Ball Hopper Monitor - Casting Project

[GreenTwin]

I guess we move on to the other pattern half for the water hopper.
I hope to be able to reduce the print time some on this print.
28 hours is excessive in my opinion, and I may even increase the layer height to shorten the print time.

I watched a tutorial on supports, and while I still don't understand them very well in the Prusa slicer program, I understand them better than I did yesterday, so perhaps I can tweek these settings.

Pattern print #2, here we go.
I used a almost a full roll of filament on the first pattern print, so I have to go find my box of filament rolls.
It will look like this screencap.
I added registration holes, but they will not be used since this is a pattern used to make a permanent aluminum pattern, and the holes won't show up in the casting.

 

[GreenTwin]

Played around with the paint-on supports in Prusa slicer.
Increased the layer height from 0.15 to 0.2, to save about 6 hours print time.
Hopefully this print will be a good tradeoff between print time and quality, and will work with these very minimal supports.
And reduced the infill to 5%.
Print time about 12.75 hours, which is a lot shorter than last night's 28 hours.

Here we go, print No.02.
.

 

[GreenTwin]

I am going to do a hot swap when the filament runs out, so I can use up the rest of the purple roll.
.

 

[Jasonb]

Pat , have you done anything different to the imported file such as using a Step file as that should eliminate the facets and just leave the layers to contend with?

What about the draft on the ends? did you add that this time?

 

[GreenTwin]

I tweeked the STL file that is generated by Solidworks, to try and minimize the facets.
We shall see if this helps.

I did not add any draft on the ends of the coreprints.
I don't think this is necessary.

There is no draft on the face of the flange either, and I may use a piece of mylar on this surface, since it will have to break free of the mold at 90 degrees.
If the end of the coreprint does have a lot of breakout, it would be simple enough to just run a knife or something similar down that face at molding time for a slight amount of clearance.

I like to use the least amount of features as a possible.
Less = more simple in my book.
And I don't want to use features that are not really needed.
As I mentioned in other posts, I don't try to solve problems that I am not having.
Better to focus on actual problems that may crop up.

I had a few false starts on the 2nd print, with the filament either breaking (I changed to a new roll), or the filament not feeding.
Apparently the filament sensor needs to be cleaned with compressed air every so often, but I don't have any compressed air at the moment.

Worst case I plan on buffing out the aluminum permanent pattern with a sanding sponge in a tool and die grinder.
There are three different grades of sanding sponge.
Typically the fine in a tool and die grinder will do a great job of buffing out surfaces, but if a slightly more agressive cleanup is needed, the medium or even coarse will do the job, finished off with the fine.

I would like to do as little as possible to the permanent aluminum patterns, and hope that I have to do almost nothing to them.

The frame have to be printed in sections, since it is longer than 14 inches.
I am going to squeeze the flywheel onto the 14" plate, and live with a slight amount of shrinkage, so the flywheel will only be 13.45 inches after machining, which beats having to print flywheel pattern halves for such a small difference.
Visually, the flywheel will look ok.

This print is working so far.
Fingers crossed.
At least it is just a 12.75 hr print, so worst case I lose that amount of time with a misprint.
Two prints a day is a reasonable print rate, given how many prints I need to make.

Edit:
For areas with no draft, if I use mylar, I would like to be able to pull the mylar before stripping the pattern.
The mylar I have is 4 mil, which is 4 thousanths thick.
I will have to figure a way to pull the mylar strips.
I don't think the mylar will adhere to the sand, it is pretty slick, although it does have a matte finish.
In extreme cases, I would use two layers of mylar, since I am sure they would slide against each other easily.

Edit02:
Interesting how the Prusa slicer puts a footprint at the base of the supports.
The plate adhesion has been really good, with no bed lifting at all, and no draft shield.
I have turned on the airconditioning very cold at times, with a cold draft blowing past the printer, and it seems impervious to draft.
My old Prusa was very sensitive to draft, and was bad about bed lifting.
.

 

[Jasonb]

Less = more simple in my book.

