---

Day 3: Wednseday, June 11

Next year I'd like to see Sun continue to provide the workstation (with laptop disk compatibility) and Internet connectivity. But more people than just me report on events (if I'm even there). Would be time enough to work that out just in the month before the conference, when schedules solidify.

Topics:

• Customer issues (individual opinion)
• Education intranet
• Sheetmetal tips
• Managing change and quality with PDM
• Experts panel
• Data exchange and quality (individual opinion)

---

Customer issues (individual opinion)

• proliferation of modules
• the 6 month release cycle

On modules: customers at all levels seem to be increasingly frustrated with the steady flow of new modules, or even the division of single old modules into multiple new modules (sheetmetal headed that way on R19, apparently). Jokes about modules were the most successful jokes of the conference, by far (remember that if you're talking at the conference next year)

But turns out PTC people joke about modules too, at least among themselves, make up silly names for imaginary new modules, just like everyone else. Seems it's just not humanly possible to contemplate the current list of modules (much less the new modules in the works) without laughing, or at least chuckling. That's probably good, gives PTC and customers something in common ("Heard any new module jokes?"), there's some room there to open discussions. Certainly plenty of room to find a common solution, just because there are so many modules to work with.

On the 6 month release cycle: customers seem frustrated by the relentless flow of new versions, similar to the flow of new modules. But here positions seem more static, frozen, nobody is joking. However in that kind of a situation, when change comes (and change always comes, nothing remains static forever) the change is just likely to be more sudden and rapid. A key part of the PTC position was this statement:
 
  The 6 month release cycle improves PTC's internal efficiency

But isn't the big question something else, isn't it this question:
 
  Does the 6 month release cycle improve the internal efficiency of PTC's customers?

For any vendor, if a process improves their own internal efficiency, but decreases the internal efficiency of their customers, that process is at risk. The 6 month release cycle might be in that category.

Could be the 6 month release cycle was more appropriate when PTC customers were smaller, like, 5 people on average per site. Easy enough to upgrade 5 people, especially if they're in earshot of each other. But now we have Texas Instruments still on R16, and another even bigger customer I heard from a PTC consultant jumped from R12 to R17. Upgrades don't come easy when you have 100's of people and dozens (at least) of major processes affected.

Should be a reasonable way to approach the 6 month cycle question. Take that question, "Does the 6 month release cycle improve the internal efficiency of PTC's customers?", and survey a statistically significant part of the PTC customer base. Statistically significant is often surprisingly small.  Survey would be probably best done by a consulting type firm, there are many that do exactly this kind of survey for companies. Then everyone would have some objective data.

(end of individual opinion, just had to put in something of my own after all this writing).
 

---

Education intranet

TI has made major investments in training, which are more fundamental than the intranet itself. Such as, developing a 3 day (only) basic training class---every new user starts with that.  After that class and some experience on-the-job, there's a Workshop.  After the Workshop, and some more experience on-the-job, then there's an Evaluation. And the Evaluation is a one hour written test, and a seven (7) hour practical test on a workstation.

This class/workshop/evaluation sequence continues through three levels, from level A to level B to level C. At that final level C stage, the evaluation is just 4 hours instead of 8. Now that this sequence is in place, the new course development work is for courses for casual users, nonusers, and managers. All the courses are proprietary to TI, not available for others.

People are free to schedule training as they want, but it is also included as a subject in their performance reviews. If someone hasn't completed the training for their job level, that's a subject for discussion with their manager.

The initial home page includes site news, FTP download of appropriate files, special topics (seminars, videos, books), other Pro/E releated links, and the Learning Institute staff.

Then there are separate picks to see the recommended course sequences, check class schedules and register, and review all courses and the prerequisites.

TI uses CADTRAIN's COAch product for self-paced on-line training to follow up on classroom training. CADTRAIN converted their courses to a HTML format so that TI could offer them over the intranet, and also record their use. The COAch modules they offer are Basic Training, Advanced Topics, Surface, and Sheetmetal. After students register, then they can use these modules whenever they have time.
 

---

Sheetmetal tips

The initial deployment did not include any implementation planning. As you might expect, the resulting parts were not acceptable to manufacturing (which included both internal TI sheetmetal manufacturing, for lower volume, and outside shops for higher volume. Bend tables weren't acceptable, punches and notches weren't shop standard, and bend relief wasn't specified correctly. Some of these problems, like tools not being standard, had been around for years before.

The same problems about bend tables and bend relief also appeared on the modeling side, as well as problems with an appropriate startpart. The PTC standard sheetmetal training class assumed some sheetmetal theory which many new users were missing, and it also included material not needed at TI, like, rolling sheetmetal.

So, the first step toward a solution for the problems was setting up a team of people from all the areas affected, including a PTC consultant and TI internal training. This is what they did:

• developed 2 bend tables, for small and large bends, exactly matching manufacturing requirements including standard radii and thickness, and replacing the 3 bend tables that are the Pro/E default
  
  
• defined bend relief to be a notch, a user defined feature. The notch is first inserted into a bend area on either side of the bend, then the bend operation is done with the No Relief option (no relief needed, since the notches are already there). The Pro/E bend relief rip and stretch features were exactly what TI wanted to avoid by using bend relief, no help.
  
