Where is CAM's "Killer App?"

I hear you. I was headhunted by System House, way back, to help the sales staff sell electrical utility GIS. The sales staff, all engineers, didn't have a clue how utilities worked and needed someone who could answer the utility specific questions and speak utility.

When AutoDesk bought the division, they decided to try it another way, from what I have heard the whole package was shelved.

Someday these companies will understand the value of use specific sales help. Let the sales people sell, but have an expert voice to answer the specific questions.

The machinist's world is all about tight margin$ and efficiency... Anything that reduces waste is helpful. I think this is a very good question, and could have significant impact on the industrial world...

Step 1: Talk to machinists and manufacturers about their frustrations.. The things that hurt the most are the areas of opportunity. (reducing waste is probably top priority. Reducing setup time might be reasonable... What about integrated CAM file management with ERP vault systems?

Step 2: Analyze the Data you get from step 1.

Step 3: Brainstorm some solutions.

Step 4: Test the proposed solutions.

etc.

I would suggest that improvements could be made in CAM code & process authoring tools at the machine level. (more user friendly)

Great question. 5 stars.

Chris

I suspect a major issue with the efficacy of the CAM packages is the same as the major issue with most packaged software- the developers are totally out of touch with the users, and haven't a clue as to how the package will be ultimately used. When "improving" the package, they most likely are adding eye candy rather than serious capability enhancements. Until you get machinists involved in the development process, you most likely are not going to develop your "killer app"...

cwarner7_11:

One thing with the CAM package (other than support files for translating it into code), is that you basically have to virtually machine the part. And by doing that you have to set-up all the parameters as well as the environment.

One definitely has to be knowledgeable with the machine tool itself. At a time I thought it was a little different way of thinking between a CAM operator (I do not call them programmers because they can't) and machinist.

But the place where I am working at, the machinists are picking up on it quickly, so that they do the programming with MasterCAN...(I'll call the machinists programmers thou)

phoenix911

After so many years in the CAD software business and a six year stint selling Mastercam, I've concluded that the biggest problems are focused around two phenomenons:

1. Management makes purchasing decisions without consulting the end users. This applies to both CAD and CAM software as well as FEA, operating systems, hardware and networks.

2. Buying from salespeople who are clueless about the application.

In our organization, we had two man teams: one the applications expert with background as a designer; the other a salesperson who knew how to convert the prospect to a sale. Rarely do the requisite skills exist in just one person.

Experience has shown that in any given market, two competitors eat up about 75% of the sales. While there may be many others trying to gain market share, the first two into the game usually pick up most of the marbles.

MasterCam has been one of those top two for many years. I no longer know who is competing for second slot but Virtual Gibbs was making meaningful gains last time I looked, on the basis of ease of use and the ability to import and edit surface models in their native format. It's been 10 years since I last looked at this issue so my memory is suspect and things have likely changed some.

As for post processors, standard posts were available for most every CNC machine out there as part of the Mastercam package and while customizing them to suit odd situations did occasionally occur, the better dealers were able to do so without having to call Connecticut.

I find that the best thing to do is contact users of CNC equipment, find out how they are generating tool paths and then see if a trend develops that points to one particular software.

L.J.

I worked at a company for 3 1/2 years using Surfcam. I liked Surfcam. It is very similar to Mastecam but maybe 3-5 years behind in applications. There are however a few things that Surfcam does better than all the rest. The company purchsed Gibbs because their machine dealer recomended it and said everyone they knew who used it loved it. I, who was doing at least 65% of the programming was not consulted. I have to say that the Gibbs dealer was a great guy and so was his training. The software however I found to be very awkward and completely counter intuitive.

