Old School Drawing Exercises and 2D CAD

The dog ate my homework.....

Honestly this 3d drawing stuff you all post on here is fantastic.

I did a mechanical engineering apprenticeship in the 1970's and never got further than drawing spur gears in isometric projection with H2 pencils, set squares and other drawing instruments......

Regards to all

Derek

Further to SOD's post, the top half of the cylinder sucks in air from the atmosphere mixed with fuel, and pumps it into the receiver. This air fuel mix is ignited as it passes into the bottom half, thus increasing it's volume. Much the same as a gas turbine except the fuel is introduced into the air after compression.

I'm now expecting to get a detention as I haven't even started on the homework set by Jason, and don't intend to, 3D CAD is above my pay grade

Spur gears in isometric projection are advanced stuff to me Derek. I did science A-Levels which came bundled with an hour a week of Technical Drawing. Just a taster - no exam at the end of it, and I'm pretty sure no gears were drawn. I may have been asleep!

Anyway, on with homework. I've never attempted a job like this so it will be fun to see Jason's red-ink. The goal is to translate a 19th Century etching (which may not be technically accurate) into a number of patterns from which the castings needed to make a model can be produced. Ideally a working model!

I find this intimidating. But, another homework dodge, here's a development plan. Planning is another skilled job, but:

• Must have a defined goal (making patterns isn't the same as fabricating a working engine)
• Several Stages, each ending with defined products, peer reviewed with issues fixed or tolerated.
• Each stage broken down into a series of steps and resourced.
• Good plans have all the stages mapped out, but only the next stage fully detailed. Decisions ideally deferred so changes can be fed into later stages. Avoid developing a fully detailed end-to-end plan from the get go because these may turn into a Death March. Making it up on the fly is equally bad. OK when done by an experienced individual, comes completely off the rails as soon as a new boy tries it or expensive resources have to be coordinated. Try to find problems early; the cost of fixing mistakes multiplies by roughly 10 for each stage passed. Fixing a dimensional error in Stage 1 in CAD is dirt cheap, whilst the same boob missed until 500 castings have been sent could bankrupt the project.

Stage One: Gather Information, identify Risks & Issues, & Tooling

1. How does the engine work? A. Brayton Open Cycle, issues for later:
   • How is the engine be fuelled? Pressurised Gas/Air is risky
   • How is the fuel ignited? Not clear from the drawing
   • How is exhaust managed?
   • Does it matter there's no governor?
   • Is there an important relationship between fuel-pressure, reservoir size, and fly-wheel weight.
   • As the cylinder burns and compresses fuel at the same time, what piston seal will keep flame away from the fuel mix. This is my only serious worry so far.
2. Tools, chosen because I know them. You might prefer different
   • Paper, pencil & calculator
   • QCAD to collect dimensions from the image and approximate 3D coordinates. Also for doing 2D profiles.
   • Fusion 360 for 3D modelling.
   • Maybe a 3D printer to check scaled down parts.
3. Establish dimensions, noting the drawing isn't isometric, & although probably realistic, is likely lacking details.
4. Decide development strategy:
   • Is it possible to model the engine realistically in 3D and then retrofit the model to produce patterns? (Noting Jason's hint about 'shrinkage allowance' )
   • Should development 'fast-track' patterns, designing other parts to fit castings, or design patterns last to suit the working parts. My inclination is to give working parts top priority because it feels easier to design a frame to suit moving parts rather trying to squeeze a mechanism into a fixed frame.

Stage Two: Translate 2D Dimensions into Obvious 3D Parts

1. Decide what the parts are. Many obvious, like cylinder, piston, flywheel, others less so. Is the bedplate/frame 1 part or 2? Is the top bearing block part of the frame or separate? Is the reservoir part of the frame or separate? Can the reservoir be cast into the frame directly, avoiding machining?
2. Produce obvious parts in 3D and assemble to check rough fit. Adjust cosmetically so the engine looks authentic whilst making sure it will work in the real world. Candidates: Bedplate and frame, cylinder, wheels and crank-arm

Stage Three: Complex 3D Parts

1. Candidates: piston assembly conrod, conrod fixings, fuel-intake fixtures, throttle, crankshaft, bearings, pipework, cross-head & guide. Adjust for standard size stock.
2. Check fit by assembling with Stage 2 components, revising if needed. Note interference, assembly, fixing and other issues likely to cause practical problems later.

Stage Four: Nail Uncertainties

1. Candidates: Exhaust, Lubrication, Fuel

Stage Five: Convert Preliminary Fit into Final Fit, Generate 2D Drawings, Review, Correct and Refine

Stage Six: Build and Test Prototype (Not necessarily the whole engine, or full-size, or in the real world.)

