Ironcad: not re-creating dated 1980s thinking!

Are today's Solid Modellers productive tools, or have most CAD software vendors blindly copied dated 1980's modelling software which defeats many of the primary objectives of using CAD?

Are today's Solid Modellers productive tools, or have most CAD software vendors blindly copied dated 1980's modelling software, that whilst being very clever, defeats many of the primary objectives of using CAD? Alan Goswell of Leonardo Computer Systems gives his views.

The 1980's saw a new revolutionary crop of Solid Modelling products emerge.

The new genre featured 2D semi-intelligent sketches, which when dimensioned, and given rules concerning geometric relationships, magically re-jigged themselves into a scaled and dimensioned shape.

This was a very impressive trick, which harked back to the pencil sketch, except that the sketch was now accurate, and could easily be adjusted to new sizes, simply by adjusting dimensions.

Parametric modelling had been made simpler.

Unfortunately, this brilliant new trick had some downsides too.

The relationships between dimensions and geometry had to be carefully defined, otherwise the trick would not work.

This was partially solved by making the software cleverer, it prompted the user to apply the requisite number of dimensions, angles and other 'constraints' that defined it fully.

But even so, on very complex parts it became somewhat difficult to determine what was needed.

Another downside was, that if the design needed to changed, it was often time consuming and difficult to make alterations.

This was particularly a problem where insufficient thought had been given to what might change.

The new crop of modellers used constructive solid geometry, where models are built up using 'features', like blends, holes, bosses etc, that were applied in a strict order.

This has great advantages because individual features can be easily adjusted in size and character at a late stage in the design (adjusting tapers and blend radii to suit the requirement of a mouldmaker etc).

So in effect, these features became parametric too.

But problems could be encountered at a late stage in the design where an alteration had to be applied, which caused features to interact in an unpredictable way destroying the stability of the model.

Almost all of these 1980's modellers, and their 1990's copies have also failed to solve the problems of working in 3D space.

Most rely on a set of '2D work planes' that have to be positioned in 3D.

Thus, the primary 2D 'sketches' that define extruded and lofted shapes, are constructed on 'work planes' which must themselves be positioned and orientated at the correct angle by rotating each one around its axes.

Complex parts can have literally dozens of 'work planes' defined which often have to be named and saved with the file.

Switching on several at once, and having to interpret a wireframe outline, can produce a view of the modelling scene which is very difficult to interpret.

To make matters worse, the software is usually too slow to allow the user to work on a fully shaded model, so realistically, a wireframe view with wireframe workplanes is the only view available.

Many of the 1980's /1990's modellers are not capable of constructing multiple parts in a single file.

Each part therefore has to be constructed separately and finally brought together in a assembly file.

(Known as 'bottoms up' design).

In the design of a big machine with many parts, having to construct each model part by part, each in its own file, is an unnatural and slow process, far removed from the way that designers work.

Just think back to the drawing board, where the design was almost always worked up as an assembly drawing, with adjacent geometry being used to provide starting points and faces for mating parts.

Without this approach, tedious to-ing ad fro-ing between files to check on size and form becomes the norm! Constructing a new part by trimming adjacent faces with boolean modelling operations is impossible.

There is yet another 1980's chain to bind the designer.

Each part must have a geometric 'mate and align' relationship to fit with a mating part, in the correct manner.

So, 1980's models consist of multiple files each with its own set of features, each with a positional relationship with other parts, developed from 2D sketches with complex interrelationships between dimensions and geometry.

Thus with a 1980's modeller, or clone, instead of designing parts and their relationships naturally together, starting from an assembly, the designer must now face working with a wire frame image, surrounded by outlines of work planes, with the distraction of whether he has too many or too few constraints, and worry whether features are in the right order.

He also has the chore of constantly switching between multiple files to pick up, and check on, geometry of mating parts, at the same time bearing in mind each part's relationship with another.

It is no wonder that the commonly accepted learning curve for such modellers is 9 months! It is no wonder that many people, who just want to get on with the job of design, and not concern themselves with irrelevant distractions, are beginning to question this muddled state of affairs, and look for better solutions.

Whilst introducing some excellent ideas at the time, this 20th Century method of working involves the designer in an almost ritualised distraction, often with no practical benefit whatsoever, since a parametric capability is often unnecessary.

This cumbersome modelling method can seriously divert him from his main purpose, which is to determine and lay down a functional (sometimes visually attractive), economical and produceable design, in the least possible time, which is in itself easy to alter and adapt as the design requirements evolve or change.

Unfortunately, many of the 1990's Solid Modeller authors have failed to see the wood for the trees, and many have blindly aped the technology of the 1980's, rather than innovating.

Virtually all of them have followed the path of the originators of this ritualised tedious and cumbersome pattern of what could be termed as 'Solid Muddlers'.

Fortunately there is at least one exception.

The story of how it came into being is as follows.

A few years ago there was a $500 Solid Modeller called Trispectives.

It had a unique, friendly solid modelling user interface.

The user worked in fully rendered mode continuously, overcoming wire frame confusion.

Trispectives also had superb photorealistic and animation capabilities.

Trispectives' patented Triball cursor was a revelation, there had never been anything so easy to use.

The Triball solved all the problems of working in 3D space in a stroke, anything could be positioned and orientated in 3D space, without workplanes, using one simple yet versatile tool.

CAE Magazine, called the Triball "the most useful tool in the history of CAD".

3D models were built up fast, using 3D shapes, dragged and dropped from fly out catalogues located at the side of the screen.

But...

Trispectives, whilst addressing the needs of the 'design' community (TV logos, and 'concept art' models etc), was not able to produce drawings, and was not parametric.

It didn't have the features one would expect to see in a serious engineering CAD modeller.

In 1997 the biggest Mechanical CAD dealer in the US (VDS now called Alventive) merged with 3D EYE, the makers of Trispectives.

With finance from a source connected with Microsoft, Alventive set about transforming Trispectives into a 'dirt under the fingernails' modeller with all the bells and whistles (linked drawings, constrained geometry, parametrics, features based etc).

Like others in the Solid Modelling world today, Alventive built in the very best third party modelling technology (ACIS kernel, constraint engine, IGES translator), and added in their own special features, the very different, but outstandingly simple Trispectives user interface.

They called it IronCAD.

Now into its 3rd year of development, IronCAD has evolved into a superb designers tool.

Rather than re-creating the 1980's thinking embodied in so many other products, Alventive has kept designers needs firmly in mind.

The constraint based parametric design features have been kept in their rightful place - available when needed, rather than applied as a matter of course.

To ensure that IronCAD had the very best modelling capability, Alventive added in the Parasolids modelling kernel as well as ACIS, thus ensuring that IronCAD had superior modelling, and modification capability compared to any other mid-range modeller.

They also recognised that the ability to communicate was of vital importance and added in the widest range of data exchange tools ever seen in a solid modeller (native in ACIS and Parasolids), as well as a built in web and VRML capability, so that realistic models can be published and viewed by anyone.

But, most of all, Alventive kept, almost unaltered, the ability to design easily, naturally and accurately, from easily recognisable rendered 'drag and drop' shapes, in true 3D, from within a single file.

In brief IronCAD offers: Superior modelling capability with dual ACIS / Parasolid kernels.

Freedom from having to design using slow, constricting 1980's 'design intent' modelling methods.

Design mating parts and assemblies naturally, within a single file.

Work faster in true 3D with intelligent 'drag and drop' solid shapes.

Banish complex datum planes.

Make modifications at any stage, even on imported parts.

IronCAD is 21st Century Solid Modelling! With apologies to all those 20th century modellers.

You know the ones we mean!.

Back to top