Mastercam is still at the top, but they continue to offer things that I don't need and fail to offer more simple but useful things or fix things that have been problems for a long time. The new X versions now up to X3 have taken things that were once easy and made them awkward.

i use x3 and x4 now, but the continous axis is a headeck and is far a way from nx. i t

I started programming in 1981 using a pencil and calculator to program complex surfaces for molds and fixtures. I avoided the the first versions of the CAM softwares that were available because the cost was prohibitive and just didn't make sense. My attitud was that when they come up with a software that can do the things that I cannot do such as lofting and surface patches within a contour then it will be worth it. I started using Mastercam in 1993 and later worked with Surfcam and Gibbs and have done some modeling with Solidworks.

I watched demos of the first offerings by Techsoft in 1981 that ran on an Apple computer and my response then was that they needed to find out what machinists wanted and give it to them rather than come up with their own ideas and expect machinists to adapt. Unforutnately this issue has not changed.

The CAD/CAM companies are bound to the Windows interface which creates the need for constant updates to keep up. More and more fancy junk is added each year that we first have to pay for and then waste our time having to learn just so we can produce CAD models and CNC programs. If you don't pay for the annual upgrades which for me run around $3k not counting support you run the risk of not being able to import files from customers or vendors and not being able to get help when you need it.

As to the question about killer APPs, my response is: How about making existing apps work properly before charging me for a bunch of junk that I rarely need. It would be great if pocket and pocket facing routines could produce efficient toolpaths instead of wasting 30% of my machine's time cutting air and still not cleaning up the pocket. How about if such features as "limit tool burial" or "gouge protection" actually worked. How about if they figured out a way to make the various surface roughing routines to do a continous climb cut so that cutters don't load up by conventional cutting when it is not necessary. The simple act of putting a fillet on a non uniform solid could be made to work a lot better. These are just a few.

Most machinsits use their computers as a tool. We just want to get the job into the machine and get it done. Anything that slows down that process is costly. "Killer applications" are nothing but a waste of time and money unless the baisic applications work well. I am certainly not saying that new features are not needed, just that the goal should be to make the job of producing parts easier and faster, not slower and more costly just so some computer geeks can charge us more for software upgrades.

I agree with most everything you wrote, especially this one:

"As to the question about killer APPs, my response is: How about making existing apps work properly before charging me for a bunch of junk that I rarely need."

Unfortunately, the only way software developers can stay ahead of the competition is by adding "stuff".

Peredo proved that we only use 20% of our resources to produce 80% of our results. If a software, such as Mastercam, is stil in the marketplace and successful after so many years, then it's safe to assume that they long ago provided the 20% we need to get our job done. So why add anything? Simply to stay ahead of the others who are nipping at their heels.

Most added bells and whistles are of little value to the average user. They are often discipline specific and geared to narrowly defined markets most of us won't use.

Regretably, these add-ons clutter the interface and make it difficult for the experienced user to find his way around a once familiar software. The intuitive responses common to the creative process get disrupted and productivity suffers. After many years of using one really intuitive 3D Solid Modeler, I found myself faced with a 2009 "upgrade". It took 3 months before I trained my automatic reflexes to the new environment and only after I customized the interface to get rid of all the clutter.

In their early days most of today's CAD and CAM softwares were driven by end users needs. Now the trend is for the sales departments to drive development and the result looks a lot like you described.

L.J.

Maybe the CAM suppliers should take a lesson from Corel Draw, once a very strong player in the computer generated drawing industry. Unfortunately they fell into the additional app hole and it eventually killed them. they still exist sort of but no where near the market leader they once were.

The stories you write are remeniscent of the war stories Pat Hanratty told us when he was still with GM.

You at least had the benefit of an already listening for CAM applications. Pat had none of that and eventually founded Anvil.

Not sure if he is still around but he was a very impressive and forward thinking man. I have no problems crediting him as the founder of CAM technology.

L.J.

Yep, ComputerVision's CADDS system was based on Hanratty's Adam, as were several other early CAD systems. I spent twelve years with that stuff. Then I left CV and went to work for Hanratty at MCS, Manufacturing and Consulting Services, Inc, doing tech and sales support of Anvil 4000 and 5000.

