[Pharkas]

Hi,

We run a small shop.

We do over 1200 different parts for a particular client, and several other parts for other customers, programmed in Mastercam.

I was wondering how you guys managed all your files. I currently store mine in folders using drawing numbers and add special comments & instruction in a text file in the same folder.

Since we do a lot of parts in low volume, this is starting to get a bit overwhelming and I am looking for a suggestions to try and improve.

I searched and found the following add-on for Mastercam from Camaix : http://www.camaix.net/projectmanager.html . Does anybody here use this ?

Any help and suggestion would be great !

Thanks !

[HIRAH]

we have a little over 2700 parts. each part has a specific # assigned to it.
there is a folder for each part that contains ALL the information for that part. cad files, cam files, operating instructions, inspection etc.
we have a database set up in microsoft access. search for the part # and the # comes up. open 1 folder to find whatever you need.
simple and effective. pain in the ass if you are trying to implement it into an existing system though.

[JerryFlyGuy]

This is something that I've always struggled w/. I have several hundred customers and I've got over 41000 files in over 3800 folders [ I checked ] I've never found a good numbering system that works for me. Being as I generally design all my own stuff and rarely get drawings from others, how did you guys figure out your part numbering systems and how well do they work?

Finding files can be a challenge at times, however usually I can find what I need in under 5 minutes. I do use the search function under the start menu but it can be cumbersome depending how many other item's you have that are simularly names [ "doubler plate" for example..] If I start to get too indepth with the part names, they get so darn long that it's like reading a book. Ie; "IP-Mill3-Main Drive-Idler Bracket-Shock Mnt"

It's working the way I do it now, but I'm open to suggestions on how to make it better

[HIRAH]

we started at #1. every part has a different #.
there is also another # associated with the job for each po.
works great. all the routing info has the number on it.

[JerryFlyGuy]

so you have to look at a print to see where in what sub assembly each part goes? I've tried to get some type of sequential numbers system to work, where you have sub assm 01 which has all of its compent parts as a 01-01 etc.. but it starts to get really confusing when you have third or fourth order level's in your sub assemblies, ei: 01-01-01-01 as a part number. and then organizing drawings etc..

Then there are lots of parts that are 02-01-0-0 etc.. and then there are parts which are labeld 02-01-0-0 which also used in the 03-0-0-0 assembly.. but the number doesn't belong in the sequence.. etc.

Ahhh.. the fun of all of it..

[ajmoir]

Hey something I know a bit about :-)

I work as software guy on large database systems.

The problems you are having are related to using the same field for more than one purpose, that's bad :-(

You need to have a unique number for each part and then the same part can be used in two or more drawings. Organizing all this is where you use other tables e.g. customer uses these parts, PO uses these parts, this drawing uses these parts etc etc.

I don't have a lot of drawings but I do have a lot of source code. The way I'd go about it is to set up a standard unique number for each drawing. I'd then have a program that could automatically read a drawing and tell me which other drawings are being used, then I could very quickly scan all my drawings for the use of a drawing.

I'd also use a nice version control system to track all my drawings and all my changes and allow me to roll back to a previous version. I'd probably mirror my VCS to an off site vendor so I have real time backups :-)

Then I could implement a bug tracking system to handle changes to drawings. I'd also have an automated order system that allowed me to quickly see what drawings are needed for what order.

But like I say I'm a software guy what do I know :-)

[Switcher]

This is something that I've always struggled w/. I have several hundred customers and I've got over 41000 files in over 3800 folders [ I checked ] I've never found a good numbering system that works for me. Being as I generally design all my own stuff and rarely get drawings from others, how did you guys figure out your part numbering systems and how well do they work?

Finding files can be a challenge at times, however usually I can find what I need in under 5 minutes. I do use the search function under the start menu but it can be cumbersome depending how many other item's you have that are simularly names [ "doubler plate" for example..] If I start to get too indepth with the part names, they get so darn long that it's like reading a book. Ie; "IP-Mill3-Main Drive-Idler Bracket-Shock Mnt"

It's working the way I do it now, but I'm open to suggestions on how to make it better

How large are each of your files (approx.)?

Also, how many different type of files, in each folder (approx)?







.

[JerryFlyGuy]

The files range from a few KB to several hundred MB [ a few are in the GB range] there can be from 1 to several hundred files in a folder. The files can be dwg,dxf,pdf,sldprt,sldasm,slddrw,pdf & jpg. There may be a few others in there but those are the majority. There is over a 100GB of data just in CAD files. The picture files are about that much again.

Old Data and projects are zipped to DVD and stored in two locations off site.
This leaves me w/ the data I need or which is current or repeatable in the forseeable future..

Ajmoir, do you work for NASA? I have the feeling that source code can't be cataloged the same way that drawings and model files can be.. but then.. i've never had any experiance w/ that type of thing too much.. well.. since GWBASIC in highschool anyway..

