Ball Hopper Monitor - Casting Project

[GreenTwin]

This was JasonB's 12 inch diameter flywheel, and I buffed off several spokes with a slow speed sanding sponge, and this method actually worked surprisingly well.
The speed has to be kept slow on the disk, and the pressure has to be pretty firm.
Any faster than a slow speed just makes a black melted smeared rough surface.
.

 

[GreenTwin]

Disaster strikes.
I am going to have to go watch a video to learn how to do supports.
I am not sure how to do those in the new slicer program.
The bed adhesion was excellent, and the print was actually difficult to remove from the bed, and required the metal sheet to be flexed to break the print away from the sheet.

The sides are pretty smooth, so they will be easy to sand.
The print did not progress to the curved part, so we will have to wait and see how smooth that is.

I will refine the model a bit, and make sure I have both coreprints on the next print.
I will probably thicken the wall to 0.25" if I can.
Thickening the wall on the window side will not be easy in Solidworks unfortunately, I don't think; maybe I will get lucky.

Anybody a "support" expert in Prusa slicer ?
.

 

[GreenTwin]

Here is a hand and soda can for scale.
I would really like to use a 16 inch flywheel, which woud give a 10 inch diameter hopper, but I think that would be a bridge too far.
I will stick with 14 inch flywheels and 8 inch diameter water hopper.

Even with 15% fill, the print below is very strong.
I can bend the open end apart 4 inches, and it shows no sign of breaking at the thin parts on the other end.
I may reduce the infill back to 10% or even 5%.
.

 

[Jasonb]

So did I win the bet as you have not even got the pattern to print correctly let alone cast

I was replying as I looked down the various posts, I can see now you have a small flat on one end of the print.

From earlier posts I did not think you had finalised a smoothing method that worked in all situations, I just see the big areas being done an not the detail.

Pattern was only 90mm deep and with draft n all sides.

Before you print another think about whether the pattern actually needs to be a hollow sphere. As you say now that you are not using it to produce the core it could just be made solid like you would with a wood pattern. The print time may be quicker as it is not having to do a detailed inner face just the percentage fill. You then also have a good flat surface to put your location pins in as well as holes to put in a screw to pull the pattern out of the sand with. You can put one of the cores from the printed corebox into it to save the amount of aluminium you have to pour so will still get just a "shell" as your aluminium pattern. that will need about 2 1/2lbs of aluminium plus whatever the gates and risers will want.

While your workshop time is limited I would put the time into the pattern design so that you have minimal additional work to do, once you start adding bondo and filling things with clay etc it is time you don't have.

EDIT, you posted while I was typing, shows what I said about the print at the bottom not being attached as that has disappeared.

 

[GreenTwin]

Place your bets now folks; will the 3D print be usable to create an aluminum permanent pattern half ?

No, you lose the bet because the question was "Will the 3D print be usable to create an aluminum permanent pattern half ?".
The question was not "Will the 3D print fail from lack of supports?".
The 3D print pretty much "as-designed" will make a good pattern for casting a permanent pattern, and no, I would not want to fill the interior because I want the permanent pattern to be a thin shell, to avoid shrinkage issues.

We will send you a consolation prize though; 50 cents.

The problem with the hopper shape is that the window really interferes with everything.
I went through three versions just getting the window to where it is now.

The flat side on the bottom opening is also causing lots of problems when extruding the coreprint.
Par for the course with 3D modeling and patternmaking.
There is a lot more work required to go from a 3D model to a pattern, especially on an engine with a lot of curves like this one.

This piece would be really easy if it were not for the window.
Since I have to reprint, I will revise the 3D model and tweek it to get a thicker wall, and fix the thin part at the coreprint.

.

 

[GreenTwin]

From earlier posts I did not think you had finalised a smoothing method that worked in all situations, I just see the big areas being done an not the detail.

The finer details will have to be hand-sanded, such as the waist bead.
The larger open surfaces can be smoothed with the sanding sponge, whether curved or not.
.

