I’m sure that like me, many of you have one or more stories associated with power supplies and the consequences of not getting something right. That could be as simple as undersizing a power supply or forgetting to check the available capacity because we are “only adding one more loop,” or overheating because it was mounted either too close to walls or at an incorrect orientation.
As you can see, it’s quite easy to quickly come up with a list of how the plain old power supply can be the source of many surprises. I recall that based on their experience in troubleshooting fieldbus network problems, physical layer manufacturer Relcom has claimed that 80% of the time, the reason for the difficulty was due to the power supply or grounding. As we know, Foundation fieldbus (FF) and PA Manchester encoded systems require a “power supply” (really an isolator) between the bulk 24-VDC power supply and the network. This is because, if the bus power supply is not installed, the linear bulk power supply will be working to maintain a steady 24 VDC while the bus communications are trying to communicate using a ±0.75 V or larger signal on top of the base voltage.
Despite the requirement for FF power supplies, after making a presentation in Medellin, Columbia during the early days of FF, one of the audience members came up to ask a question about why his FF network was not working properly, as it had a lot of noise. Having seen a number of home-made terminators on other projects, I asked about terminators and then, of course, what FF power supply he was using. His response was, “I need a FF power supply?” and then went on to explain that he was simply using a conventional bulk power supply. I suspect the reason he saw any signal at all was because he had a small network of limited size.
Not all power supplies are created equal, and in fact, some of them can create quite a bit of noise on their own. Another story to demonstrate this was when I was working with a colleague using a USB-based oscilloscope on their laptop. Though we were able to measure a useable signal with the laptop plugged in, we noticed quite a bit of baseline noise. On a whim, we unplugged the computer so it was running on battery, and magically, the baseline noise disappeared. What made us suspicious is that we were working in an office environment that should not have been affected by noise from other machines. Because of this lesson, whenever I’m using a laptop-based measuring device, I always do so while running on battery.
Power supply and possible ground loops also are a big concern in my present role at CIMA when we are migrating from one control system to another, since, in many cases, the I/O cards and power supplies of 25 to 30 years ago don’t always route signal power the same way as today’s newer systems. Since the intent of the system we use is to not disrupt the signal, having a complete understanding of how the system works is critical to our success.
Though I offered to answer any additional questions for the gentleman in Columbia if the FF power supply didn’t work, I didn’t hear anything further, so I assume the problem got solved. I’m still not sure how the gentleman in Columbia even got his system to communicate. Perhaps ground was very good, and of course, because Manchester encoding measures the transition rather than the peaks, the devices on the small network were able to overcome the minimum voltage threshold, which can be ± 0.25 V.
Despite the importance of power supplies, other than digital network supplies, most systems have no means of providing warning of impending failure. Designers resolve in their minds that installing a redundant power supply or sourcing from a UPS that is fully monitored overcomes the need to monitor at the field level. However, if you don’t monitor, even with a common trouble alarm that requires a dedicated signal to a remote location, how will you know if your redundant system is no longer redundant because one power supply failed? I don’t believe that everyone in this situation conducts regular preventive maintenance on all their cabinets, especially the more vulnerable field cabinets powered from lighting panels and similar, to see if the “light is on.” Some suppliers do provide the option of intelligent power supplies or an external monitoring system that can report back, though I suspect the number of such installations is limited. If we want to move to the vision of remotely distributed I/O, controllers monitoring power supply integrity will be one of the requirements, because when it comes to control signal reliability—surprises are not wanted.