The Ship Model Forum

David,

I am using a Nikon 105 mm macro lens for these photos. I have a Zeiss laboratory microscope that I could use to measure the actual diameters of these parts but I don't have a micrometer scale to use with it.

Phil

_________________
A collision at sea will ruin your entire day. Aristotle

Tom,

Right you are!

I took your advice and tried rotating the part before printing. I also changed the step size to 0.02 mm on the assumption that these tiny thin parts were allowing extra resin to be cured and this caused the horizontal parts to build up too thick with the 0.01 mm steps.

I tried three different orientations and printed 15 of each (I need 10 for the bucket vent):


The part on the left was rotated 45 degrees on the X axis and supported at three corners. The center part was rotated 45 degrees on both X and Y axes. And the part on the right was just supported horizontal.


Just rotating 45 degrees on the X axis actually produced the best parts with pretty uniform dimensions. However, only 11 of the 15 pieces survived washing. Here you can see that I added small stubs at each corner for the supports to attach to. On the earlier print with the supports attached to the grill itself the corners of the grill broke off when I tried to remove the supports.


This is the configuration rotated 45 degrees around the X any Y axes. I thought this would produce the best results. That's what I get for thinking. Only 5 of the 15 survived the wash (and I was pretty gentle, just letting the parts soak in ethanol for a few minutes). And of the 5 parts still attached to the support this was the best example. As you can see it has broken away from the support on the upper left and lower right corners. All of them had detached on at least one corner, and most were "folded" out of shape as you see here.


In this test the part was horizontal and supported at all four corners. 14 of the 15 were still intact after the wash. As you can see here, some of the openings right at the center are plugged. This was the best one of the lot. All had some blocked openings.


Although many of the grills separated from the supports while washing and handling I did recover most of them. This is a pretty high magnification shot of one of the loose grills. The divisions on the scale are millimeters, and the grill is 3.7 x 2.2 mm and measures 0.11 mm thick on the micrometer. Considering the size (about the size of a rice grain) it is pretty good. Certainly good enough to use on the bucket vent - assuming I can handle them and glue them into place without destroying them!

I took this photo into the CAD program and scaled the drawing to the actual size of the part. Then I measured the openings and the grid parts. The design size of the openings is 0.2177 x 0.3055 mm. The average size of the openings in the printed part is 0.1417 x 0.1752 mm, or 65% of the horizontal and 57% of the vertical dimensions.

The design dimension of the horizontal and vertical bars is 0.05 mm. They measured about 0.12 x 0.11 mm, or at least twice as thick as designed. I really can't see making them any thinner, especially when printing at 0.02 mm steps. Maybe they would have closer dimensions to the design using the black resin Tom mentioned.

There are many vents on the cruiser that had screens over them - some quite a bit larger than these. Using these dimensions, angling them at 45 degrees on the X axis and printing at 0.02 mm steps I may have a good chance of making them all.

Phil

_________________
A collision at sea will ruin your entire day. Aristotle

Phil:

Glad those worked out! One can see from this episode that there is a lot of product development involved for many parts. If it is appropriate that the screens be painted black, they will almost disappear. Printing time for many small parts is considerable, at one point thoughts of having more than one printer starts to occur.

Given there is an acceptable loss rate and some items may have further uses, often considerable number will be printed. Acquisition of plastic divider boxes (fishing sports) is a great way to (sort of) classify the small bits and pieces. For instance I have enough quad 40's extra for a major warship.

Cheers: Tom

It perhaps also shows that one technique does not fit all needs.

Such grilles perhaps would be better done as etched parts. Etching, of course, has its own issues, such as rounded corners for instance. It may be difficult to do this at home (I know what I am talking about), but a professional spray- or foam-etching plant should be able to turn out reasonable grilles with 0.2 mm square holes in 0.1 mm brass or nickel-silver.

_________________
Eberhard

Former chairman Arbeitskreis historischer Schiffbau e.V. (German Association for Shipbuilding History)

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Phil,

Awesome to see you pushing to find the limits on the designs.

Todd

Aha! I discovered the cause of the scaling problem with parts coming in 10x oversized into Chitubox!

Operator error!

