As most design engineers well know, a big part of their job is playing a kind of real-life Sherlock Ohms. Here’s a great tale from Technical Consultant Wayne Miller on a fiendish design problem he was confronted by — a case that even the fictional Sherlock might find most perplexing:
“I was working as a Staff Consultant for a small microwave company. One of my contributions to the company was a design for a line of very low noise phase locked oscillators featuring the company’s signature air strip resonators. The line was going well under the leadership of a super tech who knew it well and could fix any problem they encountered. I was therefore surprised one day when the Director of Engineering came to me in a panic: “The phase locked oscillators are all very noisy at cold operating temperatures.“
I walked slowly over to the lab. As I approached the set up, I could see the noisy spectrum analyzer display from across the room. I could see some other things about the set up as well. The set up included a refrigeration-driven temperature chamber and was set up with a stack of power supplies so that multiple units could be run over temperature at the same time. I approached the set up, carefully adjusted the stack of power supplies so that they rested squarely on top of each other and walked away. Problem solved.
The Director of Engineering, who had followed me nervously into the lab, couldn’t believe what he had witnessed. He detained me for an explanation:
The temperature chamber was connected to the power main through a long, heavy cable that ran across the floor. Not only was it a long run, but it was connected to a different breaker than the rest of the set up. The power supplies were stacked so that the top one was touching the side of the chamber. The ground of the chamber was noisy and isolated from that of the supplies, and the contact between the two coupled noise into the power sources and modulated the oscillators. The condition was made even worse by the fact that the contact was intermittent and affected by the vibration of the compressor on the chamber. The problem didn’t tend to show up hot because the compressor wasn’t running.
AC grounds are good for personal safety, but horrible for electronic equipment. Never depend on them for a ground return. Try to connect all parts of a set up to the same AC source to minimize the impedance between the various grounds. Unless there is a necessity to do so, never connect a power supply to the frame ground, even though it is a common practice to do so. It will generate indeterminate ground currents, feeding voltage offsets and noise into a set up. And most of all, remember what you third grade teacher told you, “Neatness counts.”