D model.

This front cover and crankcase will require some more work. As I add more parts into the assembly model, there are more interactions among them and more time is spent making them all work together.

Here is where I ended up today. It is going to be raining the rest of the week, so I hope to be spending a little time each day continuing the work on the Midget Offy.

 

[Eccentric]

Side Covers

The side covers are very distinctive feature of the Off and will be fun to model. The crankcase breathers are afixed to the top. I start by importing a picture with a side view, scale it and overlaying the outline.

Here is a side view showing how the middle is raised above the edges and the fin detail. The side cover and breathers are one of my favorite features of the Offy.

Then take a shot at the swept vent feature.

Then begin to lay in the cooling fins. There are a total of 9 on the Crankcase cover.

And the Block cover plate in yellow. I also worked a little on the timing gear tower and added the housing for the bevel gears that drive the magneto. That is the extension on the very left middle. I also cleaned up the front cover. The magneto is mounted atop the small shelf at the top of the front cover.

 

[petertha]

I was looking at your 88-deg included angle (44-deg per side bank). It also corresponds with the real world photo. Wondering out loud if it was 90 & 45 respectively. Have you been able to find any documentation that states the angle, or an independent method of verifying? It may well be 'odd' like that. I don't know Offy's well but I recall other models are noticeably shallower angles so there must be an underlying design reason.

Reason I mention this is I went through a similar exercise on a few radial engines: imported a digital image into CAD, adjusted axis & aspect ratio as best as possible, did overlay drawing on a single cylinder, typically vertical #1, then rotate copy. Result was many of the image cylinder axis did not line up to the CAD model which was even multiples of 360-deg/number of cylinders. Similarly, if I extend the centerline down to the oil pan, it doesn't align there either.

I suspect between back in the day hand drawn drafting, multiple copying & state of printing technology, there are many opportunities for distortion so one shouldn't expect perfection. It's actually impressive its this close. I guess in the absence of hard data, you as the designer are in charge. Just wondering out loud if 90-deg included angle would make your life easier from machining standpoint? Looks like lots of other design features cascade off of that initial angle decision - cam gear train, valves, seat, head...

(not a critique, just food for thought)

 

[petertha]

What are you contemplating for fuel & ignition system?

 

[Eccentric]

Peter.


You make an excellent point. I "thought" I read that the midget had an 88 degree included angle between the banks of valves, but just now I have not been able to put my finger on it. I measured several drawings and I "confirmed" the 88 degrees. But now I am wondering if it was confirmation bias that lead me to measure the 88 degrees; the tendency to have new information conform to an existing belief. But I really like your point with respect to the manufacturability of 45 degrees vs 44 degrees. Even if there is clear documentation showing the included angle is 88 degrees, it is in my best interest to round that off to 90 degrees. If I used 44, I am sure one would end up 44 and the other 46. I have tooling for 45 degrees but, 44 degrees would be a nightmare.

I do have several drawings calling out the valve bank offset in the larger 255 and 270 cu in Offys being 36 degrees.

I never take input as a negative critique, I appreciate your comments. Keep 'em coming.

As far as ignition and fuel I had not given it much thought other than to use hall sensors triggering the spark and a functioning distributor to distribute the spark to each of the four spark plugs. As far as fuel goes, I am not going to make this a high compression engine, that will not be to scale. I like Coleman fuel with a splash of WD40 (shaken, not stirred). I am planning on two carburetors.