The DesignCAD STL files (and files from some other programs) have problems with leaks, overlapping triangles, too many or too few triangles sharing a point, etc. These can all cause some printing programs or slicers to fail. There are several programs to fix these problems available on the Internet, and some printing services that will do it for a fee. Microsoft's 3D Builder is often recommended as a free and reliable way to repair these problems, and it does work with the DesignCAD STL files.

However, Builder was the source of the oversized parts. When I take the original DesignCAD STL files into Chitubox they are exactly the same size as the CAD program says they are. But the repaired files from Builder were 10x the design size!



The original DesignCAD output STL using millimeter units is the small blue object at the left - the 48 inch bucket vent. The enormous oversized part is the Builder output file using centimeter units.

Why? At the top of the Builder window is a dark blue bar with small black lettering (perhaps the worst possible choice for a control box). I failed to notice that it is asking for the file's units, and the default is centimeters. When I select millimeters the Builder output files are the correct size. Duh!



I think just about everything about the Builder graphical user interface is just weird, as if the programmers were from another p;anet. It bears no resemblance to any other program I have seen on Windows, Apple or Linux. But it does work.

Mostly. I discovered a peculiarity in the program. Not necessarily a bug, because it may have been designed this way. Perhaps it is part of the STL file definition. If you have a continuous closed edge it automatically fills in a surface. So if you have a plane with a hole in it the hole will be filled in IF THE PLANE IS A SINGLE OBJECT. If the plane is split into two objects that enclose the hole, then the hole will not be filled in. It took me 2 1/2 days to figure this one out. DesignCAD will allow you to create planes (surfaces of solids) either way, and the single plane/surface object surrounding a hole has fewer points and therefore a smaller file size. Since my CAD model was getting too bog (over a gigabyte and 22 million points) I was very careful to use any method I could to reduce file sizes. Now I have to watch for these things and redraw them.

Phil

_________________
A collision at sea will ruin your entire day. Aristotle

Wefalck,

I have been thinking the same thing, that the grills should be done with photo etch. And I also agree that it is difficult to accomplish this at home (ask how I know!).

If the metal is thin enough it should be possible to etch these parts. Since the bars in the grill are 0.05 mm (0.002 inch) wide, from what I have read it would be best to use brass/stainless no thicker than 0.05 mm. And to allow for under cutting and over etching it might be necessary to make the bars wider. This is far too fine a scale for what I have been able to achieve myself!

This test grill is pretty simple - just a part of a cylindrical surface. But some of the vents have conical surface sections. That presents some interesting 3D geometry problems to unfold them into flat patterns for etching. I have done this manually for a few parts and it is very time consuming! Fortunately, one of the DesignCAD users wrote a BasicCAD (the DesignCAD macro language) program to unfold and flatten complex 3D surfaces. It works nicely. It can be entertaining to take a really gnarly surface and run it through the macro. You might never imagine that the 2D pattern that results was once a 3D object!

For now I will continue to try to print these more complex grills on the 3D printer. I may experiment with some other resins.

***

I wonder if there are laser cutters that can work with such fine detail?

Phil

_________________
A collision at sea will ruin your entire day. Aristotle

Phil:

For fine grills, I have found at yard goods emporiums, open mesh ribbon which can be stiffened with a light coat of paint (airbrushed) and when cut to size works very well. It remains flexible enough to be curved for vents etc.

Cheers: Tom

I have been working on the TALOS missile launcher for the OK City, and ran into some unexpected and annoying problems. Here is an image of the CAD model I am working from:


I am still working on the launcher body but I have completed a first pass at the missiles and launcher arms:


The arms are just sitting on the missiles at the proper location, but I need to add some locating pins and sockets for fastening them together. Even at a small vertical step size the slots in the rails and the lugs on the missile do not mate (the lugs are extremely small).

The arms must slip into openings on the sides of the launcher body and then be fastened together.


Slots on the arms and lugs inside the launcher body will limit the angle of rotation from horizontal to +55 degrees - the highest launch angle for the missiles.

To get the port and starboard arms to mate and lock together I devises a "lock and key" mechanism:


This is where I had the unexpected problem. The starboard arm has the "key" and it usually prints correctly (I have printed these things several times now) although there have been a couple of failures. But the hollow "socket" in the port arm has failed several times. Here is an image from Chitubox showing the default supports the program adds:


Because the bottom of the cylindrical interior arm is pretty thick (3 mm) I thought it would be fine for supporting the upper part of the opening (remember this things prints upside down so the supports are on the bottom of the part). But when I printed at 0.05 mm and 0.03 mm vertical steps this is what I got:


The opening "caved in" and it has a oval cross section (instead of circular). All of the half dozen parts had this failure to some extent, and this was the worst example. Obviously the key wasn't going to fit in this lock!

