|UltraCAD Design, Inc|
Barry Olney's comments about the "disastrous" effects of orthogonal trace corners, where "electrons virtually fly off the sharp corners of the bend" ("Design For EMC", Jan. 96, p. 28) reminded us of the various ways our customers have tried to solve that problem through the years. We thought we'd share some of them with you.
One customer asked us to point all orthogonal trace corners towards one corner of the board. Then he installed an electron grabber there. This caught the "flying" electrons and dropped them to the power plane. The beauty of this arrangement was that as frequencies went up, more electrons were captured and returned to the plane, thus increasing efficiency. At very high frequencies, efficiencies were so high (so many electrons being returned to the power plane) that no external power supply was required at all!
Of course, some systems contain both analog and digital electrons. These MUST be kept separate. The solution is to point all digital orthogonal trace corners to one corner of the board and all analog orthogonal trace corners to the opposite corner of the board. Then the respective digital and analog electrons can be grabbed and returned to their own respective power planes. An electron reflector should be used at the plane boundaries to ensure that the digital and analog electrons can't mix.
Another customer wanted us to chamfer the corners. Then, the electrons, rolling along like little marbles, would simply reflect around the corner instead of reflecting back or flying off.
Finally, a customer concerned about electromagnetic energy being emitted at a sharp point (which begins to resemble a poorly designed antenna) asked that the edge of the trace be sculpted, much like the serrated edge of a knife. The electromagnetic waves then reflected in so many different directions at the same time that they all got confused and canceled each other out.
As Olney does point out, however, the easiest way to avoid all these problems is to miter the corners. That is why knowledgeable designers NEVER EVER use orthogonal trace corners. And then, no one has to watch out for "flying" electrons.
Douglas G. Brooks
To: Douglas Brooks
From: Pete Waddell
Thanks ever so much for the technique tips on controlling those pesky flying electrons. It seems to me that there is something in this perpetual energy rejuvenation at high frequencies, but I can't quite grasp it. But I would be careful, were I you, that the energy companies don't find out about it yet. Remember the rumors of a 100 mpg carburetor? Also I thought you may be interested in a large mayonnaise jar full of spare electrons that I recovered several months ago, from a poorly designed board. I have not sorted the analog from digital electrons, but thought you might be interested, if the cost of separating them is not too great.
I'm glad to see that people out there are concerned about these issues, and hope that we can bring problems such as uncontrolled flying electrons to the attention of OSHA. I've heard that electrons flying from a board at high enough frequencies can cause severe damage if people are not wearing the necessary protective gear. One technician in Peaudunque, Ohio was nearly castrated by a ricochet electron that flew off of a board at the speed of 1.8 gig. These things CAN be prevented. Perhaps you would care to write up an article on preventing accidents in high speed designs?
Thanks for the fax, keep in touch.
Thanks for your response to my letter to the editor re flying electrons.
I understand your not quite being able to grasp the incredible efficiencies we were able to achieve with the electron grabbers, but that is not the first time I've been involved with such feats. Funny you should mention the 100 mpg carburetor. In an earlier life I was president of a predecessor company called UltraCAR Design, Inc. But, that's another story.
I have bad news about your mayonnaise jar that you filled with flying electrons. It is my understanding that it is impossible to recover analog and digital electrons after they have been allowed to mix. By the way, I hope the jar was clean. If not, electrons covered with mayonnaise are REALLY messy. The semiconductor guys call that "mayonnaise doping." If the jar was clean, there are three problems with catching them. First, the little critters are REALLY fast. Once they are unconstrained, they are hard for normal humans to catch up with. But more than that, being fundamentally different, the results are uncertain when they collide with each other, which they most certainly will when locked up in a jar.
In some cases, since they are fundamentally opposite, they simply cancel each other out and leave a void. In other cases, they combine as a useless hybrid. But, when they combine, the combination is twice as heavy as is allowable under the laws of physics, and they shed (give birth) to another ill-behaved, although very noisy, electron. Jars containing flying electrons MUST be vented to accommodate the changes in pressure as a result of these possibilities; otherwise they may either explode or implode, depending on which is the predominate effect. You are right, flying electrons can be dangerous critters, and they should only be handled by people experienced in such matters.
These two effects are precisely why analog and digital electrons MUST be kept separate in PCB designs. If they combine and then cancel there is an obvious change in the electrical signal. If they combine and generate new, ill-behaved and noisy electrons, these play havoc during FCC compliance testing. Good signal integrity and low EMI emissions cannot be obtained if analog and digital electrons are allowed to mix.
The story about a near castration from a flying electron reminds me of another board we did for someone. I never wanted to talk about it since we treat such information as confidential. Also, we didn't want to be known as accomplices. A professor from a local school asked us to design lots of orthogonal trace corners all pointing ... actually focused is a better word ... at a point about 18" off the board. When he energized the design he was going to use the incredible energy of the flying electrons to cut into a bank vault at the local branch of Seafirst Bank. Unfortunately --- or fortunately, depending on your viewpoint --- he accidentally turned it on before going into the bank and destroyed the ignition system of his getaway car. He was arrested for blocking traffic and was subsequently discovered.
Since then, we either point the flying electrons into a corner of the board where they can be grabbed and used beneficially, or simply leave them trapped in their traces by mitering the corners. Designers who don't follow these rules may inadvertently be endangering their customers and also risk FCC compliance testing failures. I wonder if the IPC has included this type of risk in their designer certification exams?
Note: Readers might want to refer to our Design Note on 90 degree corners entitled "Right Angle Traces" for a more enlightened discussion of 90 degree traces!
Click on logo to return to UltraCAD's Home Page.