3D printer reviews

[crackerjazz]

I used Soiidworks for the parts above but I would recommend you try learning Fusion 360 which is free for hobbyists.  One advantage of Fusion 360 is it's easier to do complex curves which  I still have trouble doing in Solidworks after so many years.  I myself haven't really used Fusion 360 but Dave and other guys in the forum use it and swear by it.  I did give it a try once but found myself always comparing functions, movements and shortcuts with Solidworks which made it confusing -- it was fighting me so I gave up -- I didn't even last an hour, lol.   It does have the same functions like extrudes, revolves, etc but it just felt like cooking at someone else's kitchen, heheh.  But for you who's starting fresh there won't be anything to compare functions with so you'll absorb more.

 

I'm down with helping you out with simpler designs -- just no complex surfaces like fuselage sections, lol.  Or let's tackle an issue you have one step at a time.  Some cool looking parts can be done pretty quickly but others, like Electrosoldier mentions, can be really labor-intensive. But do it the way you'd eat an elephant -- one bite at a time -- and you'll end up with pretty cool results.    I used to say learning new things is easy to do -- just dive right in --  but I don't say that anymore because I'm having the same issue with stuff that I need to learn.   Take Python for instance -- no matter how I much I force myself to learn it I'm not making much progress.  I can't seem to remember anything and I just keep starting over.  My excuse, though, is that Python isn't visual.  3D modeling is great for us scale modellers because you get to see what you're building which is definitely more fun than learning abstract concepts : )   And with the desire you're showing right now this seems like a great time to start.

Edited April 2, 2023 by crackerjazz

[ElectroSoldier]

1 hour ago, crackerjazz said:

   And with the desire you're showing right now this seems like a great time to start.

 

This is without a doubt the biggest and hardest step.

Starting.

 

If you really want to do it you will learn it, if you dont then you wont.

 

Personally I have no desire to learn to use any of these programs to make my own parts for my models. I have thus far been quite happy to buy some STL files and print some minis I wouldnt otherwise have been able to own.

[Hoops]

I personally started down this path only very recently, and have mostly been learning how to use the software to generate 3D models.

 

I went with Fusion 360, as it's free for personal use. To get a basic idea of how to use the software, I watched a number of YouTube tuitorial videos and that was enough to get me going. Whenever I come up with a situation I don't quite know how to handle yet, some googling has been sufficient to find me the answer.

 

I am not fast at it by any means, and some of my models are probably not up to professional standards. That being said, It has allowed me to create 3D models for things that would be very difficult or time consuming with "traditional" methods. Or, correspondingly things that I plan on using multiples of. By doing the work once in Fusion 360, I will eventually be able to print as many as I need. Examples below are the Sparrow 501 pylon for the F-16ADF, and the GE GPU-5/A 30mm Gun Pod:

 

 

 

 

 

I don't have a printer yet, but will get one eventually. In the meantime I can build my catalogue of 3D models and print them when I am ready.

 

Cheers,

Hoops

[crackerjazz]

Those parts look really good, Hoops!   Very inspiring : )    That seals the deal for me.  Serendip and myself will try to get together and learn a thing or two about this software.   Hey, you have to get a printer yourself and try those beautiful creations out.   I notice some features need to be "exaggerated" to be visible when printed.   You'll learn some good printing techniques as you go and see which areas and features need enhancing.    Some holes, for instance, tend to close up if less than .75mm in diameter.   Some engraved details tend to fill up if less than a certain depth, some rivets if too small tend to disappear, some protruding parts need to be thickened to print and prevent breakage, some clearances need to be met for parts to join -- things like those -- and you'll get to learn to modify your 3d models accordingly and make them print-ready.  That aspect too of creating supports takes some time to do properly.    But, yeah, those parts look really nice it would be great to see them printed out.  

[GeneK]

Great thread with outstanding comments (facts and opinions) from some real sharing experts.

