Low cost, intelligent controls, ‘smart’ factory devices and a raft of maintenance and commissioning tools could ‘explode’ onto the market very soon on the back of Microsoft Windows CE and NT Embedded. Brian Tinham looks at the benefits – and the likelihood – of the sharp end of manufacturing IT finally joining the mainstream A standard ‘open’ operating system – in fact, virtually the same for manufacturing automation and visualisation as for mainstream business IT – is an appealing prospect. Cheaper, a unified development environment and services, reusable components, no proprietary lock-in, easier integration and so on: it could open the floodgates to a plethora of low cost, intelligent machine controls, ‘smart’ factory devices, hand-held maintenance and commissioning tools and so on. And although over the last several years it’s been moving somewhat hesitantly from dream to reality, it seems the stage may finally now be set for a sea change. Heard it all before? The arrival of Microsoft’s Windows NT, its range of ‘real-time’ extensions from independent real-time operating system (RTOS) developers, their own device RTOSs and latterly also Microsoft’s diskless mini-operating system Windows CE – each was ushered in with the promise of a wholesale move to ‘open systems’. If the pundits had been right, we should by now have seen the curiously termed PC-based ‘soft controls’ and open smart control devices (actuators, drives and the rest) leading the sharp end of factory and plant floor manufacturing IT out of the historical proprietary wilderness and into the mainstream of IT. But it didn’t happen. At least not in the numbers predicted by so many who apparently didn’t have their feet on the ground. Although, according to market research firm Datamonitor, PC-based distributed controls software, for example, is the fastest growing sector of that currently $1.9bn global market, it still only accounts for just 16%, despite strong growth in smaller markets. And the picture with ‘soft logic’ is much weaker. Quite simply, the predicted demise of PLCs (programmable logic controllers) and the rest has resolutely refused to come to pass. Now, however, enter uprated Windows CE, with the promise of real-time control extensions very soon (delayed version 3.0 may yet come out this summer), and the larger Windows NT Embedded 4.0 version (NTE, launched last summer). These, say those pundits, will effect the sea change – hastening the open controls revolution. True or false? First, what precisely are Windows CE and Windows NT Embedded – and where is each appropriate? Microsoft’s Web site at www.microsoft.com/windowsce/embedded/ helps. In a nutshell, both are ‘hard real-time’ Windows-based operating systems, with the familiar drag, drop and click environment, but scaled for use in embedded systems that don’t require the functionality or ‘footprint’ of a full-blown PC – like machine controls, motion controls, intelligent sensors, ‘smart’ motors and the plethora of other potentially ‘smart’ devices. Both offer the benefits of a fast, interactive development environment, the ability to run the application on the same platform and the availability of off-the-shelf software components. And most significantly, both provide access to the whole spread of Microsoft COM/DCOM (Component Object Model) and DNA (distributed Internet architecture) for Manufacturing connectivity and the massive range of bundled software we all take for granted. Dick Slansky, senior consultant with analyst ARC Advisory Group, says “NTE targets more complex, lower volume applications where higher end features (like advanced HMIs, security, multi-processing, etc) and interconnected networks are important”. Meanwhile CE, he says, “aims for the smaller embedded applications, where cost, size of memory and low power are constraints” – like control devices. There is debate about which matters most to manufacturing. Slansky sees Windows NTE heading the open IT revolution, driven by the rapidly expanding market for embedded systems in everything electrical – cookers, fridges, dishwashers, cars … you name it: all of which will be going ‘smart’. “RTOS developers want to change their image to embedded systems companies,” he says. “Embedded systems on NTE are going to be huge.” He sees CE as a side-show. “The only reason Microsoft is devoting time and energy to developing CE 3.0 [with real-time functionality] is to give credibility to CE in the embedded market. The soft control market is minuscule; Microsoft envisions CE as part of the embedded systems battle.” Kevin Prouty, senior consultant for manufacturing strategies at analyst AMR Research, agrees but goes further. “Users don’t care which operating system they use – as long as it’s easy, ubiquitous, has lots of tools and is