Seems less is going to entail more work, fiddling about with the sand and/or using mylar and having to cut that to shape and work out how to remove it. When a couple of mouse clicks would have put draft on which seems the simpler option to me. Plus the time factor is far less and you are tight on free time.

 

[GreenTwin]

I really like to have flat surfaces to start with when machining.
When I was 3D printing some of the patterns for the green twin, I did not really have a good feel for pattern making in the beginning, and so not all the patterns had draft angle.

So some of the green twin parts chucked up perfectly square and round in the 3-jaw lathe chuck, with the cast faces mating perfectly with the lathe chuck face, and so no tedious 4-jaw centering, and that has spoiled me forever.
I really dislike trying to grip a surface in a lathe chuck when the surface has draft, and I really dislike the egg shape that rapping on a pattern produces in a casting.
Bound sand castings pull out without any rapping at 90 degrees to the pull angle, and so the casting is as round as the pattern, which with a 3D printed pattern can be very round.

And the intent is to get a near-net casting, where most surfaces only need a very light skim.
That is the trend in the foundry industry, and many commercial castings these days don't require machining at all; the tolerances are that good.
I would like to have some surfaces that can just be buffed off smooth without machining, where a precise fit is not required.

Resin-bound sand can be pretty forgiving with lack of draft angle, especially if you strip early in the strip time period.
Petrobond (tm) and greensand is not as forgiving if you omit the draft angle.

Near net castings are really nice though.
So many model engine castings are more like blobs of metal, and often noticeably oblong.
I don't have any reason or excuse why I can't make near-net castings; it is not any more difficult for me to make near-net than to make oblong castings with draft angle.

I started out reading the Navy Foundry Manual, and other books about how to do foundry work.
Pretty much 100% of the ideas in the old books are obsolete and very dated.
If you use bound sand, most of the traditional oldschool foundry methods should be avoided.
Bound sand opens up an entirely new and radically better way to make castings, and the castings made with bound sand can be very accurate, and with very repeatable accuracy from one casting to the next.
.

 

[GreenTwin]

This print is progressing a LOT faster than the first one.
So far so good.
The surface finish looks perfectly acceptable, and I can't really tell that I changed the layer setting.
.

 

[GreenTwin]

As I suspected, the curved surfaces of the hopper pattern do not need any supports, at least so far.
Omitting the unneeded supports saves a huge amount of time when printing.
.

 

[Jasonb]

You must still be at the height where the supports were not doing anything on the other half, as the printed area becomes more horizontal that is where they are needed. There were doing nothing here to the main hopper ball

 

[GreenTwin]

So explain how the dog printed with no supports.
Apparently the bridging ability is significant, else it would have shown problems on the dog print.
Look at the overhang under the chin, or between the front legs.

.

 

[GreenTwin]

I think as long as it is printing an arched shape, it can span a long way without a support, mainly because it is additive in mostly the horizontal direction.

One area I have seen problems if no supports are used with a long flat span, and while it will keep on printing, a long unsupported span will sag.
I think it ties into the "overhang threshold" value, which I think I set to 55 degrees.
I will have to research the overhang threshold to figure out how that ties into supports.
.

Edit:
Here is some info.
https://forum.prusa3d.com/forum/ori...saslicer-setting-for-only-necessary-supports/

.

 

[Jasonb]

The dog is an interesting example, viewed as your photo then the print seems to be near horizontal between the legs so you think the overhang is large. Look at the dog from the side and it's chest is sloping up at an angle that is maybe 60deg from horizontal so the support is coming from one side.

The link also mentions you can't get the angle as steep when the layer hight is increased, so interesting to see how this one goes.

 

[GreenTwin]

The dog print demonstrates that you can get away with a lot of overhang on this printer.
Sort of defies gravity.

So far this print looks good.

And I did not support one end of the coreprint, and it seems to be doing ok, but is round.
I supported the other end of the coreprint, since it is actually flat.
.

 

[GreenTwin]

Print is done.
Seems like the changes to the resolution took care of the facets.
This print is slightly more coarse than the last one, but with a huge time savings.
I think this print is perfectly usable, and the high ridges can be knocked off with the sanding sponge.
Zooming in with the camera exaggerates the lines to some extent.

