Tuesday, February 24, 2015

Bypassing Capacitor Lore

As an eager new homebrewer still learning the fine details of how RF circuit stages are designed, I find the wealth of information online and at my finger tips to sometimes be overwhelming if not even contradictory. One topic that I have found to be of the latter category is that of decoupling capacitors.

Working with low speed digital logic circuits in the past taught me the importance of adequate decoupling. Stabilizing DC supply lines is one thing but decoupling high frequency RF stages is a whole new ball game.

Many different respectable sources have described that the trick to proper decoupling is to pair a low value capacitor with a higher value one. The idea behind this seems to be that a low impedance path to ground will be presented to a wider range of frequencies due to the the reactance of the pair. Oppose to a single value capacitance (and its reactance) decoupling a smaller bandwidth. Although I may have not described it all that elegantly, the idea seems perfectly logical to me.

Soon after learning such a juicy nugget of homebrew wisdom I found out that sometimes the logical may just be lore.
Looking for a better explanation of how to select the proper capacitance values to achieve a wide decoupling bandwidth I turned to my copy of EMRFD.* Page 2.28 if you're following along at home. For those without a copy I present the following excerpt below:

Traditional lore tells us that the bandwidth for bypassing can be extended by paralleling a capacitor that works well at one frequency with another to accommodate a different part of the spectrum. Hence, paralleling the 470pF with a .01-uF capacitor should extend the bypassing to lower frequencies. ...when tested... The Results were terrible! While low frequency bypassing is indeed improved, a high impedance response is created at 63MHz. This complicated behavior is .. the result of inductance (of the capacitor). Each capacitor was assumed to have a series inductance of 7 nH. A parallel resonance is approximately formed between the L of the larger capacitor and the C of the smaller. The Smith Chart plot showed us that the impedance is nearly 50-ohms at 63MHz. Impedance would be even higher with greater capacitor Q. This behavior is a dramatic example of lore that is generally wrong!   Bypassing can be improved by paralleling. However, the capacitors should be nearly identical. This was observed by using paralleling a 390pF with a 560pF capacitor. Only a hint of resonance was observed, not significant enough to cause any complications. These anomalies disappear as the C values become equal. The ideal solution is to place a chip cap on each side of a printed circuit run or wire at a point that is to be bypassed.

 Matched capacitor pairs form an effective bypass over a reasonable frequency range. Use two parallel chip (SMD) capacitors of nearly identical values (i.e. 390pF + 560pF) at the points to be bypassed on each side of a wire run or strip/track on the circuit board.  Two .01-uF disks had a reactance magnitude less than 5-ohms from 2 to 265MHz. A pair of 0.1uF chip SMD caps with wire leads attached were even better, producing the same bypassing impedance from 0.2 to 318 MHz. Even better results can be obtained with multi-layer ceramic chip capacitors.

I am beginning to better understand the importance of not just following the traditional design wisdom but to actually experiment, test and measure. The terminus of the homebrewing experience I seek is the ability to truly understand, witness and control the ethereal forces of nature within a system of my creation. I can take it on faith that the circuit I build is working as the designer intended it. I can take it on faith that when I follow the principles of good RF design, the stage that I've just built will perform as it should. Or I can measure, test, experiment and truly develop an intimate understanding of the nature & behavior of the circuits I build. As a homebrewer on a budget, I may not always have the test equipment necessary to make such observations, but I will always strive the greater understanding.

  If you would like to share your thoughts on decoupling RF circuits or can contribute to a better understanding of the above, please share your comments below. Do you have a spectrum analyzer to perform the same measurements? Give it a try and let us know your results, maybe they missed something critical when they performed the experiment and you will be the one to find it.

Thank you for taking time to read my post.

*'Experimental Methods in RF Design' by Wes Hayward W7ZOI, Rick Campbell KK7B & Bob Larkin W7PUA

NOTE: If you do not have a copy of EMRFD, I highly re

commend that you obtain one before it meets the fate of Wes (& Doug DeMaws) earlier work 'Solid State Design for the Radio Amateur' which has been unfortunately discontinued,