 

[GreenTwin]

You can put one of the cores from the printed corebox into it to save the amount of aluminium you have to pour so will still get just a "shell" as your aluminium pattern.

I suppose that is true.

Edit:
But if this 3D print is done correctly, then it can just be molded on both sides as a thin shell, just like the base of the green twin was molded.
There was no corebox used for the green twin base, but I will use one for this hopper.
Given the print time, I may rethink using coreboxes, and instead of coreboxes, just print two pattern halves, and have some coreprints that can be tacked into place to allow the two pattern halves to be used as either patterns or coreboxes.
.

 

[GreenTwin]

While your workshop time is limited I woudd put the time into the pattern design so that you have minimal additional work to do, once you start adding bondo and filling things with clay etc it is time you don't have.

That was a last ditch effort to salvage the print, if it had not failed.
.

 

[Jasonb]

Difference between THE print and A Print. The print is obviously not usable but A modified one should be.

Yes I know you want a shell master pattern, that is why I said " You can put one of the cores from the printed corebox into it to save the amount of aluminium you have to pour so will still get just a "shell"

Are you using Boolean subtract, that usually works well for jobs like this where a complex shape is needed for the core. and will simply repeat the shape inside the pattern.

 

[Jasonb]

You have been wanting to do a ball hopper monitor for years, slowing down a bit now won't make much difference to the overall time scale. Post the CAD and get a few useful comments. Print a few smaller test pieces while doing the CAD and get to know the new printer. Then print the part. Time spent now will save time in the long run.

Edit:
But if this 3D print is done correctly, then it can just be molded on both sides as a thin shell, just like the base of the green twin was molded.

It should not take long to do a comparison of a hollow print vs a solid one and see what the slicer says the time difference is. Then decide which way to go. That way you get to see what is quicker regards 3D printer time and ease of pulling the pattern out the sand.

 

[GreenTwin]

Post the CAD and get a few useful comments.

I am not going to post the CAD, because we all know that it would show up very quickly as drawings by "you-know-who", like has happened to many of my other engines.

The details of this engine are in the 3D model though.
The way it works is the basic geometry is hashed out in some fundamental sections in 2D CAD, then those sections are imported into Solidworks, and then the details of the remaining shapes for that part are worked out in Solidworks.

So the final 2D drawings will be derived from the final 3D models, with all of the pattern work suppressed.
The 2D drawings I have now are just the base geometry.

That reminds me, I need to add a bit on the bottom of the pattern for machining allowance.

The 3D models are done. I am probably shy a few small bits and pieces.
I have put the time into those models to get them correct, including close attention to parting lines and draft angles.

They don't require any further work, except adding machining allowances, fillets, coreprints, etc., in order to 3D print patterns.
I could actually derive the 2D drawings from them now, since they show an "as-machined" state.
I don't show fillets when deriving the 2D drawings from the 3D models, since it makes for an extremely busy 2D drawing.

The gears are commercial units, and I have those in-hand, as well as 3D models of the exact gears that I purchased, so I know the gears are both available, and also will fit exactly on this engine.
Everything rotates as it should in simulation without binding.
This is a very accurate Ball Hopper Monitor design, and far more accurate than anything that has ever been cast as a model prior, since I was able to look at both exterior and interior photos of a disassembled engine, and I matched the small details of a full sized 4hp engine.
This is not a rough approximation of a BHM.
If these parts were 3D printed at full scale, they would be interchangeable with an original BHM, and I think would be indistinguishable too.
.

 

[GreenTwin]

Are you using Boolean subtract, that usually works well for jobs like this where a complex shape is needed for the core. and will simply repeat the shape inside the pattern.

I am not having any luck with Boolean and Solidworks with the complex shapes on this engine.
Does not work.
We will get the first pattern half 3D printed successfully and buffed/sanded, and then see how we feel about trying to print a corebox.

