|
Post by joanlluch on Dec 15, 2017 23:48:25 GMT
I've found over time that one physical part per root folder is the best way to keep the files in some kind of order. Chaos is always just around the corner if you don't! The use of explicit folders is increasingly regarded as obsolete by some younger folks. You just need to look at the ways content tends to be stored and retrieved in mobile phones or modern computer applications. Advanced searching and filtering seems to be the new trend. In my case, I use a mixed system of folders and ‘structured’ file names to store CAD part drawings and other related files. This makes easier to find particular parts, and you don’t end having one thousand folders. For example I may use file names with a structure like this: assembly-subassembly-partcategory-partname. By following this schema, all parts are explicitly related by its own name and they appear on directory listings next to their related ones. You get all the benefits of structured top-down classification of parts, without the hassle of multiple folder nesting. The CAD software also seems to make this approach easier than dealing with many folders. Essentially, when you create new part within an assembly, the part is stored by default on the same folder than the assembly, therefore you just need to give the part the right ‘structured’ name and you are done without having to create first a new folder for it. I found this way to be more productive. If you later decide to move a part from an assembly to another, then you just need to “save as” the part with a new appropriate name, again without having to bother about multiple nested folders or folder structures that would need tedious updating/refactoring both in the file system and the CAD assembly references.
|
|
|
Post by Oily Rag on Dec 16, 2017 1:16:59 GMT
Each to their own as they say, but what are you basing your 3D designs of existing components on ? I am redrawing Brian Wilson's Eric from his book for my current garden railway steamer, he has 2D cad and 3D CAD pictures in the book Steam Trains in the Garden. I am very pleased I am as there is much missing and many little hicups. In the end there is not much the same, from wheels to motion. But the book is a good starter and motivator but needs a rework with a good technical editor and during that process a loco built at the same time. Document control process and with the beginner in mind. For my C17 5" gauge to build in the years to come,(I did the drawings when I had health problems and had to work reduced hours and not having a ME workshop to play in) For those I started with Neil MacKenzie's drawings but they left out loads and loads of details and even the frames had a major problem. I also wanted to build an early C17 which was not the drawings intent. So I married those up with the original Queensland Railways drawings of 1919 and also the full size loco. The assembly discovered mismatches etc. I spoke to Neil, he lives locally about sharing my drawings with him and updating but he was not very interested and he is getting on in years so I respected that. The castings exist but looking at the wheels for example, looks like I will go down the 3D print to wax and have them investment cast for high detail. I have thousands of QR drawings, from my involvement with full size steam and trips to the state archives.
|
|
|
Post by Roger on Dec 16, 2017 8:44:04 GMT
I've found over time that one physical part per root folder is the best way to keep the files in some kind of order. Chaos is always just around the corner if you don't! The use of explicit folders is increasingly regarded as obsolete by some younger folks. You just need to look at the ways content tends to be stored and retrieved in mobile phones or modern computer applications. Advanced searching and filtering seems to be the new trend. In my case, I use a mixed system of folders and ‘structured’ file names to store CAD part drawings and other related files. This makes easier to find particular parts, and you don’t end having one thousand folders. For example I may use file names with a structure like this: assembly-subassembly-partcategory-partname. By following this schema, all parts are explicitly related by its own name and they appear on directory listings next to their related ones. You get all the benefits of structured top-down classification of parts, without the hassle of multiple folder nesting. The CAD software also seems to make this approach easier than dealing with many folders. Essentially, when you create new part within an assembly, the part is stored by default on the same folder than the assembly, therefore you just need to give the part the right ‘structured’ name and you are done without having to create first a new folder for it. I found this way to be more productive. If you later decide to move a part from an assembly to another, then you just need to “save as” the part with a new appropriate name, again without having to bother about multiple nested folders or folder structures that would need tedious updating/refactoring both in the file system and the CAD assembly references. Hi Joan, Things like gmail use this filtering approach, but Windows doesn't support that kind of thing so there isn't much alternative to using folders. Personally I don't see any issue with having a lot of folders, if they are laid out logically it's easy enough to find your way around. In an ideal world, you'd be able to tag any Windows file and none of this would be necessary, but that's not the way it is at the moment. I know that Fusion360 handles the grouping of assemblies from the inside out, compared to my system which groups together parts that have already been created. There's nothing to robustly connect them, and you can get into a lot of trouble if you just copy and paste a part and then insert it into an assembly. This is because you copy what ought to be a unique ident for that part, and now you've got two potentially different parts with the same ident. You have to use the Save As route to avoid this problem. If you only have part models in the structure then life is a lot simpler. I typically have parts, drawings, notes, pictures and many g-code files for every part. It's hard to see how you could organise those in a Windows environment without folders for each part. I'm sure structured names are a reasonable way to name things, but it's by no means essential. I think it would be fair to say that most people are nowhere near as disciplined are either of these approaches. Anyway, I'm sure there are many ways to do this, I'm just sharing what I do, not trying to convert anyone to my way of thinking. We all do things in ways what suit our own situation.
