By Dan Hebert, PE , senior technical editor
Each May, OUR AutomationXchange event gives our machine builders a chance to share best practices with one another, and to meet privately with senior personnel from leading automation suppliers.
Before the event, we interview the machine-builder attendees about their main areas of interest and upcoming investments. The information helps to formulate a customized agenda for each attendee that includes guest speakers and exclusive sessions where machine builders can share best practices.
One of the subjects of common interest looked at building versus buying. This generated a best-practice discussion centered around human-machine interface (HMI) software.
Many of the machine builders had horror stories to tell. One of them said his company decided to hire an outside firm to write its HMI software based on an internally generated specification. The results were very disappointing. The budget and schedule both more than doubled from initial estimates. The builder says his companys HMI software now is beholden to two guys working out of their basement.
Another attendee also had a bad experience outsourcing his HMI software development, but he largely blamed his own company for the problems. The specification generated as a basis for software development was incomplete, and was revised constantly throughout the project. No surprise that the project exceeded budget and schedule, though his company generally is happy with the final product.
The machine builders who were happy with their homegrown HMI software had sufficient internal resources to develop and maintain applications. The consensus was three to five full-time employees were needed, with the exact number depending on the complexity of the HMI software.
Long-time Control Design columnist Jeremy Pollard provided a great example of an application in which homegrown HMI was the best solution. He was charged with installing a network of 20 PCs, each of which would be running a similar version of a relatively simple HMI program.
Site licenses alone would have cost about $200,000, and annual license fees would have been another $20,000. Jeremy wrote an HMI application in Visual Basic for less than $20,000, saving his client more than $180,000 up front and $20,000 annually.
The OEMs who chose to buy off-the-shelf HMI software also expressed dissatisfaction with vendor claims about no programming required. It seems HMI vendors have banned the use of the P-word and have substituted the term configuration.
Dont worry, they say. Our HMI software doesnt require any programming, just configuration. One machine builder after another said the line between programming and configuration constantly had to be crossed to adapt off-the-shelf software for their machines. This was especially true when the level of complexity exceeded a few simple screens.
Discussion continued around build-or-buy for controls hardware and software. Most participants were content to buy, with the favored solution being off-the-shelf controllers programmed in one of the IEC 61131-3 languages such as ladder logic, function block diagram, or structured text.
Those who favored an internal solution had a formidable staff of in-house experts well-versed in C++, C#, or other high-level languages. Everyone in this category used PC-based hardware.
An interesting subtext was robot control. Many of the machine builders purchased and installed robots as part of systems delivered to their customers. Most had let the robot manufacturer supply the controls. We heard two main reasons for shying away from robot control, even though their machine controllers had more than enough available MIPS. First, most of the robot manufacturers simply didnt offer the option to buy a robot without controls. A second concern was liability for a robot run amok, such as in the movie I, Robot.
One machine builder/system integrator with extensive robot experience said, however, that the situation was changing. At least one leading robot vendor now lets customers control their robots. This vendor sells robots with a lower-cost no controls option that includes all information required to write robot control software.