 

Gene K

[serendip]

Hi all,

I'm curious what techiniques people use for ensuring the best possible fit of self designed and printed replacements.

 

For example if I need to design and print a new redome for an existing kit, how can I best make sure the fit is good?

 

Thanks,

Marc.

[crackerjazz]

For me I lay a piece of paper against the forward fuselage's mating surface (where nose is glued onto) and do a pencil rubbing (like coin or leaf art),  I then take a photo of the pencil rubbing and re-size it accordingly in Solidworks -- then I can trace over it.   Optionally you can take a head-on photo (which might be harder to do) of the mating surface and trace over that in the 3D modeling software.  For any inner tabs, lips or alignment rods you need to have some clearance.   A 2mm-diameter rod will not fit into a 2mm-diameter hole.  You'll need to make the rod smaller or enlarge the hole.  For the resin I'm using .18 mm (.09 on either side) is usually sufficient for a good fit -- neither tight nor loose.

 

Hey, Marc, I'm really enjoying the Saturday morning lessons we're doing together to force ourselves to learn Fusion360.   I started following along to some YT videos on my own as well.    Sometimes, though, I come across instances where my software's behaviour is different from what I'm seeing in the videos.   I know some of those videos are old and we've got the latest version set up but the change seems to be for the worse.  I've got some questions that I couldn't find answers anywere and I'm wondering if I should post my Fusion360-specific questions here or if it's better to start a new thread.   

Edited April 2, 2023 by crackerjazz

[crackerjazz]

I take back what I said about scanners.   Just saw this video about importing a mesh into Fusion 360.    Gives me hope about being able to scan that Hasegawa A-7E forward fuselage again.

 

https://www.youtube.com/watch?v=imGrla3b3Mo

 

 

[serendip]

On 4/2/2023 at 1:44 PM, crackerjazz said:

For me I lay a piece of paper against the forward fuselage's mating surface (where nose is glued onto) and do a pencil rubbing (like coin or leaf art),  I then take a photo of the pencil rubbing and re-size it accordingly in Solidworks -- then I can trace over it.   Optionally you can take a head-on photo (which might be harder to do) of the mating surface and trace over that in the 3D modeling software.  For any inner tabs, lips or alignment rods you need to have some clearance.   A 2mm-diameter rod will not fit into a 2mm-diameter hole.  You'll need to make the rod smaller or enlarge the hole.  For the resin I'm using .18 mm (.09 on either side) is usually sufficient for a good fit -- neither tight nor loose.

 

Hey, Marc, I'm really enjoying the Saturday morning lessons we're doing together to force ourselves to learn Fusion360.   I started following along to some YT videos on my own as well.    Sometimes, though, I come across instances where my software's behaviour is different from what I'm seeing in the videos.   I know some of those videos are old and we've got the latest version set up but the change seems to be for the worse.  I've got some questions that I couldn't find answers anywere and I'm wondering if I should post my Fusion360-specific questions here or if it's better to start a new thread.   

Hi Joe and apologies, missed this message.

Sessions are excellent, really enjoying and learning a lot.

 

Regarding your questions I'm happy to keep it here and maybe change the title of the thread - we're in good company here and I'd hate to lose the other posters.

Opinions welcome all,

Thanks,

 

Marc.


 

[MoFo]

16 hours ago, crackerjazz said:

I take back what I said about scanners.   Just saw this video about importing a mesh into Fusion 360.    Gives me hope about being able to scan that Hasegawa A-7E forward fuselage again.

 

https://www.youtube.com/watch?v=imGrla3b3Mo

 

 

Note, though, that the scan... sucks.  None of the holes are round, smooth surfaces aren't, and the raw scan data is basically unusable - he's just using it for gross measurements to completely rebuild the part in CAD.