So, that said, I suppose I share many of the same war stories as Pat. A brilliant, though eccentric and tough to work for, man. And, yes, I'm don't think he has turned over the reins of MCS to his sons yet. Anvil development continues on, and I'm sure, with close association to their customers' needs and wants. MCS is one of the best when it comes to co-developing with their users.

Marketing is not their strong point so they are not widely known. Development drives the company and Doctor Hanratty keeps it tightly under control so it won't grow beyond his personal pervue, which is mechanical CAD/CAM, without all the bells and whistles that other CAD companies get distracted with.

IMO, if you want the best (but certainly not cheapest) parametric solid integrated mechanical CAD/CAM system around, you can't do better than Anvil.

Hooker <--- who has no current association with MCS or Anvil

per the original request... killer app.. how about this. (bit of pure fantasy ) part hardware, part software.

add a cnc mig welding and a cnc manipulator to the set of tools that can run in the cnc machine.

then you could perhaps build up complex parts from different pieces without unmounting the primary piece. or you could perform weld build-up and machine-down type of repairs on known parts.

additionally you could add a rep-rap type of head that could build up plastic parts in place, on existing metal pieces.

for complicated shapes and surfaces you could have an EDM head... you could also have grit polishing heads...

and lastly, you could integrate a simple inexpensive 3d scanner system that could autmatically perform in-place qualilty checks, and in conjunction with the weld system, fix any errors. (reducing waste)

another idea might be to design a type of software and hardware that can bend light sheet metal into custom shapes, punch holes, spot weld together... etc.

basically anything a fixed robot can do... wiring harnesses, assembly, should be fair game for cnc programming. after all cnc really is robotic. Think outside the toolpath!

On another tangent, I was/am a designer, working in facilities that purchases machined products from local machining companies. I am a 3d drafter and quality manager. I would like to be able to have an electronic connection between my files and revision control system, and the machinists cnc code.. and have that certifiable and auditable, as perhaps part of a document vault system.

just some ideas, as I don't know too much about it.

Chris

Some of these are really great ideas. Polishing or grinding heads are already in use. Amada already has the automated CNC sheetmetal machinery as well as hybrid machines that do both waterjet and wire EDM. Your thinking must be on the right track. You just have to beat the big boys to it.

I know what you mean about the welding idea because I have had to weld repair mold inserts and then re-machine them but I don't think the machine would be cost effective. Also, I'm not sure I want weld spatter in my machining center. Although come to think about it, it just might be a great application in a shop where they repair shafts on a lathe.

Just curious, why do you want access to the machinist's CNC code?

thanks for the comments. in addition to taking away material, or adding material.. the only other things I can think of are either changing material properties (ie heat treating, anodizing, etc), coating/finishing material. (ie paint, polish, varnish, ) and maybe inspect (radiographic?)

all that being said, I see similar types of code being required for any type of 'material removal' or 'material deposition' machine.. say for example you wanted to make a giant machine that could build you a house out of concrete and other materials.. you could have it build the house walls in a rep-rap style, and then have it trim and surface the walls to required finish textures and colors. (even artwork)... but who knows... when I get into this headspace, I'm not necessarily connected to reality

as per your question, I don't want access to supplier cnc (as in, to change it), I just want to ensure consistency of data between vendor and customer. From a quality manager's point of view, revision control of any 'document' which impacts the manufacture of the product has a bearing on quality and customer satisfaction. I'm just suggesting a more direct link between the revision number of the cnc code, and the customer supplied drawing revision number and/or 3d model. (I've had extensive quality issues with a particular vendor which brought these issues to my awareness)

I'm aware that this is no guarantee of quality, and does not replace receiving inspections where or when required.. but it is a start. making sure that the machining vendor and their customer are on the same page, sharing the exact same data.

