Pennsy Centipedes

[kelticsylk]

Yikes!
Maybe I should just see if I can get a shell made first and work up to the other stuff. Baby steps, so to speak.

Marc mentioned creating Baldwin DR-4-4-15's using ER shark mechanism (can you use Life-Like FA if sharks are not available?) and a cast shell. This is kinda appropriate because the first generation of Pennsy sharks were DR-4-4-15's.

We could probably use a more accurate Shark for that matter. If my 3D models can be used for 3D printing I could whip up just about anything.  Aside from the Baldwin series...

Not sure how to do this thing yet, but it sure would be nice if it worked.

Frank

[GimpLizard]

Frank, I'm not at all familiar with the TSM format. Or the software that uses it. I do have experience with 3D solid modeling (Solidworks & Keycreator) and STL format used in rapid prototyping. I managed to drop a couple of images into photobucket, and then upload them here, to illustrate what I'm talking about.

This first imade is of a simple, solid model (I know, I know... the slope of the nose is all wrong. ) that I created in CAD. (Keycreator) Notice that there are flat & curved surfaces. All of which can be manipulated, or edited, to change the shape and/or size of the model.



Image 2 is of the same model, exported as an STL model. Notice how the flat & curfed surfaces are made up of multiple triangles. To the best of my knowledge, CAD software will not allow you to manipulate those triangles. (And, at the micro level, these triangles leave a faceted surface.)



Becasue of this, when I import an STL model, I cannot edit it. The best I can do is create a new model over it, using the original as a tempate. By drawing new lines/arcs, and using selected points of those triangles as a sort of jig-fixture. (That might not be the best way of explaining it, but i hope it makes sense.) And that doesn't guarantee the finish model is 100% accurate. If you pick the "wrong" triangle points, you could be off a hair or two. (Or more?)  Then again, maybe it's not that big a deal.

[pnolan48]

I am well behind the leading edge of 3D design--in fact, I'm probably at the trailing edge.

For the body shell, it seems to me that we are talking about a box, without a bottom, that has a difficult to model nose attached to the front. So the sides are basically slabs with some details such as grilles, doors, and other miscellaneous items. The top, or roof, has a camber (or curve) with more details upon it. The back is also basically a slab with details, whose top has to match the camber of the roof. These four parts could be easily mastered in styrene (with details perhaps from commercial suppliers), rendered in resin, and joined together. I think 2D drawings and processes would be sufficient. I think there's no reason to cannibalize other shells, given cheap digital 2D cutters.

Now, the nose. If we are having such difficulty rendering it in 3D, why not just make a master by hand, and then mold and cast it? And then join it to the detailed box from above? While less satisfying to those who enjoy the 3D modeling challenge, this approach might be more economical to actually producing a shell in small numbers. What is the total market for a Centipede these days? I'd guess less than 100 pieces.

That's the way I will eventually do the shell, if I ever get done with the ships.

For the mechanism, which I prototyped a number of years ago (and can't find the images), it's not too difficult to build a box frame from brass "L" channel, and mount the DD40X trucks in the middle, using their built-in swivel tolerances. The motors go at the far ends, and can swivel with the trucks (or not). It's basically the reverse of the way they were mounted in the DD40X, with the motors mounted toward the center. The "ball with spikes" drive shafts can accommodate quite a deflection.

I never got beyond the prototype because all my D-D trucks had the old Bachmann cracked axle gear problem on at least one axle per truck, and I never got into trying to find the good ones and switching them out for the bad ones. It took only one per truck to mess things up.

[kelticsylk]

First off, I want to thank you guys for your responses. I'm learning a lot I did not know and finding ways to improve what I do know.

GimpLizard,
TSM (Train Sim Modeler) was developed from a similar program for Flight Simulator. It does allow me to manipulate the points on the triangles. In fact I use the point mode more than any other function. It has a tool for manipulating curves that I haven' seen in the other programs I have tried. That one function is the secret to creating the complex shapes that appear on all those first generation streamliners.

This is the nose of the Baldwin DR-4-4-15

The Union Pacific M10000

I have actually seen the triangles you're talking about. When you screw up the way you select the points to create a polygon on a model, you will see one of the triangles in the poly "disappear" which means it's "normal" is not facing the wrong direction.

pnolan48,
As I mentioned above, it isn't too hard for me to make the contours of cab unit noses. My one "claim to fame" modeling for Train Simulator was the fact that I could create shapes other folks had trouble with, like this DL109...

It's easier for me to do these shapes on the computer. I would never be able to make these using real wood or plastic. It's one of the reasons I was looking for a way to feed these drawing files into an NC machine. When I found out about 3D printing, I was pretty excited at first. But at the time I didn't know there was a way to convert my files to the format required by the printer, so I shelved the idea for awhile. When I started this thread I was going to try and create a "centipede" by hacking up an E-unit and using a GG1 mechanism. I have even thought to cast resin panels just as you describe and assembling them, but the cab would have still been a problem. Now I know I can use the DD40 trucks and maybe make the 3D model I have into a real shell.

By the way, Bachmann has replacements for those broken gears. I had to do the same thing for my PCC trolleys. I had to make sure I specified the new replacements, otherwise they send the same wheel sets that cause the problem. I believe Northwest Shortline sells replacement wheelsets for the DD40.

I'm sort of following your description of the mechanism but I have never seen the insides of a DD40. There are 2 motors?

