Thread Tools
Oct 14, 2015, 12:36 AM
Registered User
cn0rris's Avatar
Thread OP
Discussion

Critique my DLG Wing Mold Model


I am teaching myself CAD and (soon, hopefully) CAM, with the eventual goal of cutting my own molds for DLG wings and tails. Tail molds were pretty easy using some of the great directions Tom Siler gave in another thread, but I struggled quite a bit making full wing mold models. I'm new to this, and I'm coming up the very steep learning curve of Rhino 3D and having a blast.

With wings, having multiple airfoils, plus twist does evil things to the parting plane, and I found I had to learn a whole new set of Rhino skills in order to model them correctly.

I think I may have cracked it now, but since I've never done this before, and I still have yet to cut molds, I'm looking for any advice you more experienced CAD guys may have for the new guy about how to design good wing molds.

Hopefully you can help me skip some of the mistakes you probably made when you were learning this stuff.

About this wing:

The wing is an absolutely stock Synergy II wing, designed by Gerald Taylor (a.k.a. "G_T"), the 20141225i version (the latest as of this writing). Full details here. The wing planform and geometry are exactly as Gerald specified to the best of my ability to model it.

Tom Siler's Rhino project of the Synergy II wing (here in the thread above) was extremely helpful and essentially got me over the hump figuring out wing modeling, but since my goal was to teach myself CAD and mold design, I chose not to just use his shapes to construct my model. His model is full of interesting hints and techniques that I'm still learning from, and I recommend checking it out if anyone is interested.

The model's too big to upload here, so here's a link to the file on DropBox:

https://www.dropbox.com/s/xyeq4s0po2...i%205.zip?dl=0

Any and all comments and suggestions are welcome - especially advice that will save me headaches in the future.

EDIT: My latest version is down at post #88. This one has a few problems that I since corrected. - Chuck

Chuck
Last edited by cn0rris; Nov 29, 2015 at 08:36 PM.
Sign up now
to remove ads between posts
Oct 14, 2015, 07:35 AM
Barney Fife, Vigilante
tom43004's Avatar
Keying holes.... don't forget the keying holes so you get good top to bottom alignment of the molds.

Sanding the bump with a piece of hard material and 220 grit 3m77d to one side is a piece of cake. Milling it is an option for tails (where the parting plane is flat) but isn't much of a feasible option for a washed-out parting plane due to re-alignment hassles on the machine.

Congrats. It would appear that you've overcome many of the large challenges to get this far.

A few things I typically do but am not sure I've shared before:

I always split the airfoil into three curves after the point cloud import. One for the top surface including the leading edge point, one for the bottom surface including the leading edge point, and one for the trailing edge. Group them for sizing / positioning etc then ungroup them for lofting.

OPTIONAL - I like to do a "rebuild" on the top and bottom curves separately after importing the foils and before lofting. This makes sure that all curves have exactly the same number of points so the lofting is clean. I use 600 points and 1 degree. This helps a lot if you're working with solids and isn't quite as critical if you're working with surfaces.

If you are going to draw the wing as a solid then Boolean "difference" it from the mold stock, loft it with top and bottom surfaces "joined" then add the trailing edge so your lofting doesn't pick up artifacts from the sharp corner at the trailing edge. It helps keep you sane while trying to "align curves" correctly.

No LE channel is necessary unless you're really heavy handed with the splooge. It doesn't hurt anything, just adds steps to the milling process and makes the molds a little less sturdy.

Save your work about 8x as often as you think you should in Rhino. Surprises happen and they're generally NOT good surprises.

Draw your resin channel squareish (as you have) but mill them with a radiused endmill. This makes the removal of the wing much more of a non-event and you'll be less tempted to use sharp stuff to get excess epoxy out of the channel.
Oct 14, 2015, 11:20 AM
Registered User
cn0rris's Avatar
Thread OP
Several good nuggets here - this is great.

1) Keying holes - I have them in earlier versions, but just didn't include them in this version for some reason. I see you're using 4 holes - I presume that's enough to make sure Corian-type material aligns well?

2) Splitting into 3 curves - this should help me. I was splitting into two, with the trailing edge cut in two with half going with each side. I was getting some odd artifacts with the little bit of TE, and this should eliminate it.

3) Rebuild: I had noticed you were converting the airfoil from a polyline into a curve, but I ended up doing a "Curve Through Polyline Vertices" command instead. Either I don't know how to use Rebuild, or it's got bugs in the Mac version. My rebuilt curves kept coming out very rough - like it's ignoring the point count value.