in my case, I built 3d models, and adapted them to fit into the designs.. and they were revision controlled.. we then submitted drawings and prototype parts to the machining company.. who then re-modelled the part, and then transformed that into code.. and all along the chain, there were slight deviations. that coupled with new and poorly trained workers, who did not understand the product, or its context, created extensive quality issues. It would have been simpler if we had been able to transform the original model directly into cnc code. As it was, I ended up just making the vendor completely repsonsible for it all, and insisted that they modify the drawings as we made changes, and then provide us with a paper file copy for auditing purposes. I just kept thinking there had to be a better way. this is the era of wide area networks.. why can't a company have a more directly connected file sharing system between 'custom' part vendors and their manufacturing customers, in a way that is ISO compliant, traceable and auditable... ?? (ramble ramble ramble on )

My reason for the response actually ties in with the original question. There are people out there who honestly believe that if they come up with a sufficiently comprehensive software package that they can eliminate the need for the machinist. I swear that I have met a surprising number of people that think that all you have to do is feed a drawing into a machine and it will make your part.

A CNC code is for a specific machine to make a specific part. There were many decisions that had to be made before the program was written. There are things to check and decisions to make as part of the setup. Cutters must be selected and inspected. Adjustments must be made after first article and adjustments made if necessary during production. So, having the machinists CNC code even if it is referenced to a Rev. number is really usless information unless you plan to look over his shoulder the whole time he is working.

As an example, I have programmed setup and run jobs successfully and then had the shop setup guy run the job the next time around and have problems and want me to change the program. After a bit of investigation I then find that he is trying to run the job with dull tools. Change the tools, problem solved.

You really cannot do your vendor's job for them. I have made this statement before on this forum that once a purchase order is issued against a drawing, that drawing becomes a contract. Your drawing must constrain the vendor to make the product you want. It is their job to meet the requirements of the contract or be replaced as a vendor. It is your job to see that all of your requirements are covered in the drawing/contract so that if the product is "to print" it will work.

TQM, Mil-I-45208, Mil-I-4565, ISO...... are all designed to force companies to produce quality work efficiently. In my humble opinion, no matter what program you have, or what fancy software applications you have there is no substiitute for coompetence.

yes... you are totally right... I should know better.. thank you for speaking the truth so well.

as for replacing machinists... I don't know what is possible and what is not... (not connected to reality ) but I know that the technology will continue to evolve... computing power will certainly continue by moore's law. artificial intelligence research will continue. smart people (like yourself) will continue to innovate... so I predict that some time, the robot will match the human capability, in that, while not perfect, neither is the human, and therefore, the robot will have passed the cnc equivalent of the Turing test. I think that what the human has is good visual feedback and problem solving analytics.

Chris

The technology of computers and robotics can certainly do amazing things and will continue to do more. The more applications you add and the more complicated the parameters the more they can do. However, a person has to determine and design the apps. and fill in the paramaters.

I have yet to see or hear of a computer comming up with a clever idea on it's own.

I have seen and worked with automated systems that have closed loop feedback mechanisms to monitor and update machining lines in real time.

One, specifically, was an engine block machining line that was capable of inline checking of machined surfaces/areas, determining deviance (primarily based on cutter wear), and then feeding back that deviance to the machine tool cutter offset registers to return the operation to a predetermined mean.

In particular, a bore honing operation provided feedback to the hones realtime as the operation was being performed. While certain bores were being honed, probes examined adjacent bores for real vrs expected results and modified the live hone operation accordingly. The system often worked to a tolerance of +/- .0005".

The system also monitored spindle horsepower for each tool and assigned a new tool from an automatic tool changer slot when the power consumption and deviation from the mean of the measurements reached a certain stage.

Lastly the system provided indications to the operators that tooling needed to be resupplied based on wear and/or breakage, and rfid chips in the blocks stored data on the condition of each individual block; deviation from the mean that could be read into downline operations with adjustments automatically incorporated if necessary. Blocks that encountered problems (such as tool breakage) or potential unresolved undercutting or overcutting were flagged in the system (and in their chips) and sidelined for individual attention; this could mean manual machinist intervention and then return to the line, or trashing the block as unusable.