Stage Seven: Identify castings & develop patterns.

Stage Eight: Build a Real One

Stage Nine: Blame the Innocent and Reward the Guilty

Stage One progress:

Jason's Flywheel has me worried. I'm not sure how to produce the 'S' shaped spokes in Fusion 360. Might be tricky. Time will tell.

Dave

Dave.

I did a two-rail sweep (in MoI) to create the spokes.

Draw two curves to represent the spoke outline with an elipse at one end. Sweep said elipse along both curves and it will scale in accordance with the distance between the curves.

If Fusion 360 doesn't have that feature, then you should be able to Loft a number of various sized/positioned elipses to get what you are after.

Martin.

Thanks Martin,

Firing up my Windows 10 Laptop triggered a big update after which Fusion no doubt will also have to be refreshed. In the meantime, I tried lofting in FreeCad, similar to MOI I suspect except there's only one rail.

Two ellipses on different planes joined by a spline curve, producing:

Although the curve's not correct for Jason's Flywheel the method works as a demo I expect Fusion to be easier to use, but my Windows machine is still downloading, installing and rebooting. Oh, and the Windows mouse is missing!

Dave

Going back a bit to the 'simple' exercise, is it safe to assume that both circles are in the same position on the Y-axis? Although they appear to be, the dimension line does not pass through the right-hand one. I am genuinely interested to know if there is any way of calculating its position from the dimensions given, except that it is 24mm from the left-hand one on the X-axis. If you centre both circles on the same horizontal the exercise seems very simple using just T-square and compasses.

Graham's cylinder, head and frame pattern with half a corebox along side

part of the top bearing yoke pattern I cut earlier today

I tend to draw to the size I want the model in Alibre including machining allowances on the faces that need it and draft angles where required then scale the model at the end to increase the pattern for shrinkage allowance. Fusion makes it a bit easier to draw the finished part then add draft and machining allowances before scaeing the shrinkage.

Delighted to see I've not got it completely wrong. This bad-boy has got to be the James Dean of Internal Combustion engines! Every home should have one!

I guessed wrongly the engine didn't have inlet and exhaust valves! The gubbins that works them must be round the back.

This clip from Jason's photo reveals what's on the mysterious bottom of the cylinder, which I was ignoring:

I now think the claw holds a rocker arm driven from the back of the engine, and the Inlet Valve is the spike thing. Still mot clear to me where the fuel connects, or how the flame works.

This is a fascinating engine, a nice mix of clever engineering and genius design, but with many early imperfections. Very inventive considering it's a piston engine based on the same thermal cycle as a modern gas turbine. (I'm sure Duncan is right.) I'd pay money to watch a few of these ticking over at an Exhibition.

Dave

Here's my attempt in Fusion.

Is it safe to assume both are on the same axis? No, but it's not unreasonable. Drawings are supposed to avoid unnecessary clutter, but it could be a mistake. I can't see a way of calculating the second circle's position unless it's on the same axis as the other one.

Have you tried doing the exercise with just a T-Square and Compasses? It's not hard to approximate the drawing by eye, more difficult to get it spot on geometrically so that everyone produces exactly the same drawing. Consider Points A, B, C, D and E added to Gary's original:

A is easy, but the exact length of A to B is harder because point B meets the arc radius R100 at a tangent. Not a blend from a near miss, but a bulls-eye. And the centre point of R100, point D, has to be found exactly so that the arc is tangent at both ends. Likewise the centre of the R50 arc, point C, also has to be found such that the arc intersects at an exact tangent with circles at both ends. Point E (arc R15) presents the same problem, but is easier to solve.

The exam problem is finding points A,B,C,D,E accurately rather than drawing something like Gary's original. (Even though a blending approximation would be 'good enough' to make one of these in the real-world.)

It can be done accurately with T-Square and Compass but I don't think it's simple unless the draughtsman already knows about tangents.

A decent CAD package makes tangents easy because tools are programmed to calculate them. Having been told the ends of an arc, radius R, must be tangent to other objects, it does the sums and finds the centre automatically within a few microseconds. Otherwise, the centre is found by knowledgeably drawing construction lines, which an expert can do in several minutes, while an amateur might fumble for hours, and fail!

Dave

Thanks for your answer, Dave. I genuinely wondered if it was a trick question and I had missed something.