4) Solids: I could never get the boolean difference approach to work correctly - your comments about modeling the top and bottom surfaces separately were the main "trick" that allowed me to finally get workable models. Booleans worked fine for tail surfaces with a flat parting plane, but I although I could subtract the airfoil solid from a block to make a mold blank, I could never get it to split along the parting plane for some reason.

5) TE channel: good advice - I was tempted to model them as half-elipses, but I figured that would just increase the cutting time without a lot of benefit. Curves are definitely better for durability.

You're right about the surprises in Rhino, but for me they've been mostly been learning my way through the program and having to restart from earlier steps to try a different approach. I've taken to squirreling away the basic curves and saved pieces in a "Construction" layer so I can rewind more easily.

But I'm still getting lots of practice importing airfoils and setting up the basic geometry when I give up on an attempted approach and start from scratch. It's all good, though - my main goal for now is to learn Rhino and how to produce workable mold models.

A few questions:

* Does modeling at higher precision, or with more than necessary control points slow anything down at the cutting phase? Should I be trying to get to "just enough" control points for any reason? I'm tending to over-specify the precision during my learning phase because I don't really know which what constitutes "good enough" and what will affect the final cutting results. I'm pretty sure it makes my model files a lot bigger, but I have plenty of disk and RAM, so that's not hurting me at the moment.

* My top mold gets pretty thin at the thickest part of the airfoil - do I need to worry about the stiffness in a material like Corian? It looks like it comes in 1/2" thick panels, and I was careful to leave as much material there as possible (the 1mm LE bump ends up being the high point which limits the thickness at the root). But I could laminate two pieces together, or use some other reinforcement. Or I could reduce the "bump" to 1/2mm, and gain some more thickness at the root.

* I would like to have molds that I can eventually join together to make one-piece wings. How do you handle the joints and supports to maintain the dihedral angle?

Lots of questions - at some point I just need to cut some molds and then I'll have a lot better intuition about how to do things right. But right now, every pointer and suggestion is likely saving me many hours of time (and materials cost) by avoiding mistakes.

Thanks for your help!

Chuck
Last edited by cn0rris; Oct 14, 2015 at 11:35 AM.
Oct 14, 2015, 12:01 PM
Barney Fife, Vigilante
tom43004's Avatar
Try these:

Quote:
Originally Posted by cn0rris
Several good nuggets here - this is great.


A few questions:

* Does modeling at higher precision, or with more than necessary control points slow anything down at the cutting phase? Should I be trying to get to "just enough" control points for any reason? I'm tending to over-specify the precision during my learning phase because I don't really know which what constitutes "good enough" and what will affect the final cutting results. I'm pretty sure it makes my model files a lot bigger, but I have plenty of disk and RAM, so that's not hurting me at the moment.

I always have the precision turned up and the display meshes turned down. This makes drawing and manipulating things accurate but quick. I usually turn everything up for final checks before generating my code.

* My top mold gets pretty thin at the thickest part of the airfoil - do I need to worry about the stiffness in a material like Corian? It looks like it comes in 1/2" thick panels, and I was careful to leave as much material there as possible (the 1mm LE bump ends up being the high point which limits the thickness at the root). But I could laminate two pieces together, or use some other reinforcement. Or I could reduce the "bump" to 1/2mm, and gain some more thickness at the root.

It's fine for drawing but you will want to laminate a couple or three pieces of corian together, especially if you're doing solid core. You can do it before or after machining. I usually do it before to avoid any distortion in the clamping of the mold.

* I would like to have molds that I can eventually join together to make one-piece wings. How do you handle the joints and supports to maintain the dihedral angle?

Make your molds a few extra mm in span and tablesaw the correct dihedral angle in. Join with epoxy or Weldon #16 and sand / polish the center. Google "tiltbox" and buy one. A good inclinometer is worth it's weight in gold. The one from HorribleFreight is junk.

Lots of questions - at some point I just need to cut some molds and then I'll have a lot better intuition about how to do things right. But right now, every pointer and suggestion is likely saving me many hours of time (and materials cost) by avoiding mistakes.

Thanks for your help!

Chuck
Oct 15, 2015, 09:47 AM
Aurora Builder
The photos look good! I can't really read native .3dm files outside my CAM program and manipulating them is a bit tricky.

Tom's points are all spot on. Any chance you can upload .STEP or .IGES or .XT files of the 4 molds halves and the two wing halves? If you only have the right wing half I can mirror that. If they are clean this will save me a fair amount of work and drive me to machining some top MDF molds and cores. I could re-pay your CAD efforts with the same setup if you like.
Oct 15, 2015, 11:47 AM
Registered User
Excellent comments. Thank you for sharing the experience. Certainly sparing us from rework and frustration.
How the washout is done in the model? Should we create a polyline around the edges and create a flat surface using rail command and use the surface as the parting plane on the two half molds?
Oct 15, 2015, 12:03 PM
Registered User
cn0rris's Avatar
Thread OP
I'm happy to share my work. Based on Tom's comments, I'm going to rebuild my model and tweak a few things. It'll still be the same geometry, but things like the trailing edge will hopefully be cleaner. Plus, I've been examining my model more closely and have found a few small issues that probably wouldn't affect tool paths, but I'd still like to fix them up.

