An 80-person manufacturing plant has had problems with a particular production machine. For example, they’ve replaced nine circuit boards in it already this year. The plant manager called the manufacturer, and the support tech said he’d never encountered such a problem before. However, he said it sounded as if voltage transients were taking out the circuit boards. He suggested finding the source of intermittent spikes and sags, correcting the underlying problem, and then seeing if the rate of parts failures starts to drop off.
The plant manager called your firm, and your boss assigned you to the project. What are some elements of troubleshooting this problem?
Ground connections
The support tech may have identified an issue, but his solution may not be practical because plants are full of transient voltage sources. For example, every motor that starts and stops across the line will contribute to this problem. Still, the first step in troubleshooting this is to characterize the power supply. Put a power analyzer on the feeder to that particular load. If you can secure it against theft, leave it connected to collect data for a few days. If you can connect another one to the branch circuit(s) for this equipment, it may provide more insight.
Installation errors are more likely here. Why? Because circuit boards are on the load side of a DC power supply. In addition to a transformer, the power supply should have a Zener bridge or other voltage regulation scheme, plus filtering capacitors and coils (tank circuit). It most likely has its own transient protection circuit, too. The power supply should provide more than ample protection against normal electrical noise.
However, undesired current can enter the electronics by finding a different way in than through the power supply. Look at each chassis where circuit boards are mounted. You may see several terminals or connection points labeled with an identifier that includes the word “ground.” For example, there may be chassis ground, power ground, signal ground, and common ground (or just common), but not one of these is actually ground. Art. 100 defines ground as a connection to the earth.
Begin by determining whether neutral and ground are interconnected anywhere. This is a common mistake, but correcting it may involve significant wiring. You will need to estimate the time for repairing this. Proceed methodically.
Next, look at the chassis ground. The chassis is a nonconducting metallic body, which means it needs to be at the same potential as other significant nonconducting metallic bodies. Else, you get differences of potential. If the chassis is connected to ground instead of to an equipment bonding conductor, it’s not at equipotential to other metallic objects and, thus, undesired current is circulating.
It would not take much of such current to zap a circuit board. This fact is why computer techs wear wrist straps alligator-clipped to the computer chassis; their hand is then at the same potential as the chassis, the chassis is at the same potential as anything plugged into it, and the tech’s hand subsequently won’t be introducing current that zaps a motherboard or expansion card.
And it could be the plant’s very own techs are inadvertently zapping those cards through poor electrostatic discharge practices. Ask to see the preventive maintenance procedures and if a maintenance tech can perform one or dry-run one as you observe.
If part of this equipment is actually connected to ground (e.g., somebody drove a ground rod) or connected to building steel as the sole return path or bonding path, you have a big red flag that not all the ground connections in the equipment are correct. Identify the deficiencies and estimate the time needed to fix them.
You’ll need to review the information from your power analyzer(s). If you have a higher-end model, you can let it convert the raw data into useful information. If it shows that significant transients appeared on the feeder or branch circuit, the source is likely in the plant. And it would likely be a large motor starting across the line. Typical culprits include fire pumps, plant air compressor motors, and stamping press motors. A scrap grinder subjected to overload also may be a contributor.