I should obviously have included a rule and possibly a set-square with my T-square and compasses! For quickness I used the digital version (TurboCAD) to produce my 'approximation' of the drawing based on the assumption about the second circle, but I didn't use any of its other functions. Point D may be determined by the intersection between a parallel line 100mm below A-B and a 91mm arc centred on E. Point B is where this arc meets the line A-B. All other arcs were similarly centred as shown on my drawing. Incidentally, my coordinates for point D seem to agree with those mentioned in answer to Michael Gilligan's earlier question.

While none of this really matters here, it might have serious consequences in a real-life situation to make assumptions about missing dimensions on a drawing. No doubt someone will now point out some grave error in my work, on which I have now spent far too much time

Point B should not be too hard to find, I've added a couple of dimensions onto my hand drawn one posted earlier.

First draw a line parallel to the top horizontal AB 100mm below it

Then set compass to 91mm (100-9) and swing an arc with compass point in the far right point so it intersects the 100mm down line.

The intersection not only gives you the springing point for the R100 edge but if a line is projected vertically upwards this will be the point where the horizontal top edge joins the 100R arc.

Gary set the question and it had me going! My construction is similar to yours but I got Point D slightly wrong (the centre of the R100 arc). Martin (Blowlamp) showed how to do it with a parallel 100mm from A-B, which is the method applied by Jason. Gary's drawing tests our ability to draw an arc tangent between two circles and also between a circle and an intersection with a straight line. As always, it's easy when you know how!

I agree about ambiguities in drawings. The more I draw the more I appreciate how hard it is to get them right. Working in an old-time drawing office must have been tough. Canyon Diablo was the most violent town in Arizona after it became an accidental railway terminus. The railway was supposed to cross a bridge over a deep canyon. Unfortunately, when the metal parts arrived, they were too short to cross the gap. During the time it took to fix that little detail an anarchic wild-west town boomed in the middle of nowhere. The town died as soon as the bridge removed the need for trains to stop. All because someone made a little mistake...

Dave

Edited By SillyOldDuffer on 26/07/2020 15:04:31

Full marks and a gold star to those who managed to produce an image of the flywheel.

As Dave has not handed his in yet he gets the detention project of manually writing the 27,500 lines of G-code to produce this. One down..............

Edited By JasonB on 27/07/2020 13:07:24

The reason Dave's homework hasn't been handed in:

Got into a fight with Fusion 360, which I have to say is messy in this department. It's easier to do in FreeCAD.

No problem drawing two ellipses on separate planes. Likewise, seemed straightforward to draw a spline curve as a rail connecting them, except it isn't. The issue is how to connect two ellipses in 3D space when the model has three 2D sketches in play. The sketches aren't connected automatically, nor is there an obvious 'External Geometry' button. Whilst editing an active sketch, the create toolbar (click it) has a Project/Include option hidden away. It includes projections from another sketch, making them available for snapping in the current one. The two ellipses first have to be selected by this tool, turning them purple (as above) before a rail line can snap between them.

I may have made life difficult by switching on the 3D sketch option because moving spline points & angles in 3D is like driving on wet ice with bald tyres. For added fun, Fusion doesn't allow spline points to be grabbed directly by the editor as I expected, instead the Move Tool has to be engaged.

Mostly Fusion behaves as I expect it should, but this particular feature was hard work. Maybe I'm making life difficult for myself by not doing it right! I'll try again this evening.

As can be seen from the images I haven't got the spoke into its correct S shape let alone set dimensions and repeat copied it into a wheel. If Sir is in a very good mood I might get a C minus for effort...

Dave

This is how I did it in Alibre

Firstly the cross section of the rim & hub were drawn and rotated about the central axis

Next ellipse drawn on what is the horizontal plane in the above image. (blue)

Then the path that the ctr of the spoke is to follow (red) was drawn on what is the plane facing you

Then I used "Loft" to make the ellipse follow the path and added 3degrees taper so that the spoke kept the same proportions but got smaller as it radiated outwards.

The single lofted spoke was then repeated in a pattern of size about the central axis

Finally fillets added where spoke meets hub and rim

If you want to look a it in F360 click this link for the spoke, you can also go to manufacture if you want to look at the tool paths and simulate the cut and the g-code is there too for you to copy out as that may be easier than writing from scratch.

Hmm, the link just produces a blank screen for me.l

Not sure if you need Fusion or not, it should open as an image that can be moved about with an option to open in F360 so the file can be worked on by anyone, CAM viewed etc..

This is a screen shot of how it opens for me.

This is all I get. It must be something to do with permissions.

Works OK for me in Linux/Firefox which definitely doesn't have Fusion 360 loaded.

Looking at how Jason did his spline, I may have made life difficult by doing mine in 3D. Back to the Drawing Board.

Dave