When I've rebuilt it, I'm happy to share any format that Rhino can export.

Chuck


Quote:
Originally Posted by samc99us
The photos look good! I can't really read native .3dm files outside my CAM program and manipulating them is a bit tricky.

Tom's points are all spot on. Any chance you can upload .STEP or .IGES or .XT files of the 4 molds halves and the two wing halves? If you only have the right wing half I can mirror that. If they are clean this will save me a fair amount of work and drive me to machining some top MDF molds and cores. I could re-pay your CAD efforts with the same setup if you like.
Oct 15, 2015, 12:16 PM
Barney Fife, Vigilante
tom43004's Avatar
Quote:
Originally Posted by eghelman
Excellent comments. Thank you for sharing the experience. Certainly sparing us from rework and frustration.
How the washout is done in the model? Should we create a polyline around the edges and create a flat surface using rail command and use the surface as the parting plane on the two half molds?
For years I didn't share how I did this. It's not rocket science but it requires some thought.

Model the wing with no washout and create a plane through the leading and trailing edges, extending beyond the root and tip. Select the wing and the plane and do a "twist" with a starting point at the quarter chord at the wing root, and ending point at the tip perpendicular to the root chord. Twist by the washout amount. If you can't tell if it twisted the correct direction, put 10 degrees or -10 in and note the direction, then undo and twist again with the correct angle. The plane (parting plane) is now twisted along with the wing surfaces or solid. Done. Now you can use parts of the parting plane to find intersections with the wing etc to draw the rest of your tool.

Some twist the wing around the hingeline. That produces a straighter hingeline mechanically but it does some other things with the parting plane that may or may not be desirable. Also XFLR models the washout based on quarter chord, which is why I twist there.
Oct 15, 2015, 01:35 PM
Registered User
jbarnhardt's Avatar
Tom - thanks for sharing everything you do. One observation on the technique you describe for modeling washout: this will result in a linear washout distribution across the span unless I'm missing something. That is probably just fine when it comes to observable flight characteristics, but if one wants to exactly replicate the model in XFLR I think that technique will be insufficient. It seems common to have the washout be progressively more extreme as you approach the tips (i.e. there are more degrees of washout applied per mm of span). In that case you'd either need to do segmented twists between different stations (which could introduce discontinuities at the station breaks I think) or rotate the individual foils before lofting and then create the parting plane that incorporates the wing twist in a manual and much less elegant way than you've described. That's how I've been doing it as my high-AR Synergy wing has progressive washout. Tradeoffs...

Chuck, thanks very much for publishing the details of your work here - it is great to read through and compare to my processes (which also owes a huge debt to everything Tom has shared over the years). You have made very rapid progress! I don't know if you recall but we met at IHLGF the year before last and talked quite a bit about building techniques. At the time I was bagging wings and you were hand-crafting pod molds IIRC. It's amazing what a couple years of regular building and refinement will do!

An additional comment on earlier posts: with respect to the precision of your modeling in Rhino - the precision of the drawing itself should be more than accurate enough for anything you're going to do. As long as you're specifying enough control points in your curves and surfaces to reflect the geometry of the wing you're drawing you should be fine there. The less obvious part has to do with your Rhino rendering mesh, which controls the display of your surfaces in the program, but in many cases also will determine the resolution of the machine paths your CAM program generates. I use MadCAM (a Rhino-integrated CAM plugin) and the render mesh settings directly affect the precision/fidelity of the toolpaths. I tend to just leave the render mesh settings high which slows down screen update speed a little but makes sure I never generate toolpaths at lower resolution. I doubt this applies to every possible CAM solution out there (particularly those that are external to Rhino) but it's very important to understand how your chosen CAD and CAM systems work together when it comes to toolpath fidelity to your drawing. Tom, does VisualMill work similarly to MadCAM or is it independent of the render mesh settings?

Keep posting updates, Chuck, great stuff. Thanks,

-John
Oct 15, 2015, 02:16 PM
Barney Fife, Vigilante
tom43004's Avatar
John,

For non-linear washout, you can simply twist in sections by starting the section at the quarter chord of the root side of that panel and then twisting only the panel. Where you end the twist makes a big difference in the end result.