|
|
|
Post by joanlluch on Dec 16, 2017 9:40:08 GMT
Hi Roger,
Of course everybody do things the way we think it’s more convenient for us. I don’t think none of us want to “convert” others. We are all just expressing what works for us and what not. For me, thousands of folders just doesn’t work that well.
My approach precisely overcomes the lack of proper file search features in Windows. All parts are appropriately prefixed with names that relate them with their assemblies and other parts, so searching is only a matter of listing them sorted by name. All related parts will just appear together. I can still use folders for parts or assemblies that belong to multiple parent assemblies where I want uniqueness and consistency among them.
I do not have g-code files, but I also maintain pdfs, dxf, and other external file types in the same way. Since these files have different file extension names, I just keep the same identical naming convention with their already distinct extension, so they appear next to their related part or assembly on the file listing. In my case there’s no need to put them in folders most of the time, but if I feel the need I would just create a folder with the ‘structured’ part name and put whatever in there, except the part itself, which would remain at the same level than the said folder.
A further advantage of my approach is that the same straightforward finding of all relevant parts applies on assembly edition, and during the creation of 2D drawings with embedded part listings in them. Part listings will already get structure information implicit by the structured file names without the need to add explicit metadata. All that is just not that easy to have by putting part files in separate folders even if these folders keep some structure.
I do not know how your system works, but I would assume most 3D CADs deal with files in a similar way. There's no "robust" filesystem links between parts or between parts and assemblies. The software just looks for parts (and subassemblies) with the names specified in its parent assemblies and loads them from the disk. It first searches in the directory that was specified for the part when added to their immediate parent assembly. If not found, then looks into the assembly directory. So having parts in subdirectories is prone to break assemblies when moving them around because they are not in the location that the software would expect by default, unless you are very careful and know what you do. On the other hand, having them all in the same folder makes refactoring easier because the software will just keep looking in the assembly directory regardless of anything else.
Said that, I agree that keeping a "structured" name convention requires some discipline, but I found that’s the only way to deal with the literally thousands of parts and assemblies that my locomotive drawings already have, in a manageable way. Using one folder per part would have been a lot more tedious and fragile. However, to be honest I have no idea of what's the preferred approach in the industry, I learned CAD just after I decided to join this hobby, and had never used it before. I just applied what looked reasonable to me based on my previous experience on software projects with tens of thousands of source code files.
|
|
|
Post by jon38r80 on Dec 16, 2017 14:20:22 GMT
There is other very good reason for using Rogers file structure approach in that file names have a finite length in search algorithms and if you use the lazy long name dustbin approach to storage you will find the file names rapidly get too long for searches to function proper.y.
|
|
|
Post by simplyloco on Dec 16, 2017 16:09:19 GMT
Are we not talking about Bills of Materials (BOM) as used in Inventory Control systems? There must be loads of software out there, free or otherwise. John
|
|
|
Post by joanlluch on Dec 16, 2017 16:17:45 GMT
Are we not talking about Bills of Materials (BOM) as used in Inventory Control systems? There must be loads of software out there, free or otherwise. John My 3D CAD software can create BOMs from assemblies based on a top-down approach. You have several options to configure them and decide what they contain based on general options, or individual parts or subassemblies metadata and properties. I am inclined to think that most softwares would do. Not sure if this is what you are asking.
