Seductive software?

With the fast growing realisation that PCs offer an unbeatable combination of versatility and value in control applications, the idea of developing real-time extensions to the Windows (or any other operating system) kernel is, for software developers, a very seductive one. Unfortunately, it so often leads to a situation where engineering development is completed and a project sits waiting for a control system, sometimes for months or even years.

For developers, the real lure of a project which involves writing real-time extensions is that it is interesting, and perhaps will keep them in work for a long time. However, the arguments put forward to justify such a development to management are more likely to be that any software will be tailored to meet a company's exact requirements, and once it has been developed, there will never be any licence fees to pay.

As with all attempts at seduction, however, the consequences need careful consideration before taking the plunge. The truth is that developing real-time operating system extensions is a complex task, and it will take many programmer-years of effort to get it right. Many who have attempted the task have found it not too difficult to produce a system that does, say, 95% of what it should, but getting that last 5% is both taxing and time-consuming.

This protracted development cycle has important consequences. First, it means that development costs are high and difficult to estimate accurately in advance. The question that needs to be asked, therefore, is: how many licences for a proven real-time package could be purchased for the cost of carrying out an in-house development?

The second consequence of a long development cycle is delay in bringing to market products that will use the new system. Almost everyone must know of companies that have missed excellent market opportunities because of delays in software development.

Even when new software has been developed, and is working reasonably well, the story is far from over. That software is now going to have to be upgraded on a regular basis, and be maintained for many years. Nice work, perhaps, for the development team, but potentially a major headache for the company behind the development. Quite apart from the ongoing costs associated with providing a realistic degree of support, what happens if one or more key members of the development team leave the company? How much will it cost to recruit suitable replacements, assuming that they can be found?

Those software licences for established software must, by now, be starting to look rather inexpensive! But does off-the-shelf operating system software really deliver everything that's needed to turn a PC into a powerful and reliable real-time industrial control system? There are well-proven systems available now that are being used in tens of thousands of applications for important end users like Volkswagen, Ford Motor Company and Whitbread. They are exceptionally versatile and allow multiple virtual PLCs to be implemented on a single PC.

Beckhoff's TwinCAT, supplied in the UK by Hayes Control Systems, is an example of this type of system, where the same PC can even host multi-axis motion control systems. But, what about those "costly" software licences? In fact, the one-off price for a TwinCAT licence is around euro1,100 (£800), with even lower prices for multiple users. How does that stack up against the costs of in-house development?

And does using an off-the-shelf system mean that your in-house software team is redundant? Of course not. It simply means that their time would be better spent on what they are best at - using their specialist knowledge of your market sector to develop application software which gives you a clear competitive advantage.

In short, I strongly recommend that you resist the seductive appeal of developing real-time software. Instead, adopt an established, proven software package so you can concentrate your own development efforts in areas where your in-house know-how will produce real benefits. After all, re-inventing the wheel is a most unlikely path to success and riches!

Hayes Control Systems
p101@industrialnetworking.co.uk