Then I added extra supports to the Chitubox default set:


Notice the additional internal support and several supports that are above the maximum diameter of the arm. Here is the result:


So the extra supports did the trick - mostly. However, even with all the supports some of the prints of both the port and starboard arms have had the ends of the cylindrical arms squashed a bit to oval cross section on the side with the supports! Apparently the part is separating from the supports near the end of the cylindrical arm and this allows it to "droop" while printing. The attachment points are sized "Medium" with a 0.80 mm contact diameter according to Chitubox. Maybe I should use "Heavy" supports or larger contact size?

I am using the default motion timing for the Anycubic Photon Mono printer (lifting speed = 240 mm/minute). I wonder if the force of lifting the printed parts through the resin is breaking the attachments?

****

Another issue is apparent over exposure during curing. You can see the white flaky patches in places. From what I have read this indicated over curing.

I was using a 300 Watt photocopier exposure bulb (mostly white light with some UV) that I used for exposing photo etch. It took about 30 minutes to get a good cure. However, this incandescent bulb burned out, and I noticed the replacements have only a 20 hour life expectancy.

I ordered a Tresbro 20 Watt LED UV lamp with 405 nm wavelength. The adds said it was a "200 Watt equivalent" but of course the marketing morons fail to tell the luminosity (lumens) of the lamp.

https://www.amazon.com/dp/B085NMPW7B?ps ... ct_details

I read several articles on line saying that the 2-4 Watt UV lamps used to cure fingernail polish could cure resin in 1 to 5 minutes (thicker parts may take longer), and there were a number of 4 Watt versions of this same lamp. I cured these parts in a bottle of water for 5 minutes and then another 2 minutes upright and 3 minutes inverted on the turntable. Apparently that was a bit too long!

So that is my latest adventure in 3D printing. If anyone has suggestions for proper support or information about how to know the correct curing time I would appreciate it.

Phil

_________________
A collision at sea will ruin your entire day. Aristotle

Phil:

Perhaps a simpler design, round stud in round hole? Square hole, round peg, big hammer? Drop of CA to secure when ready for assembly? I think items that need to remain dimensionally precise and stable should be printed vertically. Really the horizontal axis are the only ones accurately constructed.

Good luck! Tom

Another interesting lesson, indeed, that dimensionally crucial open features should be build up in the direction of printing.

I wonder, whether one could not have printed a thin membrane in front of the opening, keeping it in shape and which then would be easy to remove without damaging the actual opening.

Another option could be to leave two thin holes in each part, which are less likely to substantially distort, and then to insert a metal pin connecting the two parts in a defined position. If needed, the holes could be reamed out with a drill. Not sure, whether this would work in practice.

_________________
Eberhard

Former chairman Arbeitskreis historischer Schiffbau e.V. (German Association for Shipbuilding History)

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Another lesson is to keep it simple when assembling.

This piece should have been printed vertically as Tom says, it's longer of course. But at 0,05 mm of layer it would come out very well, I rarely go below that. It doesn't really seem relevant with 4K. Maybe with the phrozen 4K printer which has a horizontal resolution of 0.037 mm, you can then tune the vertical to 0.037.

_________________
Pascal

•Battleship Bretagne 3D: https://vu.fr/FvCY
•SS Delphine 3D: https://vu.fr/NeuO
•SS Nomadic 3D: https://vu.fr/tAyL
•USS Nokomis 3D: https://vu.fr/kntC
•USS Pamanset 3D: https://vu.fr/jXGQ

Thanks for the suggestions.

I considered printing vertically (with the cylindrical part and the opening closest to the print platform) but thought there would be more support problems because the long parts of the arms would be hanging off from the cylindrical part.

Printing like I did I have lots of supports under the launcher rail.

I suppose I could print it with the cylindrical part pointing away from the print platform - hanging down in the printer. This would put a lot of short supports under the long launcher rail part and the cylinder should support itself.