Print No.02 is complete.
.

 

[GreenTwin]

I am very satisfied with how these two pattern halves printed.
Not perfect, but completely usable.
You can see from the photo under the unsupported dome that it has some sagging unsupported lines in the beginning, but then it bridges nicely to finish up the top.
I will probably not modify the print settings for the coreboxes, since the interior can be cleaned up pretty easily.

I guess I will print the coreboxes, and then see if I can cast permanent patterns for these two pattern halves.
If all goes well, then this will be proof-of-concept about how I will approach the remaining patterns for this engine.

This build is a definite go.
It is a bit sureal to hold a mockup of a 1/2 size hopper in the hand.
I have dreamed of an larger scale Ball Hopper Monitor build for a long time.
Casting replacement parts for original full sized engines is not out of the question, but there are so few of these engines around that you would not need many.
This engine seems to be very rare.

Edit:
The methods I use now for casting an engine were worked out over a period of about six years (often very painfully) with the green twin build.
At this point I know what to expect from the 3D model, the 3D print, the molding sand, and the castings, and I fully understand how to run the burner and furnace at its optimum settings.
All of these items were new to me back in 2012 when I started the green twin.
This build starts at the end of a very long learning curve, so I anticipate this engine build to go much faster, and be predictable at each step.
I don't have to learn the methods as I go on this build; I know what works.
.

 

[ddmckee54]

Supports are always a P.I.T.A., especially on large prints. I try to avoid them where possible. Personally, I don't use supports if the angle of the overhang is over 45 degrees above the horizontal. However, once that surface starts to flatten out the printer REALLY hates it when you try to tell it to print a curve in mid-air. I really despise printing something that has a hole through an overhang for that reason.

I don't know for sure since I don't use it, but I think the Prusa slicer will let you manually add - or remove supports. That way you can have the support where you need it, and still reduce the print time by not printing supports where they aren't really needed. I use Simplify3D, so I when I do want to use support I tell it to automatically generate the support - then I manually delete the supports I don't want. I will also add supports to be sure a curve has a nice base to start from, the slicer can sometimes be a little stingy at times in this regard.

Don

 

[GreenTwin]

Yes, I added a few manual supports on the 2nd print, and that reduced the print time dramatically.
.

 

[Jasonb]

Facets seem to have been significantly reduced
Coarser surface is no doubt the increased layer height which really only shows towards the top, would be worth looking to see if your software does variable height like the BambuLabs as it would not add much time to the print if say only the top 1/2" were done with a finer height.
Shape and size has probably helped with the overhang, the dome is a self supporting structure and being large one layer has had time to cool before the next, Might be different if it were say just a pie slice of the dome.

Casting by the end of the year

 

[GreenTwin]

The facets are very much reduced, and even though the 2nd print uses a slightly thicker print, the 2nd print actually looks smoother than the first, due to the lack of facets.
It is difficult to distinquish the difference in layer thickness viewing the two prints side-by-side.

I will not try to adjust the upper layer heights, but in the future, I will probably do that.

One school of thought is to finish the hopper pattern halves, and coreboxes, and cast permanent patterns of that first.
The 2nd school of thought is to 3D print all of the patterns and coreboxes now, and then have a muilti-day flask-making/molding/casting session, and pour all of the permanent aluminum patterns at the same time.

The weather at the moment is cold, and so that does not lend itself to mold making in my cold shed.
I could use the small kiln to melt aluminum in my house (wife would probably freak), but the molds and flasks must be made outside, and so the melting may as well be done outside too.
Melting aluminum with the oil burner is quite fast, even from a cold start, and that is a real plus if you are doing multiple back-to-back pours.

I guess I could use a #20 crucible, but that can get a bit heavy to pour.
I still have to calculate mass on the various patterns, to get a feel for what crucible size to use.

If I get all the patterns 3D printed this year, I would be happy.
That would get me set up and ready for a large casting session.
.