supported by lots of hardware. With NT and NTE you can do that, but with CE, because every device has to be customised that doesn’t apply. So CE for soft control: no.” And, he says, CE is too expensive to take on hundred dollar micro PLCs, smart sensors and the rest. “It’ll be a long time before CE devices are that cheap.” His view: “NTE is the issue: it’s already starting to impact the high end of CE, and it has the benefit of scaleability to small and inexpensive footprints.” And he reflects that established RTOSs like VxWorks, QNX and RTX are going to take some beating. But Bob Butts, the ever pragmatic marketing director of industrial software integrator Boward Computer Services, has a different outlook. He says that in the industrial world Windows CE is the only one to watch. “CE 3.0 will be the platform to go for in control. NTE is simply too big for most smart systems: it would have to be pretty complex; NTE will be a niche OEM market!” He insists that CE 3.0 will be the turning point for open control, open smart devices and so on. “When CE has gone real-time and it’s had a year to shake down, then it will (metaphorically) explode,” he says. He’s convinced this is what the market has been waiting for – a small, familiar RTOS with full Microsoft connectivity. Uwe Henze, product manager for automation firm GE Fanuc, doesn’t expect an ‘explosion’, users being too conservative. But he says, “there is one clear trend: vendors don’t want to invest any more money in operating systems: it’s a major investment. We want standardised OSs and we’ve chosen CE because of the connectivity.” He emphasises the general disappointment with Microsoft’s progress to 3.0, saying: “Our hardware is ready to go; we are CE ready, but Microsoft is not”. But he says neither price nor hardware are problems: “Microsoft can change the price of CE; we will negotiate the price we want for our major product lines .... We develop our own hardware and sell in volume.” Butts and Henze also believe that CE will sweep away other RTOSs – like VxWorks from market leader Wind River (which with acquired ISI’s Psos claims 45% of the embedded market) and others – that have been around since the early 90s. Butts: “If IBM couldn’t defend OS/2 against Microsoft’s DOS-based Windows, when it was technically superior, what hope have smaller companies got of defending their RTOSs against CE 3.0?” It’s a good point well made, but moot nonetheless. ARC’s Slansky says: “I don’t think Windows CE will have much impact on the RTOS companies – and I don’t think Microsoft gives a rip. The two main issues are footprint and license costs. Right now it’s a bit high: about $30, compared to VxWorks and others at $1.5—4.” And while he agrees that Microsoft will be attractive “in that it ties right into the Win 32 API and then DNA, Windows 2000 and all that,” it’s not alone. “Wind River already has VxDCOM, incorporating the COM/DCOM object model. So you can have Microsoft connectivity without relying solely on Microsoft. They realise the future of DNA too.” So let’s take stock. There is certainly a market for open visualisation and open control. Slansky, who was formerly a senior controls engineer with Boeing for 28 years, says: “At Boeing I wanted open systems; I tried to integrate ‘best of breed’, but I couldn’t do it – because of the proprietary OSs.” But Butts and Slansky both say ‘open’s’ earlier failure to thrive was the inadequacy of what was on offer. Going open wasn’t compelling enough: it was still a ‘horses for courses’ decision – with HMIs, visualisation and supervisory control scoring well on full NT and Windows 9x before it, but live control remaining mostly ‘no go’. As GE Fanuc’s Henze says, users select control systems against criteria: like “response times, reliability, operating system, networking, visualisation as well as long term maintenance, service and supply of spare parts”. Not just price and connectivity. And with these criteria, the well-rehearsed arguments for soft control over conventional were less convincing, not to say specious. Fantastic price/performance of PCs, for example, falls at the first hurdle of stability and robustness for plant installation – you just can’t use cheap PCs, and they can’t compete anyway with ultra low cost micro- and nano-PLCs. And there are issues surrounding the ruggedised I/O racks. When CE 3.0 and/or NTE is bedded down, however, will this change? Slansky says yes: “It’s over the hump. Siemens is already leading the pack, with Rockwell, GE Fanuc and the rest close behind. And the open control packaged software providers are offering whole suites too: companies like Wonderware and Intellution with toolkits and ‘total solutions’.” And Henze says: “CE has a future for the ‘little bigger’ PLCs.” With hard real time a