This system happened to be a high precision 80 plus station engine block machining system. It was designed to take a raw cast block in and spit it out at the end ready for assembly. It worked amazing well after 2 years of design, testing and installation bugs, except for one thing!!

The designers didn't take into consideration the affects of housing such a beast in an environmentally uncontrolled plant in the upper Mid-West. The seasonal differences between summer and winter required them to almost completely reconstruct the facility around the line to keep the machines in spec year round. Even much of the floor had to be reworked to keep the system isolated from seasonal ground temperature changes.

Soooo..... IMO, everything written here is technically feasible, given enough money, time and backup manufacturing expertise. To me, replacing a machinist (as I am) is more of a matter of capturing his intelligence and experience and extracting that info to automata as required. Not impossible, just terribly expensive and time consuming. Also, most of that info is already captured, in paper form, in Machinery Handbook, which has been replicated many times in various databases.

To me, there is a time, and need, for both automation and classic machining.

Hooker

You said that very well. You must capture the intellengence, experience, and I might add creativity of the machinist. How many times in the 2 year development of the process did a machinist (probably you) have to look at the result and make a decision based on experience. How many times did the machinist have to say "lets try a 3 flute high heilix of brand B instead of the 4 flute from brand A". How many times in that 2 year development did he try a different insert, hone grit, coolant, drill configuration, peck depth, smaller roughing pass, longer lead in move, faster feed rate to eliminate chatter, add a spring pass. The list goes on and on. With the changing technology in this feild, even the Machinery's Handbook can't keep up.

The scope of the machinist's job may be evolving, but it is far from being eliminated.

So now back to the original question of the blog. New applications are great. Keep them comming, but please get the baisics right first.

The machine you described probably cost a million dollars or more. I bet it was frustrating to try to get your software package to produce a pocket facing routine that didn't waste 10 minutes cutting air at some crazy amount per minute.

You must capture the intellengence, experience,

as well as comprehend it

Yep, it was indeed a multi-million dollar system. And the best part of it, that I neglected to include in my previous message, was that none of the CNC coding was done in any CAM system. All of the G & N coding was done by hand to ensure the most efficient programming possible and full integration to the overhead management system designed for overall monitoring and control. And, yes, hundreds of test engine blocks were machined to ensure efficiency, not only in the coding but also for the tooling. A small price to pay as the system was designed to handle hundreds of thousands of blocks a year.

So I guess what I'm saying is that there is little or no need for a whiz-bang CAM system for a production system of this magnitude. Not only because of the scale of the block system but also because the machining paths were 2 and 2.5 axis.

Now, I do agree with you entirely about "get the basics<sic> right first", but for other applications, say, aerospace!! Low production rate, high complexity parts and tooling do indeed benefit from CAM systems that can get it right in the first few programming attempts, particularly multi-axis. The key to a good system (as you describe, "pocket facing routine that didn't waste 10 minutes..."), IMO, is the ability to dig down into real time G/N code and tweak it there, not in the graphics. This, of course, would necessitate live graphics screen update and instant collision detection inherent in the CAM system, with the target machine's capabilities/limitations tied in (real time post processing). Two way communication between the source database and the output code as opposed to detached output code that is the way of most CAM systems today. That kind of capability prices it way out of the reach of most job shops. Job shops (internal and external) are usually long on human expertise and short on cash for whiz-bang CAM systems.

And, now, it seems we've come full loop...

Hooker

I haven't used Mastercam since version 7 and only had intro training with Surfcam so this may exist by now or may have already existed but I didn't know it.

* Classic mode (just the basics - but up to date and more efficient)

* Module select (hide what you don't want to see)

Staying up to date shouldn't make any software harder to use or less efficient.

Some Press Brakes have the ability to process the entire setup (CAM) directly from a solid model file (CAD). This is the ultimate drag-n-drop where anyone can walk up to the machine with a drawing file on a USB stick and make a good part.