[ajmoir]

Ajmoir, do you work for NASA? I have the feeling that source code can't be cataloged the same way that drawings and model files can be.. but then.. i've never had any experiance w/ that type of thing too much.. well.. since GWBASIC in highschool anyway..

No not Nasa, although if you know of an opening in the JPL... I'm just a dev with way too much experience, everything looks like software to me :-)

The reason I think drawings could be treated like source is that dxf is a text based format e.g. open them in notepad and you can read them, just like source. In fact I bet some changes could be automated by changing the dxf directly not in a cad package e.g. moving from an old part to a new one.

Now source code is just text based. We have a build step to make the stuff the computer can run. The reason we use source control is for tracking changes an audit trail if you wish. We can scan all the source and see which file is used where etc.

FYI, I work on a source base that is nearly ten years old, ancient in software timescales, not only can I track every single change, but I can rebuild the product at any particular time in it's history. I can see every xref of a file in the project. Currently we have 5 devs but in the past we had closer to 30 working together on the same product.

Now back to your dxf files. They are the source for your CNC machines. You usually put it through a conversion to GCODE etc etc This is very similar to software.

_IF_ you are having trouble reliably building/finding parts from your system then version control is an obvious first step. Second automate the conversion of dxf to gcode. That way the gcode always matches the dxf. Pretty soon you'll be using make files, it tells the build process which parts are needed and in which order.

I repeat I have not tried this in a machine shop but the more the world moves to virtual representation the more it becomes a software problem. Believe it or not the basic engineering of software is easy to control it's all the permutations that kills us.

[JerryFlyGuy]

Ok, now I see where your coming from. I'm kind of off the scale however. The reason being is that about 10-20% of my files are DXF or DWG, the rest are Sldprt, Sldasm, SLDDRW or other file formats. I have software controls which allow me to readily see what files are used in what parts and models etc. That isn't really the issue. The issue is more organizing my generic part libraries and various project's in a way which works well for my inhouse products as well as for product's which are designed and built for other's. There are things like 'lift lugs' which could possibly be used for other projects, instead of designing them from scratch each time. I do make use of my softwares library functions, however it's more the part number and or job number, numbering system's which I have no grasp of. Or at least, never found a system that worked for it all.

If I store job's under their job number, then I need a master file someplace which I can search for to find what job's had what job numbers, then go search or look for the folder w/ that job number. Of course adding customer numbers to the job numbers would help in this regard, but it seem's like its an excercise in creating and maintaining lists and databases. Neither of which I'm interested in.

What I'd like to be able to do is have a part numbering system which allows me to look at the number and de-code it in my head as to what kind of part it is, how it was made and whether it was a inhouse part or a part which was designed for a customer, or possible used in both instances. I want to get away from naming things with names, yet have the familiarity of names in a numbering system. The system must be able to flow the other direction as well. I should be able to find a part just by knowing what it's made from, or how, or by customer or by job or any combination of those things.

Anyway, when you have 40,000 files to work through it isn't just a cut and dried problem.. and I'm looking for more than would possibly be the norm.
Anyway.. I've kinda hijacked this thread and should probably shut up about it all.. as it's not contrubuting to the original question.

[Jay Kyle]

Jerry,
I've been kicking this around myself. Current thought coming out of the think tanks is to use a meaningless number to track your parts, rather than an intelligent number. Intelligent numbering schemes always break down due to the unanticipated myrid of exceptions that pour in. In the database game (my background is actually in the advanced business intelligence side of I.T. with data modeling and a smattering of engineering thrown in) creating intelligent keys has gone away for the more flexible method of abstract keys.

A big problem is when you have a part that is a member of a sub assembly, that is a member of another larger sub assembly..etc. You can end up with an unknown number of parent-child relationships. Intelligent numbering schemes work best when the level of relationships are strictly defined.

However, having said all of that intelligent numbering is compelling, and if you study the large manufactures they employ a combination of both methods (depending on the manufacturer).

As for the other issue of managing files, ajmoir's software development lifecycle (SDLC) example is a good one. What you need is a lifecycle management system. Have a look at what the bigger boy's are using to manage their environments. Features such as version control, project managment, issue management, etc are all features of these kind of systems. Look to the manufacturing industry rather than the software industry, it will be more tailored to your specific needs.

A change in paradigms is going on that aligns with this, with the introduction of Google's gmail, the emphasis is no longer on trying to categorize information, rather to search for it. Just as in your case, trying to walk through 10's of thousand of files and directories for just the right one is a headache, better to let the computer do it for you through a smart search (A single part can be stored too many different ways).

To aid this be sure you are verbose on your descriptions, this is what the search engine will rely on. BUT!!! again there are extremes even in this method, I prefer a middle of the road approach and keep all of my files grouped by project, within customer (regardless of what type of file it is). I consider myself as a customer as well. As for trying create a generic parts library or index of parts made and stored all over, that's were the searching comes in.