****

The lock and key idea is more complicated than a simple pin and socket. But the two arms will have to be assembled by inserting them into the center piece from two sides and then locked together some way. After they are pushed together in place they will be rotated so the bottoms of the launcher rails are parallel before the glue sets. CA sets too fast to allow them to be positioned correctly. I will use epoxy, which I prefer in almost every instance over CA.

The idea behind the lock and key is that after the key is inserted into the slot the arm will be rotated until it contacts an internal stop. At this point the two rails should be parallel. Should be - but I am beginning to wonder how accurate the positions of the mating surfaces will be.

I think I can build a jig that will allow me to place the launcher rails on a flat surface after they are inserted into the center piece. That way they can be oriented correctly until the glue hardens.

****

I really don't understand the reluctance to print at less than 0.05 mm vertical resolution. for nearly horizontal surfaces like the blades of my propellers and the tops of the capstan and wildcat 0.01 mm produces almost perfectly smooth surfaces - the "jaggies" are so small that they are not visible to the naked eye. I need a 10x magnifier to see them, and then they are almost invisible.

If you use larger steps you get noticeable digitizing effects on nearly flat surfaces, like the decks of Pascal's tug boats. It is very visible in the photos.

The resolution of the LCD display does set a limit on the smoothness of the edges of slices. This is independent of the vertical step size. For example, on my Photon Mono with a 2560x1620 LCD the pixel dimension is 0.051 mm. This makes pretty noticeable lines/grooves on very thin vertical objects, like the tiny grids I was trying to print. It seems to be most visible on surfaces that are almost vertical, say 5 or 10 degrees from vertical. Regardless of the vertical step size, as the print progresses pixels at the edge of a vertical object turn on or off in 0.05 mm horizontal increments.

The best way to avoid this is to try to avoid surfaces that are nearly vertical. But of course you can't always do that. For example, any way you print a sphere you will get nearly vertical surfaces somewhere. The arms of the missile launcher are another example with rounded surfaces everywhere. But if you are printing objects with only perfectly vertical and horizontal surfaces, like cylinders with the axis vertical or cubes, the vertical resolution isn't as important.

If I always printed at 0.05 mm I could never get the smooth surfaces on almost horizontal surfaces. I would be condemning every object to horizontal and vertical 0.05 mm resolution.

_________________
A collision at sea will ruin your entire day. Aristotle

It sounds counterintuitive, but are these edge phenomena the reason, why people print objects obliquely ? They seem to need more supports, but then obviously results in fewer digitalisation artefacts.

_________________
Eberhard

Former chairman Arbeitskreis historischer Schiffbau e.V. (German Association for Shipbuilding History)

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

I suspect some people do this because it is "the thing to do." However, in some cases it can make a difference.

The smoothest printed surfaces are flat and either vertical or horizontal. With vertical surfaces you avoid the jaggies caused by the finite dimension of the individual pixels. Theoretically the same set of pixels will be used from the top to bottom with no changes. With horizontal surfaces you avoid the stepping effect of having some pixels on part of the time and the off the rest of the time. As the vertical step changes all pixels for the surface change at the same time.

The worst effects are when the surface is almost horizontal or vertical. As the print proceeds either the pixels or the stepping changes while the surface is printing.

For angled surfaces there will always be pixel/step changes, producing jaggies. The smoothest angled surfaces result when the angle of the surface from the horizontal is


where s = step size and p = pixel size

So if s = p the angle is 45 degrees. Say s = 0.05 mm and p = 0.05 mm.

If s = 0.01 mm and p = 0.05 mm than the angle is


For almost horizontal surfaces a step size smaller than the pixel dimension gives a smoother surface.

However, the same is not true for almost vertical surfaces because you cannot make the pixel dimension smaller. In this case using the same step size and pixel dimension is about the best you can do.

So, depending upon the object being printed it may be possible to select an angle that avoids the worst case situation.

****

There are other options in Chitubox (and probably other slicer programs) for producing a smoother surface on some printers, at least with laser/LCD printers. in the "Settings/Advanced" tab there is an "Anti-aliasing level" setting that produced pixels along the edge of a slice that are different shades of gray instead of just black or white. In theory this will produce a reduced exposure that will not cure as much resin in the jaggie. Likewise the"Grey level" setting also influences the saturation of a pixel along the slice edge.