given, and all the rest right, it’s all about the new buzzword, ‘total integration’. And with the emphasis on near-real-time information transfer from shop floor to top floor – or indeed to anywhere – signing onto the Microsoft DNA architecture becomes very attractive: it is now compelling. “Moving live data from the controls to the enterprise and doing something useful with it is the point,” asserts Slansky. And you can do it with any of the Microsoft-centric hardware or software vendors: “it only depends on how well they do it.” But there are still doubters. Richard Sturt, business manager Controls and Information Group at Rockwell, is one. He says that although Rockwell is about to launch Windows CE-based visualisation called ‘View Anywhere’ – allowing HMI screens to be developed on PCs and downloaded to anything from plant PDAs (personal digital assistants replacing laptops for maintenance, commissioning and troubleshooting) to machine control HMIs and PCs for SCADA – he still doesn’t see control on CE ‘exploding’. “We’ll be introducing soft PLC on CE later this year or early next, depending on when CE 3.0 is released, but we only expect it to expand soft control take-up, currently on NT. The drivers aren’t there for open systems in control. Most people we speak to still want to lock it [the OS] away so it can’t be touched.” Sturt concludes that, integration notwithstanding, most will see open controls, whether CE or NT, as “too big a migration”. Omron’s manager of factory automation systems Michael Juniper concurs. “PLCs are nothing more than constrained computers. We spend a fortune on ensuring reliability so that users get solid, maintainable integrated hardware and software to run their critical controls. And the value of that still applies. Soft control is still only appropriate where flexible visualisation is the primary driver – not for mission critical control.” And there are parallels here with the arguments for IBM’s integrated AS/400 workhorse for enterprise systems. So the jury’s out. Meanwhile there is another way – at least to gain loose integration: the Web. AMR’s Prouty again: “While engineers weren’t looking, companies like Rockwell and Schneider slipped Web servers into their PLCs. This sleight of hand prepared many factories for a Web-based architecture without engineers even knowing it.” And he adds: “While some concerns still abound about the reliability, speed and security of Web-based applications, hardware and software developments have made these points moot.” Beyond the obvious – that Web browsers can take over from traditional PC-based HMIs (providing visual access to consolidated plant information without the expense of full integration) – he says the point is this “lowers the bar for [integrated] automation system architectures”. Because with embedded Web servers what we’re talking about is gaining an instant, low cost route to what’s being termed ‘highly distributed control’ (HDC). So whereas the long running, standardless fieldbus digital factory/plant network debate had been making it difficult and costly to contemplate HDC with disparate plant level systems and devices, using Web servers instead you can get that information flowing. According to Prouty, the only weaknesses remain the analysis and management tools and managing the multitude of devices. He says that although Web development tools do exist the issue is integrating these with the tools that engineers and IT in manufacturing use. Nevertheless, there is general agreement that if what’s wanted is simply loose coupling, or quasi-real-time monitoring and control, then the Web can provide a solution, given embedded Web servers or converters. But that’s not promoting openness. And before we get too excited, it’s not a panacea. Boward’s Butts warns that there are few standards on the Web. “It’s technically very fragmented: there isn’t a standard environment, so you have to choose from the Microsoft, Sun, IBM ... tools.” Ironically, he says that’s another reason why Windows CE will take over. Why? “Because Microsoft owns the desktop and the browser – and CE dovetails with that too.” And he explains: “It’s easy to create Web pages, for example, in Windows Front Page. So if you want reports, key operating parameters and graphs, you can use an ODBC compliant database; you can query it from an active server page, reload every 10 seconds and it’s available anywhere free of charge. But then you have to host it under NT – because of all the bundled Microsoft functionality on the Web server.” And that’s the point. We’re back to easy connectivity – but only if you go the Windows DNA/DNS (Microsoft’s digital nervous system concept) route. “No-one else can do this,” asserts Butts. No-one else has the financial, marketing, or R&D muscle.