As a last item of caution BACKUP EVERYTHING. Computer will fail, and you can take that to the bank! I don't mean manually copying stuff to floppies, I mean make sure you keep you're data on a RAID system, if you use a laptop, setup offline files to your server (which has RAID drives setup), and have an offline system such as an internet backup service or tapes or something that will offline data automatically, etc.

Lastly a good virus scan product, I lost just about every file on my network drive once because of a virus that got through.

Best way to justify these kind of expenditures is to ask yourself - what would it do to my business if I lost all of my files (as what happened to me once).

Jay

[CNCRim]

Create a main fold and many subfolder for each customer and part number. let's said two customers A and B.

For Example, Customer A with part/draw #1223, then main folder A-> subfolder 1223 etc..... Customer B with part/draw #11111, then main B-> subfoder 111111. Thhat is simplest and I have use it for years never need any note to remember or any log to keep track.

[JerryFlyGuy]

Currently I do group projects via customer [or proprietary ] and then break it down into subgroups and from there specific to the project, and then about 10 sub folders which contain the major assemblies and the drawings etc. I usually can find any project I want in just a couple minutes, the problem comes when you have several customers who each have a similar project, then your left wondering... was it that one.. or was it this one.. Having specific parts for specific customers works to a point. However when you want to use that part for a different project then it's no longer specific to that customer so.. you shouldn't duplicate the part but generally thats what happens. This leads to problems when the part or process is changed on one set of data but not on the others.

I've got a system that works pretty darn well... I just wish it worked for EVERY situation, however... I doubt that it is possible, esp w/ a very diverse and fluid shop like the one I work for.

[HuFlungDung]

Is there any known method in computing of making a shortcut to a file behave the same as an actual file, kind of a flexible path if you will, that intercepts a program call to open such and such a file as if it is in "Folder1" but really, all that exists in Folder1 is a shortcut to the one and only file of that part, which is stored in "FolderUniversal".

Or something

[cabinetcraft]

Ok, first, are you storing these files on your desktop or on a server? A desktop only situation is disaster looking to happen, even servers fail. Back up to CD or DVD and do it often, make an archive. As far as keeping track develope a directory structure that makes sense to you and your crew, put an NFO file in the main directory explaining the "road map" as it's easy to forget at times. One solution that I have seen is an sql data base. It works wonderfully as you can find a file by entering search parameters such as client name, job number, part number, description and so on. The down side is that the front end needs to be custom designed which will cost you a little, but from what I've seen it would be the best way to go in the long run. But Backup, Backup, Backup what ever you do

[CNCRim]

well I can feel your pain I used to work for a company that has same situations, but oh well if there is a perfect system we all will be out of work. Best way is back up everything and just make one folder for each one....... and for the one similar copy to new folder then modify to whatever to the speci.

[JerryFlyGuy]

Thats pretty much what I do, until I realize that I shouldn't have copied it as it needed to be modified and this means I have to copy it back to the other folder and over right the old part as it should match the new part. And then there are the times that I didn't copy it but I should have as I should have two seperate parts for two seperate customers.. sigh.. I'll just continue doing what I've always done, and get the same results as I've always got.. and hope it all works out in the end.....all of which is one version of the definition of insanity [if anyone wanted to know that ]

[Jay Kyle]

Is there any known method in computing of making a shortcut to a file behave the same as an actual file, kind of a flexible path if you will, that intercepts a program call to open such and such a file as if it is in "Folder1" but really, all that exists in Folder1 is a shortcut to the one and only file of that part, which is stored in "FolderUniversal".

A nice idea at first glance but these systems turn out to be high maintenance. In the windows environment they're called shortcuts, in UNIX they're called links.

If you own Solidworks you should be seriously looking at PDMWorks, if you're a programmer you would use something like CVS.

More in-depth reading can be found at http://www.ecm5tools.com. Sorry to throw this high level stuff out there but it's always good for the career to hear/read best practices.

BTW Toyota has an outstanding manufacturing QA method, and I personally like Honda's numbering scheme.

Jay

[ADucci]

for data archival you have a few options...

1: create a database, in access... with some meta-information that you would need to populate on each design - ie for each design, specify the client, application, category etc.. then create a small program in access to open up the files found in the search result- pm me if you dont know how..
then you would have a search window that would show a list of results where, if you clicked on one... it would open.

2: download some desktop indexing software, this will run through your computer (or a set of folders and archive all information about your files (names) and custom properties ( i will explain later) the searching process on a few thousand records will take milliseconds... examples here are copernic (free) and google desktop (also free)

Custom properties - if you open windows explorer, select a file, right click, select the "Summary" tab, you will see fields that you can populate (summary, category, keywords etc.. you can then search on these attributes...

the problem with a database is that you would need to maintain it... and the nice thing about indexed searching is that it is instant and doesnt worry what folder the file sits in