|
|
|
Post by joanlluch on Dec 16, 2017 18:24:39 GMT
There is other very good reason for using Rogers file structure approach in that file names have a finite length in search algorithms and if you use the lazy long name dustbin approach to storage you will find the file names rapidly get too long for searches to function proper.y. I wasn't aware of that possible issue, but in any case there's no need to perform actual searches. The idea behind my naming convention is the ability to have directory listings with many parts and yet having them fully identifiable by just displaying them sorted by name. There's no need to have very long names. My convention actually breaks name prefixes with folders. For example I have a single folder for my "SteamPump" parts and assemblies. This folder is named "SteamPump" and it belongs to a parent folder named "Locomotive". Since steam pump parts and subassemblies are already in the "SteamPump" folder they do not need to be prefixed with "Locomotive-SteamPump-". However, they may have prefixes such as "SteamCylinder-Valve-Box-" and so on corresponding to lower levels of assembly nesting. Another reason why structured names do not need to be that long, is that you can just use short words to add levels down the structure. I mean, you won't normally name an isolated file just "box", because that would be very misleading, but you can use "box" as the next (or last) level down the naming structure with no ambiguity, because the existing prefixes disambiguate it by adding context. If fact my global approach is a combination of a nested folder structure (that I keep at a very reasonable minimum) with structured naming of files. For the above example I have a single folder with all the relevant files for the Steam Pump in it. These files have structured names from this level down, but they do not inherit prefixes from their parent folders. This also enables moving entire folders to a deeper or upper level without having to rename all files in them. I hope that's now clearer.
|
|
Midland
Elder Statesman
Posts: 1,870
|
Post by Midland on Dec 16, 2017 19:38:05 GMT
I am glad they still make paper and pencils!!! D
|
|
jools
E-xcellent poster
Posts: 200
|
Post by jools on Dec 17, 2017 1:50:44 GMT
Each to their own as they say, but what are you basing your 3D designs of existing components on ? I am redrawing Brian Wilson's Eric from his book for my current garden railway steamer, he has 2D cad and 3D CAD pictures in the book Steam Trains in the Garden. I am very pleased I am as there is much missing and many little hicups. In the end there is not much the same, from wheels to motion. But the book is a good starter and motivator but needs a rework with a good technical editor and during that process a loco built at the same time. Document control process and with the beginner in mind. For my C17 5" gauge to build in the years to come,(I did the drawings when I had health problems and had to work reduced hours and not having a ME workshop to play in) For those I started with Neil MacKenzie's drawings but they left out loads and loads of details and even the frames had a major problem. I also wanted to build an early C17 which was not the drawings intent. So I married those up with the original Queensland Railways drawings of 1919 and also the full size loco. The assembly discovered mismatches etc. I spoke to Neil, he lives locally about sharing my drawings with him and updating but he was not very interested and he is getting on in years so I respected that. The castings exist but looking at the wheels for example, looks like I will go down the 3D print to wax and have them investment cast for high detail. I have thousands of QR drawings, from my involvement with full size steam and trips to the state archives. I have just about finished re-drawing Don Youngs Black 5 plans (13 x A0 sheets) in AutoCAD which will enable me to print a laminated A3 book of all the sub assemblies to take to the workshop and preserve the originals. From this A3 book I can then begin to translate them into 3D via Fusion360 as another training exercise but it helps to know what they should resemble. This redraft has been a very useful exercise at this has helped explain the assemblies, parts, names and locations - something that was not obvious to a relative beginner. I'm looking forward to starting proper, hopefully after Easter when all of the other "jobs" are done. Jools
|
|
|
Post by joanlluch on Dec 26, 2017 20:33:02 GMT