[crackerjazz]

Hmmm, you're right.  I'm really amazed at how they're able to do something like that, though  --- just going around something with a handheld device and re-creating it on a computer screen.    Would be awesome to see an A-7E scan : )

 

Hey, Marc : )     I was going to post a question about something I was having trouble with but I've got it sorted out.    Long-time Fusion 360 users might find this amusing to see 😀  but let me post it anyway.       I'm really looking at everything from a Solidworks user's perspective.   

 

The other week this was what was giving me trouble.  Let's say you want to model this simple shape:

 

  

 

 

I drew the base:

 

 

 

Drew the middle triangle.    The lines snap nicely onto the edges and corners of the base:

 

 

 

 

 

 

I was trying to do the same in Fusion 360 but the lines wouldn't snap onto the edges and corners of the base like it doesn't see it.

 

  

 

So that was what I was trying to figure out for a week until I found out it was this guy:

 

 

I have another Fusion 360 headscratcher  --- let me take pictures of it.

[ElectroSoldier]

55 minutes ago, MoFo said:

Note, though, that the scan... sucks.  None of the holes are round, smooth surfaces aren't, and the raw scan data is basically unusable - he's just using it for gross measurements to completely rebuild the part in CAD.

Yes there have been several videos on youtube over recent months that show the considerable advances made in 3D scanning.
Its not prefectly possible to scan a part and print it. However the scanners are a little pricy. 

[crackerjazz]

I've been trying to prepare for those instances where you build something and want to split them apart to build the individual model parts.    I know I should really be building parts in a way that they stay as separate components for printing but there are just some instances where you want to break components down further.

 

Solidworks has a pretty cool way of splitting parts and you're able to use faces, planes, extruded surfaces,etc.  Let's say you have this single body to break down:

 

 

 

I can do a split and it allows me to use any face or a combination of faces.  In this case I know I can break the parts apart using just 2 faces:

 

 

 

Solidworks breaks it down to the various parts using the 2 faces you created.  And you can go ahead and select those you want to separate.  In this case I can separate the box but don't want to consume/delete it.

 

 

 

and now I have my 2 parts:

 

 

 

 

I was trying the same in Fusion 360:

 

 

 

And was trying to use the same faces I used in Solidworks for splitting:

 

  

 

 

And it's erroring out:

  

 

 

So I try to do it another way by creating planes on the faces and trying to use those to do the split:

 

 

 

And still got an error:

 

  

 

 

In the end I had to split one way using one face:

 

 

Then the other way using the 2nd face:

  

 

And had to combine some of the split parts to get the 2 parts I need:

 

  

 

 

 

I can also try sketching the edge of the box and use it for splitting.  In Solidworks, for parts that are more complex I can create surfaces and stitch them together and use that for splitting.  Not sure if I can do that in Fusion 360 -- will have to explore further.

 

Here are the differences I've learned so far between Solidworks and Fusion 360.   Will try to add to this list as I go.

1.  You can automatically fully define sketches in Solidworks.  You cannot do this in Fusion 360.
2.  You can extrude faces in Fusion 360.   This cannot be done in Solidworks (you have to have a sketch or surface first).
3.  Feature tree can be searched (if you changed the names for example) in Solidworks.    In Fusion 360 you can rename timeline items but 
    there is no search function to search through them -- you have to hover over each item and see what's highlighted on the model.
4.  In Fusion 360 if you extrude a sketch but later on decide to exclude parts of it by turning a section into construction lines, your extruded shape
     stays the same -- you have to delete the unwanted portion of the sketch. In Solidworks, any geometries made up of construction lines cannot be extruded.
5.  For sweeps -- in Solidworks the Profile can only have one orientation. In Fusion 360 you can choose for the profile to stay perpendicular or parallel to the path.
6.  For wood trim along a box's edge for example, you can use sweep in Fusion 360 and the 45 degree mitre corners are automatically done for you. 

     This is not automatic in Solidworks.

Edited April 8, 2023 by crackerjazz

[Inquisitor]

8 hours ago, MoFo said:

Note, though, that the scan... sucks.  None of the holes are round, smooth surfaces aren't, and the raw scan data is basically unusable - he's just using it for gross measurements to completely rebuild the part in CAD.