As for the demand for centipedes. I'm really thinking of my own desire to have at least one pair of these monsters. The Allegheny Eastern models the Pennsy when the centipedes were still assigned to passenger trains, which is what they were designed for.

Frank

[peteski]

To me, 3D computer modeling also seems easier than making master patterns by hand.  Especially things like noses which need to be symmetrical. All you have to do is to design one half then mirror it to make the other half perfectly symmetrical!  Also all the repeating elements are easy. You need to create one then use a copy of it wherever it is needed.  Editing is also easier on the computer vs. a piece of clay or wood.

[kelticsylk]

I found out that 3D Canvas can actually export .stl files, so I don't really need 3D Exploration to convert, only to see what the .stl file looks like.

I went back into TSM and cleaned up the model a bit. There were some extra parts that are needed for a shell. I then re-scaled the whole model  to 1:160. I imported it into 3D Canvas. It didn't show up on the screen because it was so small, but 3D Canvas was still able to export it as a .stl file. This is a rendering of the the.stl file viewed with 3D Exploration...


You can't see it but when 3d Canvas exports the .stl file it automatically stands it on end. If I've been reading some of threads correctly this is actually how you want the model created. Solves some of the "zagger" issues on curved surfaces.

I'm going to try and send this off to a shop that uses the new Perfactory printer. These printers just appeared on the market in May or so and supposedly have much better resolution. I'm hoping they will at least tell me if they can produce a shell from this file and what to correct if they can't.

Frank

[robert3985]

You can't see it but when 3d Canvas exports the .stl file it automatically stands it on end. If I've been reading some of threads correctly this is actually how you want the model created. Solves some of the "zagger" issues on curved surfaces.

I'm going to try and send this off to a shop that uses the new Perfactory printer. These printers just appeared on the market in May or so and supposedly have much better resolution. I'm hoping they will at least tell me if they can produce a shell from this file and what to correct if they can't.

Frank

The Perfactory process doesn't create "zagger" issues, so the only issue with orientation is how to most efficiently fill up the build envelope.

Much increased resolution and no stepping are two of the most obvious advantages of Perfactory printers.  Increased cost IMO, is the only detractor. 

Perfactory printers have different resolution per machine, and per materials used.  Make sure you have a nice, long talk with your potential Perfactory shop about these parameters.  Also, make it clear that you're not interested in the "Rapid" part of the the shop's Rapid Prototyping moniker.  You might get a substantial price break.

[kelticsylk]

Bob,
Sent the file as an attachment to Metropolitan Works and asked them if they thought it would work and what I should change if it didn't. I'm waiting to hear back from them.

Frank

[Lemosteam]

Does any of the shape have volume (thickness)?  They may not be able to presume how thick to make the shell walls...

I'm interested to hear their response to you.

[wcfn100]

The machines are called "Perfactory" machines and one of our members (I can't remember who right off hand) found a company in New Zealand which would do his cabooses (4 of 'em) for $140 bucks or so...which is not too bad actually, especially considering the quality.

Well it's not that I found a company, it's Mark4Design who started doing shells like this years ago that was nice (awesome) enough to help me with my first pieces.   And while the unit costs on 4 cabooses may reasonable (I'm hoping to send those to Mark in a few weeks) , what you're leaving out is if I had to do a redesign, it's another $140.  As opposed to Shapeways which is $15.  Shapeways is great way to prototype your prototype - so to speak.

Much increased resolution and no stepping are two of the most obvious advantages of Perfactory printers.  Increased cost IMO, is the only detractor. 

There are also build/design issues with Perfactory that aren't present with Shapeways.  As I had talked about on the TB thread, my cabooses would need to be designed with separate cupolas where the models with Shapeways could go as one piece.  This isn't a big deal with some software, but if you are using software which doesn't allow for multiple pieces to be drawn at the same time, there is a chance for error.  And when errors could cost you $140, it's can be something to be concerned about.

Jason

[kelticsylk]

Does any of the shape have volume (thickness)?  They may not be able to presume how thick to make the shell walls...

I'm interested to hear their response to you.

I should have thought of that. I have interior walls but they are in same planes as the exterior. I'll have to add thickness.

Back to the 3D drawing board...

I did hear back from them. They referred me to an RP tech. The are also based in the UK, something I didn't read when I looked at there website.

Frank

[kelticsylk]

Latest update...
The techs looked at my file.
A) No wall thickness, which I am currently correcting.
B) Model came out to 14mm long! Thought I had scaled it properly. Once I get the walls fixed I'll have to figure out how to get it to N scale. Just reducing it by 1:160 does not a N scale centipede make. 

Frank Musick

[GimpLizard]

Frank

Frank, that looks like a "surface" or "sheet" model. Meaning those panels don't actually have any thickness to them. If so, that means there is no "volume" to it. And while I may be mistaken about this, I think volume is needed for the 3D printers. I'd be interested in what the printers say.

[peteski]

GimpLizard, Frank is on it - see his post, right above yours.

Frank, if you are reducing by a percentage then if your drawing is 1:1 scale (life-size), you need to reduce it to 0.625% to make it N scale size (i think...) 

[kelticsylk]

GimpLizard, Frank is on it - see his post, right above yours.

Frank, if you are reducing by a percentage then if your drawing is 1:1 scale (life-size), you need to reduce it to 0.625% to make it N scale size (i think...) 

I came up with the scale by dividing 1 by 160. I got 0.00625% which is obviously total;y wrong. 0.625% may be correct. I'll try it.

Frank