Speaking about 3D CAD, I just learned that Autodesk Fusion 360 is able to operate in both parametric and direct modeling modes, i.e. not only parametric mode. knowledge.autodesk.com/support/fusion-360/troubleshooting/caas/sfdcarticles/sfdcarticles/How-to-switch-between-a-parametric-or-direct-modeling-environment-in-Autodesk-Fusion-360.htmlI use Solid Edge. It features Synchronous technology that essentially is an advanced way to create and modify 3D models in a direct way which is not history based. This technology enables the direct edition of the model in ways that you did not previously anticipate. This is a great feature when you are actually designing things, rather than just drawing them in 3D. Some competitors, such as SolidWorks, claim to feature "direct edit". However, SolidWorks doesn’t really do direct model edition the way Solid Edge does. SolidWorks do something that looks a lot like direct edit, but it isn’t real direct edit. It's in fact a feature with a tiny fraction of the functionality of a real direct editor. Solid Edge Synchronous technology, is a bit more than just direct model editing because it also involves implicit relationships on faces or symmetries. It's also a patented feature, so I would expect that this technology would be available on other packages only after some time or after the patent has expired. Still, Autodesk Fusion 360 claims to feature a "direct modelling mode". I have looked at the video below and for simple edits it actually resembles Solid Edge: help.autodesk.com/view/fusion360/ENU/?guid=GUID-DCC3B4DD-8853-40E5-AA33-1839092C99BFCompared to the traditional history-based creation of 3D features on parts, the direct modelling mode makes 3D drawing a lot more fun and productive. You stop thinking on what's the best order to add 3D features on your part to help to modify it in the future. You instead start to think just on your part shape and you just work directly on it. Now, I'm not an user of Fusion 360, but if it has a true direct edit mode, as it seems to confirm the video above, then this is to me a TOP reason to chose this package over others lacking this feature.
|
|
|
Post by joanlluch on Dec 26, 2017 21:49:10 GMT
Well, if somebody is not yet fully convinced about the benefits of direct model editing, I found this example video based on Fusion 360. I just want to add that the same editions on Solid Edge could be performed with less steps by allowing the software to identify several symmetric faces that would be edited on a single step.
|
|
|
Post by atgordon on Dec 26, 2017 23:34:04 GMT
Speaking about 3D CAD, I just learned that Autodesk Fusion 360 is able to operate in both parametric and direct modeling modes, i.e. not only parametric mode. SNIP Compared to the traditional history-based creation of 3D features on parts, the direct modelling mode makes 3D drawing a lot more fun and productive. You stop thinking on what's the best order to add 3D features on your part to help to modify it in the future. You instead start to think just on your part shape and you just work directly on it. Now, I'm not an user of Fusion 360, but if it has a true direct edit mode, as it seems to confirm the video above, then this is to me a TOP reason to chose this package over others lacking this feature. I use F360 for all my work, both commercial and loco related. And you are correct, it allows direct modelling, but it is worth mentioning that it is usually a timeline based process (i.e. you have to create a sketch before creating a 3D feature based on that information). And so, if you inadvertently delete a sketch, you will quickly see that any related 3D parts are flagged with errors. You can use the Direct Modeling/Base Feature aspect too, but I have not found any benefit from doing so, well at least up to now! I look forward to learning from you how we can exploit that feature! I used to use PTC's Creo, which I don't think uses a timeline basis (I think it was feature based, whereby each element was a unique set of parameters). I haven't seen any problem with the F360 timeline basis with the exception that you cannot move timeline actions backwards of forwards without causing problems (qualifier: I haven't found a way of doing so!). Since almost everything I do is prototypical, I do not use the parametric features in F360 very much, but what little I do use has shown me that it is somewhat limited when compared to Creo or Solidworks. For our work, that is probably OK, but in a production environment, parameters that are based on formulae should be supported (e.g. scaling a BA screw by just typing 2BA into a box to create your 3D model with the correct geometry) as they are in Creo, Solidworks and Catia.
|
|
|
Post by atgordon on Dec 26, 2017 23:44:55 GMT
I have just about finished re-drawing Don Youngs Black 5 plans (13 x A0 sheets) in AutoCAD which will enable me to print a laminated A3 book of all the sub assemblies to take to the workshop and preserve the originals. From this A3 book I can then begin to translate them into 3D via Fusion360 as another training exercise but it helps to know what they should resemble. This redraft has been a very useful exercise at this has helped explain the assemblies, parts, names and locations - something that was not obvious to a relative beginner. I'm looking forward to starting proper, hopefully after Easter when all of the other "jobs" are done. Jools I'm sure that you know that you can import Autocad drawings directly into Fusion 360. It can get messy, but with careful use of layers in Acad, you can end up with the 2D elements that you need in F360 quite quickly. Flat parts without any change in thickness work very well (e.g. frames). However, for some components, particularly anything that has extruded holes or radii, such as a cylinder block or saddle are probably best drawn in 3D in F360 from the Acad drawings.