There are no irregularities between the stations when twisting in segments that I've found... as a matter of fact I just did one the other day for a F3J wing with non-linear washout and it worked perfectly. *added a quick pic twisting a rectangular surface the same amount (45 degrees) over two different distances to check accuracy. *

About accuracy... I would add to John's comments. Draw more accurately than anything you'll need EVER. It's amazing how many times I've shared a drawing with someone who had a more accurate machine than mine and they required zero rework. Lofting after twisting the airfoils can cause weirdness... twisting after lofting is the only way I could get accurate results.

...and yes, rendering meshes affect all of the plugin CAM software packages AFAIK.
Last edited by tom43004; Oct 15, 2015 at 02:27 PM. Reason: Added a Picture
Oct 15, 2015, 02:16 PM
Registered User
cn0rris's Avatar
Thread OP
I ended up doing things a little differently, and I’m curious to hear whether my way seems like an ok alternative. I’ll try Tom's way soon to figure out which way I like best.

What I did was to put the airfoils and rotate in the washout rotation given in the XFLR5 file Gerald provided. Then I positioned them as specified (LE offset, span position). This gets everything mostly correct, but leaves the hinge line not quite aligned. So I snap a line between the hinge line point on the root foil and the tip foil, and align all the intermediate foils leaving a perfectly straight hinge line.

Then, I connect up all the LE points and TE points with a polyline. This defines the wing profile shape.

Then, to make the parting plane, I sweep a line around that profile shape using Sweep1.
Oct 15, 2015, 03:16 PM
Registered User
jbarnhardt's Avatar
Tom - got it, thanks for the additional explanation. I am definitely going to try this then the next time I draw a wing up, as I have encountered plenty of weird little artifacts and whatnot when modeling and especially when working with booleans. The twist approach and the separate treatment of the TE may go a long way toward fixing these annoyances.

-John

P.S. Allright, this is a rat hole that is likely totally irrelevant in practice so feel free to ignore, but I can't help thinking about it. With section-based twisting, you would have sequences of progressively higher-sloped straight lines if you looked at the TE from the rear, correct? In that case, I think you would "only" have G0 continuity (position) as compared to the theoretically more desirable G1 (tangent) or G2 (curvature) continuity at each station. So at some level there would be a kink in the wing surface that would be more pronounced the farther you were away from the center of rotation of the twist (LE or quarter chord or hingeline) - so likely most noticeable at the TE and on the parting plane (where it doesn't matter). With the very small rotation angles we're talking about I suspect this doesn't make enough difference to ever show up in practice but I'm still curious about it because when I've looked at your models in the past, Tom, they usually look perfectly continuous in a zebra stripe analysis and I wonder if I'm thinking about this incorrectly somehow. End rat hole. Please ignore if too pedantic.
Oct 15, 2015, 03:29 PM
Barney Fife, Vigilante
tom43004's Avatar
*rathole warning*

I have had some models in the past that you could see the twist breaks in the zebra pattern. I didn't post pictures of those Some of my models had washout adjusted to be more linear... and some (like NXT) even had the airfoils adjusted as well. A couple of them had no washout (Fr3ak-TS) at all so they were smooth.

IIRC the SuperFr3aK was adjusted to linear washout without changing airfoils. I'd have to go back and look.
Oct 15, 2015, 09:46 PM
Registered User
cn0rris's Avatar
Thread OP
Quote:
Originally Posted by jbarnhardt
Chuck, thanks very much for publishing the details of your work here - it is great to read through and compare to my processes (which also owes a huge debt to everything Tom has shared over the years). You have made very rapid progress! I don't know if you recall but we met at IHLGF the year before last and talked quite a bit about building techniques. At the time I was bagging wings and you were hand-crafting pod molds IIRC. It's amazing what a couple years of regular building and refinement will do!
Thanks - I do remember meeting you and being impressed with the planes you were building. If you're the John who is working with Amardeep, then I've seen some of your more recent handiwork with the high-aspect wings. You're making rapid progress too - your wings are looking really good and fly great.

My own efforts are because I've been frustrated with my wing bagging skills and am wanting to step up to something where the final shape requires less skill shaping wing cores by hand. There's something very satisfying about the idea that I can design very carefully on the computer and then produce a product that matches those designs with high fidelity.

But I still have a ways to go - now that I'm getting on top of mold design, I need to learn how to create tool paths and G-code, and then how to cut the molds themselves.

I'm enjoying the learning process, though.

Chuck
Oct 16, 2015, 11:24 AM
Registered User
cn0rris's Avatar
Thread OP
Quote:
Originally Posted by tom43004
John,
as a matter of fact I just did one the other day for a F3J wing with non-linear washout and it worked perfectly.
Is that a SynerJ wing by any chance?


Quick Reply
Message:

Thread Tools