I don't know if my Anycubic Photon Mono printer can use these features, but since it just displays bitmap slice images the slicer program generates I suspect it can.

I have been experimenting with these settings. I just printed an object with a flat surface angled 37 degrees from the horizontal. In theory this would require a vertical step size 0.4 times the pixel dimension. With a 0.05 mm pixel a 0.02 mm step size should be optimal. However, I used a 0.03 mm vertical step and Anti-alising level = 2 and Grey level = 4.

The resulting surface has no visible jaggies, and even some rounded surfaces with near horizontal and near vertical angles came out very smooth. There still is some vertical grooving or lines visible with magnification on vertical cylindrical surfaces due to the 0.05 mm pixel dimensions. but there isn't anything that can be done to eliminate that in the print. But once the part is painted I doubt these will be noticeable, even under magnification.

Phil

_________________
A collision at sea will ruin your entire day. Aristotle

Such practical and sometime counter-intuitive observations are really valuable. I have no immediate plans to get into 3D-printing, but have followed the developments over the past 15 years. Such observations help to decide, whether the 3D-printing would be really helpful for a particuar project.

_________________
Eberhard

Former chairman Arbeitskreis historischer Schiffbau e.V. (German Association for Shipbuilding History)

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

I know next to nothing about 3D printing, except the excellent results from pieces I occasionally buy from Model Monkey, and Reedoak. Of the current generation of 3D printers, what quality results could I expect from a $150 - $300 machine? Could I reasonably expect the results of the above mentioned companies?

With good design and technique (learned) one can far exceed the commercial products. At the edges of what is possible, a failure rate is expected and you can afford to make several to get a good one, with sequential fine tuning. A commercial operator has to dial things back a little to get a reliable product.

Perhaps the biggest advantage is to produce ship/situation specific parts that would not be common enough to be commercially viable.

Certainly a mathematical understanding of the process of constructing layers and optimizing orientation and settings is important. To some extent the number of useable significant figures is controlled by resin and settings. Some resins will provide what in the chemical process of film development we call accuatence (SP?) which defines the sharpness of edges. In photoshop we call this tool "Unsharp mask". A goopy resin with a longer exposure will produce a smoother surface and less sharp detail. There is always a little "measure with a micrometer and cut with a chainsaw" effect. Removal technologies as opposed to additive will be more accurate. 3D printing is a bit more like sand casting as opposed to computer controlled machining. Each has it's limitations.

Tom

... and can be combined. The UV-hardening resins (acrylics) are hard enough to be machined. I would consider some printed parts like sand-cast machine parts, where mechanically important surfaces are machined exact specifications after casting.

Think of steam-engine or winch parts and the likes. The quality and dimensions of those surfaces that would be milled or turned on the prototype could be improved by exactly the same techniques.

_________________
Eberhard

Former chairman Arbeitskreis historischer Schiffbau e.V. (German Association for Shipbuilding History)

--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

drastic plastic,

I have an entry level LCD printer - Anycubic Photon Mono (2560x1620 pixels - and it is producing far better results than I expected. It is fine for 1:96 scale modeling, and I think it could be used for smaller scales. The pixel size is 0.05 mm x 0.05 mm and the vertical step minimum is 0.01 mm. Not bad, but I do see a few jaggies under 10 X magnification. There are better printers available. Some 4K (~4000 x 3000 pixels) have come on the market and they will have smaller pixels and finer resolution.

I have been watching the development of 3D printers for decades but I had not decided to buy one. But The Admiral's youngest son gave me the printer for Christmas - totally unexpected. He explained that he had a 3D printer and gave me one so I could learn about it and be his tech support!

So now I am learning, and there is a learning curve. But even as a novice I have been getting good results unless I push it to the limits, which I have done several times.

Although I can print things as small diameter as 0.008 inch, they are far too fragile for practical use, at least with the basic resin I am using. This is important, for the US Navy railings are 0.010" to 0.015" diameter at 1:96 scale. Some are especially complex and I really would like to print them instead of make them from wire! But this may be the limit of the current technology. I am planning on using brass wire for hand grabs, steps and such because it is far more durable.

The real learning curve comes with figuring out how to use a 3D CAD program and then learning how to design for printing.

Phil

_________________
A collision at sea will ruin your entire day. Aristotle