Engineers learn a lesson about feature creep when a remote control fails
Design engineers inadvertently impact the functionality of a garage door’s remote control and learn about feature creep when they introduce a “new and improved version.”
Daryl Gerke, PE, Kimmel Gerke Associates, Ltd.
If it weren't for the "law of unintended consequences," I might not be able to make a living chasing down EMI problems. And it continues to amaze me the new ways those consequences get created.
An example is the lowly garage door opener. You push the button, and the door goes up or down. No more getting out of the car in the rain or snow. Simple and reliable, that is until "feature creep" enters the picture. Which is exactly what happened to a client of mine awhile back.
The garage door controller worked fine until new features were added. As I recall, the change was the addition of a photosensor to augment the existing contact-autoreversing feature. Rather than redesign the entire electronics board, engineers incorporated the new features on a small board that plugged into an existing socket.
Pretty clever, right? Yes, except prior to shipping, one of the engineers installed the new card in his own garage door opener, and immediately noted that the remote control did not work unless it was within two to three feet of the opener. Warning to engineers: Always check "improvements" before shipping to customers.
So how did I identify the problem? Since I suspected interference from the new board, I first scanned the opener for radiated emissions with a spectrum analyzer and antenna. In this case, the remote transmitter operated around 350 MHz. Sure enough, with the feature board installed, emissions at the transmitter frequency were about 40 dB above the levels without the board installed. That would certainly explain the loss of receiver sensitivity.
The next step was to examine the feature board. It was a small, two-layer board (about one inch square) with a 1 MHz clock. But even better for generating EMI, the clock trace was routed around the edge of the feature board, creating a loop antenna.
But should this really be a problem? After all, this was 350th harmonic, with only an inch of antenna. At 350 MHz, a wavelength is just under a meter, so the loop size alone was starting to get pretty efficient. Furthermore, the "unintended transmitter" was only a couple of inches from the garage door radio receiver. Finally, the harmonics did not even need to land on the exact frequency of the remote control -- just being close was enough to desense the receiver.
So how did I fix the problem? Actually, it was pretty simple. I cut the clock trace, and installed a short jumper. Instead of a one inch loop, the total loop was now about 1/100 the size. This reduction corresponded nicely to a 40 dB (100X) reduction in emissions.
To further confirm the fix, I employed some highly trained help. My then-teenage son was dispatched down the block with the remote, clicking away until the opener would no longer activate while sitting on my desk. At about 200 feet, the experiment off was called off. Success had been achieved.
How could this sort of situation have been avoided? New features are regularly added in software, of course, but adding extensive hardware features usually means a board redesign. In this case, the changes were pretty simple. Furthermore, the strategy allowed engineers to continue production with the existing design, and they could also sell the add-on board as an upgrade for older units already in the field.
Had the add-on board been multi-layer (power and ground planes) instead of a two-layer design, the problem might never have emerged. Alternately, if they had just checked the layout, they might have seen what a great little transmitting antenna they had build on the board. To put a positive spin on it, the engineer who designed the board did not realize that he or she was a pretty good (but unintended) antenna designer. (Time to add a new section to the resume!)
A few years later, I ran across one other "add on board" expansion for a piece of test equipment. In that case, engineers had pulled the existing microcontroller, and replaced it with a daughter board that had a new controller and some additional I/O features. Since most of the other board stuff stayed the same, this allowed them to easily upgrade existing equipment in the field.
We've done some EMI "add on boards" for I/O, adding filters or transient protection. Typically, these are added internally as "extenders" in cables or on connectors. For DB connectors, add on connectors are readily available from several vendors that do the same thing. Once again, great for retrofits.
As an aside, did my teenage son become enamored with engineering? No, he went into finance, where he now happily counts dollars instead of electrons. Maybe another unintended consequence?
Author Daryl Gerke, PE, is a partner in Kimmel Gerke Associates Ltd., an engineering consulting firm that specializes in EMI/EMC issues. Daryl resides in Mesa, AZ, and can be reached at firstname.lastname@example.org, or at www.emiguru.com