And that's pretty much what I've seen all model companies do with their scan data of the real planes. Just a gross outline and measurements to trace the sketches/profiles/cross-sections to rebuild the plane in CAD 😉

[Inquisitor]

7 hours ago, crackerjazz said:

I can also try sketching the edge of the box and use it for splitting. 

The sketch does work for splitting bodies, either you'd already have the sketch you used to extrude those shapes or simply project the edges of the box.

As you pointed out there are many differences between how solidworks and fusion 360 handle certain things.

Also one mayor difference: Fusion 360 is free for personal/hobby use. Solidworks costs a pretty penny. 😉

Edited April 8, 2023 by Inquisitor

[ElectroSoldier]

On 4/8/2023 at 1:06 PM, MoFo said:

Note, though, that the scan... sucks.  None of the holes are round, smooth surfaces aren't, and the raw scan data is basically unusable - he's just using it for gross measurements to completely rebuild the part in CAD.

Depends on what you want to use it for.

Its not unusable at all, youre over exaggerating there, I mean he uses it in the video proving you wrong.

Or is this a way to throw you hat into the arena?

[MoFo]

I mean, he uses the scan to completely re-draft the part in CAD, because it's not even remotely useable for production purposes, but other than that, sure, it's great.

 

 

My point being, if it's so bad at recreating such fundamental geometric shapes as "circle", "flat" and "parallel", then don't expect to be able to just scan and print kit parts.  The technology, while it has its uses, just isn't there yet.

 

As for the Solidworks vs Fusion 360 debate...  Solidworks now offer a hobbyist license for $90/year.  https://discover.solidworks.com/3dexperience-solidworks-makers

[ESzczesniak]

13 minutes ago, MoFo said:

I mean, he uses the scan to completely re-draft the part in CAD, because it's not even remotely useable for production purposes, but other than that, sure, it's great.

 

 

My point being, if it's so bad at recreating such fundamental geometric shapes as "circle", "flat" and "parallel", then don't expect to be able to just scan and print kit parts.  The technology, while it has its uses, just isn't there yet.

 

As for the Solidworks vs Fusion 360 debate...  Solidworks now offer a hobbyist license for $90/year.  https://discover.solidworks.com/3dexperience-solidworks-makers

But he could have 3D printed the part with basically no modification…a mesh usually needs to be “closed”, but that’s about it. 
 

He needed to rebuild the part because of the required tolerances. However in other applications such as scale modeling, it may not matter if two axes colinear, etc. 

[ElectroSoldier]

Not sure we will have a definitive answer until somebody buys one and tries it.
All else is speculation.

[serendip]

As I understand it many kit manufacturares now use LIDAR ((laser) light radar) to scan subjects subsequently aligning the scans, getting rid of all unwanted 'noise' after which (and I quote) the point cloud is converted into a solid polygonal mesh object ready for use as a template in the CAD system.  This technique works well if a tad expensive.

 

What are the fundamental differences between this and what the guy in the video in the thread above was using:
https://www.youtube.com/watch?v=imGrla3b3Mo

 

Besides cost and portability? Is he using an entirely different technique?

 

 

 

[Nate]

The mathematics behind mesh models (point cloud data) and parameterized solid models (Solidworks, CATIA, Creo, NX, etc.) are fundamentally different, which is why you can't turn a mesh into a parameterized solid. You can convert a parameterized solid into a mesh (Fusion creating .stl files), but not the other way around.

 

3D and CAD seems super fancy and exotic, but the problem is no different than what we deal with when we work with 2D images. A vector image (.svg) can be converted to a raster image (.jpeg), but a raster image can't be converted back to a vector image without some elbow grease. Similarly, if you scan a document, it is a raster image and requires post-processing and manual drawing to create a vector image from the scan. Nobody complains that they have to manually trace outlines of general arrangement drawings to make profile art, or that OCR is riddled with typos, but now all of the sudden it's a failing of technology when you add another dimension and you have remodel things from scan data.

