After noticing the continued banter and discussions regarding the use of guard traces in both the EMC and SI discussion forums over the past months, I decided to consult a couple experts on the subject – Howard Johnson and Eric Bogatin. I summarized their thoughts in my latest blog posting on the Test & Measurement World web site: http://www.tmworld.com/blog/The_EMC_Blog/41806-Guard_Traces_Use_Em_or_Not_.php. I invited both to add any additional comment, if they wished. Feel free to add to the discussion.
Guard traces are typically grounded at both ends to the signal reference plane. There are certainly situations where guard traces can help. For example, for low-frequency audio – especially for two-sided board designs, guard traces can reduce crosstalk by an order of magnitude. However, on modern-day high-frequency digital designs, guard traces may help, but only if implemented correctly.
Crosstalk occurs when the magnetic lines of force pass from the aggressor trace under the victim trace. In other words, the lines of force must encircle the victim (Faraday’s Law). As the distance between the aggressor and victim traces increases, the coupling decreases, as you might expect. Henry Ott, in his latest book, Electromagnetic Compatibility Engineering (2009), summarizes succinctly when he states, “crosstalk between adjacent microstrip traces is proportional to the square of the trace height divided by the square of the separation distance.”
There’s a great article by SI expert, Howard Johnson, explaining how guard traces work. For more detailed info on the pros and cons of guard traces, please refer to the T&M World link above.