  
• developed a UDF library for punches and notches based on manufacturing specifications
  
  
• removed everything they didn't need from the basic sheetmetal class, which was now a custom class, taught by Rand Technologies
  
  
• updated the TI sheetmetal theory class, now suggested before the new customized sheetmetal class
  
  
• developed a sheetmetal startpart, which in addition to the standard startpart goes into sheetmetal mode, and also assigns a bend table to the part
  
  
• developed a sheetmetal modeling practices document, and distributed it on the company intranet

• use standard part parameters for thickness and bend radius, and after creating each wall change radius and thickness to those parameters
  
  
• model the sheetmetal part in bent condition (the documentation itself cautions against trying to model it flat, if you read the documentation)
  
  
• use Extend to make walls meet at corners
  
  
• flatten a part before adding cutouts or other geometry at corners
  
  
• use inside radius when creating walls, but also dimension to outer edge

---

Managing change and quality with PDM

Doug did a useful survey of the audience before he began, a survey that might only be feasible at a conference like this. He asked what people were doing, or not doing, for Pro/E database management, and he got these numbers, as far as I could count:

• 2 did nothing (brave souls, to identify themselves)
• 5 used some combination of Unix permissions and scripts
• over 100 used Pro/PDM or CMS
• over 20 used some corporate PDM standard, Sherpa or CMS or Metaphase

• geometry errors, basic
• frozen parts for datums, which affect modifying or redefining assemblies
• suppressed parts of features, which they do not allow, too difficult to evaluate whether a suppressed feature will be acceptable
• unwanted dependencies, which directly affect data management (a simple part might bring in a 200MB assembly file, just because of an unwanted dependency)
• library part changes or renames, such as, library part changed from part to family table, or, user created library parts
• not checking lower level changes at higher assembly levels, or not saving top level changes in an assembly when working on individual parts

Doug described a nightly script for batch review of a top level assembly:

• fetches the the latest or the production version of the assembly from Pro/PDM
• brings it up in Pro/E
• saves the assembly tree to a text file, with 'status' on, so that the list shows frozen and suppressed parts
• search that text file for words 'frozen', 'suppressed', and 'geom'
• writes out the results as a report

Submit always to PDM to capture a particular configuration, and attach a specification or analysis files or other appropriate documentation.

As an example of what is possible, Doug mentioned a 370MB 900 component assembly with no errors, and only 3 frozen parts (which came from suppliers).
 

---

Experts panel

Here are different points:

• there isn't a way to develop a flat label outline for a recess in a complex surface
• use Copy Draw in Sect Tools to bring imported (AutoCAD) draw entities into the sketcher, fastest way
• clip planes are only temporary, use feature cuts if you want something more permanent
• Pro/Batch has an update, but will remain the same for the future
• Pro/Intralink has two sets of tools, general and sys admin. There isn't a utility now to restore a specific past version of an object, only the latest version
• the major menu changes will be in place by R19, after R19 menu changes will be more subtle
• a PTC Web site guide for producing large assembly drawings will be available there within a couple of weks
• R18 has a search tool to find parameters by their name or a part of the name or their value
• Pro/Mesh has a secure future for associative interfaces to 3rd party applications
• R18 9713 fixes most performance issues with the menu panels, except for NT issues
• R18.1 (9709) introduced a major new capability to separate the graphics representation from geometry and from features. A graphics representation can then be used in large assemblies for speed improvements.
• add draft to a part by using Tweak Draft near the end of the design cycle. One reason is that once a part is drafted it's difficult to compress it to the midplanes for analysis.
• Pro/Designer will be on NT by end of the year, Alpha NT to follow.
• since there are no individual databases in Pro/Intralink, a utility was developed based on alpha testing input to add information to a family parts library
• on 1.0 Pro/Intralink has to be a separate Oracle instance, but on 1.1 it can be installed into an existing instance. Multiple servers are supported through environmental variables.
• R19 of Pro/E accesses the Pro/Intralink database
• R19 will separate sheetmetal into design and manufacturing. It'll include autonesting, with distribution of parts across multiple sheets, and runtime postprocessing. Future changes will be autocontour (for laser cutting) and autopunch.
• difference between holes/cuts/slots? o use holes for circular holes (easy answer) o choice between slots and cuts is up to user, cuts just require an extra menu pick to select the side
• over time, expect Pro/Mechanica to incorporate more Pro/E geometry tools

---

Data exchange and quality (individual opinion)

One of the exhibitors at the conference was Cambric Corp., from Utah (801/571-8100). They have hundreds of CAD operators at different locations, which they need, since they do data exchange jobs in the range of 10000 to 20000 hours a job (that's 5 man years to 10 man years). A job might include converting 12000 parts and 20000 drawings.

Which is reasonable, seeing that one of their main customers is General Motors, doing MCAD data exchange with most of the other car makers in the world and also many other companies. For wireframe and solid data exchange, IGES or STEP could do the job, but to bring data from another system into a parametric solid the only choice now is to build that solid manually.

Cambric has a process which includes a 20 page template for a customer to complete, to define their needs for the result they want. If Cambric just tries to do the best job they can without gathering that initial input, it doesn't work. There are just too many ways you can build a parametric solid, for different end uses. Most of us just know one way, the way we do it, but there are usually other alternatives.

So, measuring overall quality of a parametric solid might not be possible just inspecting the database. You'd have to ask, where is it going to be used, then check the requirements there, then check the database against those requirements.  With 3D explicit wireframe or explicit solids, it was possible to test for database quality just by examining the database itself. But apparently with a parametric solid you can't do that anymore, you do have to look further to say whether the part will fill the needs. Progress brings about more work, that happens.