 

Every .stl exported from a parameterized solid model is some sort of faceted interpolation of a mathematically perfect surface. Printers need this so they can have discrete points in space to drive to, since the G code is a basically just instruction to go somewhere. You can see this faceting with a simple experiment in Fusion by manipulating your export settings. Make a cylinder in Fusion - its smooth shape is defined by x^2+y^2, so you can scale it all you want and it looks smooth and also basically maintains the same file size. Type in a new diameter and it just changes the value of the variable in the existing equation.  Now export that cylinder to an .stl but change the "refinement" settings (Fusion 360 .stl settings) to low. Bring that .stl into your mesh editor of choice and then compare it to another export with high refinement. Both .stl cylinders are made of facets, the higher refinement one just has more facets so it looks smoother and is also a larger file because it has more data. You could do math to interpolate the circle created by discrete points in the .stl and try and derive the true shape, but you don't get back to that exact x^2+y^2 circle without making some assumptions (the points in the .stl typically stop at 4 or 6 decimals, and the perfect curve point from the equation might actually be 37 decimals long or something). If you scale up the .stl, you will scale up the faceting as well - just like zooming in on a .jpeg.

 

I've never used Meshmixer, but industry software like Spatial Analyzer and Polyworks can interpolate the scan data and generate 3D primitive features with some handholding - very similar to Photoshop/GIMP automagic tools to recognize boundaries and whatnot to help vectorize an image. Given the appropriate constraints, the software can make axes, cylinders, planes, etc. from cloud data. It can't insert the design intent or dimensional relationships between geometries that a designer would through a parameterized model, nor can it compensate for the natural variation in the physical object and the noise in the measurements. If you give me an engineering drawing that says 1.0" +/-.020", I can deliver you product that is .980" or 1.02". So, if you scanned a part delivered at the low end of the tolerance and somebody else scanned a different part at the high end of the tolerance, you could spend days arguing on forums over who was right and who was wrong and who's father smelt of elderberries only to both be right (and wrong). In a one-off replacement/repair/matched part situation, this doesn't matter. In an OEM, serialized production scenario, this is a problem.

 

I hate to say it, maybe it's because I've watched Terminator 2 too many times, but the magical scan to parameterized model solution is likely AI-based because there is so much noise in scan data and subjective judgement required for a robust 3D model. At the hobbyist level, it probably doesn't matter and a smoothed mesh is fine, but that's why there is a ton of calories burnt at the industrial level to rebuild models.

 

Sorry for the essay - it was mostly cathartic because I deal with people at work all the time who don't understand why CAD takes so long and why you can't just put somebody in front of workstation and get instant results ...

[habu2]

I started 3D modeling in the early 80s.  We had no CAD, we had no modeling software. We had to calculate points in a plane out to 4 digits, we had to define a list vertices to define polygons, a list of polygons into objects, and define separating planes to determine priority between objects.  The concept of Z-buffering was years away. I met James Blinn, who explained the concept of mapping bit-mapped texture patterns onto polygons. We had racks and racks of custom designed hardware to achieve a rendering rate of a whopping 1500 polygons at 30 Hz.  

 

I started modeling on a PDP-8 and moved up to a VAX-11/780, and eventually SGI Iris workstations the size of a mini-fridge. 

 

Yeah, I'm old. 

[Nate]

OMG Habu2, you are my hero.  I thought I was old because I downloaded the original Rhino 3D beta over dial-up in high school, had to save my undergrad CAD work on a Zip Disk, learned NX when it was still I-DEAS, Creo when it was ProEngineer, and (very briefly) had work in industry in CATIA V4. The entire Rhino installer was like 8.5MB and took me a couple days to download, but was totally worth it for free NURBS modeling....

 

Thanks for making me feel like a spring chicken.....