|
|
|
Post by joanlluch on Dec 27, 2017 11:18:37 GMT
I use F360 for all my work, both commercial and loco related. And you are correct, it allows direct modelling, but it is worth mentioning that it is usually a timeline based process (i.e. you have to create a sketch before creating a 3D feature based on that information). And so, if you inadvertently delete a sketch, you will quickly see that any related 3D parts are flagged with errors. You can use the Direct Modeling/Base Feature aspect too, but I have not found any benefit from doing so, well at least up to now! I look forward to learning from you how we can exploit that feature! I used to use PTC's Creo, which I don't think uses a timeline basis (I think it was feature based, whereby each element was a unique set of parameters). I haven't seen any problem with the F360 timeline basis with the exception that you cannot move timeline actions backwards of forwards without causing problems (qualifier: I haven't found a way of doing so!). Since almost everything I do is prototypical, I do not use the parametric features in F360 very much, but what little I do use has shown me that it is somewhat limited when compared to Creo or Solidworks. For our work, that is probably OK, but in a production environment, parameters that are based on formulae should be supported (e.g. scaling a BA screw by just typing 2BA into a box to create your 3D model with the correct geometry) as they are in Creo, Solidworks and Catia. Well, I use Solid Edge so I can’t really say much about Fusion 360 other than what I saw on videos. Solid Edge is an excellent product but is quite expensive so I am wary of recommending it because I know most people wouldn’t be able to afford it (including me). I am lucky to have access to a spare license of an old version of the software that a friend in the equipment manufacturing industry activated for me. In Solid Edge you can apply history-based modelling features to a directly modelled part, but not the opposite. They name “synchronous environment” to direct modelling. History-based modelling is the “ordered” environment. In Solid Edge you can create a pure history based (ordered) model in the traditional way, or you can create a synchronous model with /optional/ ordered features. There is not such a thing like “disabling capture design history”, if you are drawing ordered features then the drawing history is always available to you. Since you can not add synchronous features to an ordered model to begin with, my understanding is that this prevents failure or errors on the ordered part. Both history-based and direct modelling approaches in their pure forms have strengths and weaknesses, but the ability to mix the two in Solid Edge brings the best of all worlds. Direct (synchronous) modelling may seem straightforward, but understanding it takes longer than it may seem. I am sure that some people who dislike it or get frustrated by it, just don’t fully get it, or haven’t worked enough on it, or they haven’t yet made the mental switch from the history-based way of drawing. Regarding this subject, I found an article based on Solid Edge that covers it all in a very comprehensive way. community.plm.automation.siemens.com/t5/Solid-Edge-Blog/Get-the-Most-from-Synchronous-by-using-a-little-History/ba-p/2172While using direct modelling you will find many times that the software seems to not understand what you want to do or gives you errors of unsupported geometry. However, once you get a grasp on it you can always trick the software to do the edit that you want. Sometimes you may need to temporarily remove a feature in order to make the change then add the removed feature back. Or you may have to apply a temporary change before you apply your intended one. Or you may have to edit the part in increments so the intermediate geometries help the software to understand your final intended geometry. Or add dimension or geometry constraints so your edit is performed as you expect, and so on. In any case, I don’t think anyone would want to go back to history-based drawing, once they have grasped direct editing, unless they really need to maintain fully parametrised parts due to industry exigence.
|
|
|
Post by silverfox on Dec 28, 2017 12:29:48 GMT
Roger
Yes found it there!!
Wow, that seems like i will be on my machine for the rest of my life!!
Thanks, it is very very detailed and hopefully for me easy to understand. i WILL give it a try, but reserve the right to keep my sanity and farm it out!!!(lol)
Appreciate the time you took in putting it down for me
Ron
|
|