As of this moment, I am beginning to lean away from a corebox, since that is going to require more 3D modeling work, and a lot more print time.
I think a simple solution is to just make a one-piece coreprint that goes all the way through the pattern half, and protrudes out both ends.
This coreprint is not symmetrical on the ends, but it would be very easy to 3D print, and would have a straight cylinder between the two protruding ends.

If the ends are removeable and hollow, then they would also be used to make the core when placed on the ends of the pattern half.

I see coreboxes as just being a lot more surface work, instead of just smoothing the inside and outside of 3D printed pattern halves.
.

 

[Jasonb]

I meant screen shots not the actual CAD files.

Yes I did mention that I could not see the machining allowance (in the screen shots) so a typical example of why posting may save wasted print time.

If that is a "Done" 3D model then I would say you have more than a few bits to add. You look to be almost done with most cast parts but Persionally I would not want to commit to casting until the whole thing including all the machined parts are in the assembly and it has been tested. For example without the intake valve casting and the valve, springs etc in place can you be sure that the hopper will clear them? Does the governor weight swing out correctly, at the moment there are no bosses on teh flywheel for it to fix to, let alone swing.

Although indeed a close visual match, without proper dimensions of all the parts I think you may be pushing it as far as interchangeable with full size.

 

[Jasonb]

If you want to send me the CAD file for the Hopper I will show how the Boolean should work. If not say so and I'll rough out a hopper.

 

[GreenTwin]

This is not a typical model engine design.
Most model engine designs are rough approximations of a full sized engine.
This is a rather exact smaller scale replica of a 4hp engine.
The approach was not a casual one, but rather each part was devoloped one at a time, with the mating parts designed to fit exactly.

With the factory sections shown in the original Baker brochures, you can get pretty exact with the way the engine was originally designed, if you pay attention, and if you care to adhere to the very fine details.

You can see in this wireframe the details of the valve in the valve chamber, etc.
This is not random approximate stuff.
Even the sparkplug is the same size as the one I will use.

I could point out quite a few deviances from the Pacific and Lone Star kits and a full sized Ball Hopper Monitor.
.

 

[GreenTwin]

If you want to send me the CAD file for the Hopper I will show how the Boolean should work. If not say so and I'll rough out a hopper.

I can send you the STL file, but it is with the coreprints.
The boolean you refer to will be different, since the pattern uses the outside of the coreprint surface, but the corebox must have a new object created that is different.

I understand how boolean works.
So you can play around with the STL, but it is not yet set up for a corebox.

.

 

[GreenTwin]

STL sent.
See if you can open it.

It should look like this.
.

 

[Jasonb]

As I agreed it is more details both visually and in some internal details.

A one or two off large scale model allows this.

The PMR and Lone Star kits are smaller so not always practical to be able to cast some of the finer details at least without putting the cost of a kit up considerably. A core for that valve chamber could easily double it's cost as the money is in the moulding work, the actual amount of iron is very small.

The thinner a part is also run the risk of chill so casting as a solid piece is far more likely to avoid that. The same detail at almost twice the size is likely to have more mass so will not cool as fast.

The guys who designed the PMD and Lone star ones were also practical guys who had built a number of model engines and knew what would work and what other hobby machinists would be able to do so the "dumbed down" detail was not just a short cut but a way of producing an engine that would work and that was affordable.

Will let you know on the file

 

[Jasonb]

STP Step file is best

 

[GreenTwin]

The Lone Star and Pacific Ball Hopper Monitor kits were/are extremely impressive kits.
The Lone Star kit was unbelievably high quality kit in iron.
I saw the work that Maury put into both the patterns and the drawings, and they are first rate by any measure, and the best iron castings I have ever seen for any kit.

I have noticed from the photos from Barney, and also from the original factory brochures that there are a few differences between an original BHM and the kits produced so far, but an untrained eye would not notice the difference.

There would definitely be trade-offs as the scale is reduced, in my opinion, as you can lose some of the finer detail.
.