[serendip]

16 hours ago, Nate said:

The mathematics behind mesh models (point cloud data) and parameterized solid models (Solidworks, CATIA, Creo, NX, etc.) are fundamentally different, which is why you can't turn a mesh into a parameterized solid. You can convert a parameterized solid into a mesh (Fusion creating .stl files), but not the other way around.

 

3D and CAD seems super fancy and exotic, but the problem is no different than what we deal with when we work with 2D images. A vector image (.svg) can be converted to a raster image (.jpeg), but a raster image can't be converted back to a vector image without some elbow grease. Similarly, if you scan a document, it is a raster image and requires post-processing and manual drawing to create a vector image from the scan. Nobody complains that they have to manually trace outlines of general arrangement drawings to make profile art, or that OCR is riddled with typos, but now all of the sudden it's a failing of technology when you add another dimension and you have remodel things from scan data.

 

Every .stl exported from a parameterized solid model is some sort of faceted interpolation of a mathematically perfect surface. Printers need this so they can have discrete points in space to drive to, since the G code is a basically just instruction to go somewhere. You can see this faceting with a simple experiment in Fusion by manipulating your export settings. Make a cylinder in Fusion - its smooth shape is defined by x^2+y^2, so you can scale it all you want and it looks smooth and also basically maintains the same file size. Type in a new diameter and it just changes the value of the variable in the existing equation.  Now export that cylinder to an .stl but change the "refinement" settings (Fusion 360 .stl settings) to low. Bring that .stl into your mesh editor of choice and then compare it to another export with high refinement. Both .stl cylinders are made of facets, the higher refinement one just has more facets so it looks smoother and is also a larger file because it has more data. You could do math to interpolate the circle created by discrete points in the .stl and try and derive the true shape, but you don't get back to that exact x^2+y^2 circle without making some assumptions (the points in the .stl typically stop at 4 or 6 decimals, and the perfect curve point from the equation might actually be 37 decimals long or something). If you scale up the .stl, you will scale up the faceting as well - just like zooming in on a .jpeg.

 

I've never used Meshmixer, but industry software like Spatial Analyzer and Polyworks can interpolate the scan data and generate 3D primitive features with some handholding - very similar to Photoshop/GIMP automagic tools to recognize boundaries and whatnot to help vectorize an image. Given the appropriate constraints, the software can make axes, cylinders, planes, etc. from cloud data. It can't insert the design intent or dimensional relationships between geometries that a designer would through a parameterized model, nor can it compensate for the natural variation in the physical object and the noise in the measurements. If you give me an engineering drawing that says 1.0" +/-.020", I can deliver you product that is .980" or 1.02". So, if you scanned a part delivered at the low end of the tolerance and somebody else scanned a different part at the high end of the tolerance, you could spend days arguing on forums over who was right and who was wrong and who's father smelt of elderberries only to both be right (and wrong). In a one-off replacement/repair/matched part situation, this doesn't matter. In an OEM, serialized production scenario, this is a problem.

 

I hate to say it, maybe it's because I've watched Terminator 2 too many times, but the magical scan to parameterized model solution is likely AI-based because there is so much noise in scan data and subjective judgement required for a robust 3D model. At the hobbyist level, it probably doesn't matter and a smoothed mesh is fine, but that's why there is a ton of calories burnt at the industrial level to rebuild models.

 

Sorry for the essay - it was mostly cathartic because I deal with people at work all the time who don't understand why CAD takes so long and why you can't just put somebody in front of workstation and get instant results ...

On the contrary Nate, thanks for the essay - I just read it and will again this weekend - hoping you're game for some questions.

Thanks,

Marc.

[ElectroSoldier]

Recent video about a printer that has, as said in the video been talked about a far bit, and bares out what I said about printers not really improving the prints as much as they will be improving the features.

 

BEST 3D Printer Right Now - Uniformation GK TWO Review (great all-rounder Uniformation GKtwo) - YouTube