RFI-Ham(1).pdf

A Ham's Guide to RFI, Ferrites, Baluns, and Audio Interfacing

Revision 5a 5 Jun 2010

by Jim Brown K9YC

Audio Systems Group, Inc.

http://audiosystemsgroup.com

The basis of this tutorial is a combination of my engineering education, 55 years in ham radio, my

work as vice-chair of the AES Standards Committee working group on EMC, and extensive re-

search on RFI in the pro audio world where I’ve made my living. That work is documented in tech-

nical papers and tutorials that can be downloaded from the publications section of my website.

Chapter 1 – Some Fundamentals

To solve interference problems, we must understand them. So we'll begin by describing the ways

that RF interference is coupled into equipment and detected. There are several principal mecha-

nisms at work. You should study this tutorial thoroughly to understand how these things work.

Detection at Semiconductor Junctions Every semiconductor junction, whether part of a diode,

transistor, or integrated circuit, is quite nonlinear, especially in the voltage region where it is be-

ginning to conduct. In analog circuits, we prevent this non-linearity from causing distortion by

properly biasing the circuitry, by using lots of negative feedback, and by preventing the signal from

being large enough to cross into the cutoff region.

Thanks to this non-linearity, every semiconductor junction functions as a square law detector , de-

tecting any RF signal it sees. A good designer prevents detection by shielding the equipment and

its wiring, by filtering input and output wiring, and even by bypassing the junction by a capacitor.

Since virtually all detection that causes RFI follows square law, the strength of the signal detected

by audio equipment, telephones, and other equipment will increase (or decrease) as the square of

any increase (or decrease) in RF level at the detector. In other words, the strength of the detected

RF changes by twice the number of dB that the RF signal changes. This means that if we manage to

reduce the interfering RF signal by 6 dB, the detected audio will drop by 12 dB. This is a very use-

ful thing – it means that we may not need "an elephant gun" to solve many interference problems.

Antenna Action The most fundamental cause of radio interference to other systems is the fact that

the wiring for those systems, both inside and outside the box, are antennas. We may call them

"patch cables" or "speaker cables" or "video cables" or "Ethernet cables," or printed circuit traces,

but Mother Nature knows that they are antennas! And Mother Nature always wins the argument.

When we transmit, some of the RF from our transmitter is picked up by those unintentional anten-

nas, and RF current flows on them. What happens to that current determines whether there will be

interference, and how severe it will be. We know that antennas work in both directions – that is,

they follow the principle of reciprocity – so when RF trash from inside the box flows on those an-

tennas, it is radiated as noise and we hear it on the ham bands.

Fig 1 shows a simple antenna we’ve all used,

probably with our first radio receiver. We con-

nected a random wire to our receiver, and the

antenna current flowed through the receiver to

a "ground" that might have been a driven rod,

but was more likely the safety ground of the AC

power line (the third pin on the AC socket,

known in North America as the "green wire").

Even if the radio was double insulated so that it

didn't require the green wire connection, RF

current still flowed through the stray capaci-

tance of the power transformer to the power

line and made the radio work.

Fig 1 – A simple random wire antenna

RF picked up on the antennas we call loudspeaker wiring, video cables, the coax from the cable TV

system or a rooftop TV antenna, flows through equipment to get to the AC power system safety

ground. Hams understand that some antennas are more effective than others. An antenna that is

Understanding and Solving RF Interference Problems

Page 2

close to resonance will work better than one that is not. Long antennas tend to pick up more RF

than short ones. Think about these fundamental principles when trying to diagnose which cables

are bringing your RF into a given system (or radiating their trash into your receiving antenna).

A path to "ground" or the power system is not always needed to produce antenna action. The whip

antenna on our VHF and UHF handheld radios uses the radio, capacity-coupled to our hand that

holds it, as a counterpoise (that is, to provide "the other half of the antenna"). All that is required

for this to work is that the size of the counterpoise must be a significant fraction of a quarter wave

(or larger) so that it can "sink" the antenna current.

Common Mode and Differential Mode Signals A differential mode signal is one that exists be-

tween the conductors of a cable. At any given point along the cable, current flowing on one con-

ductor is precisely balanced by current flowing in the other direction on the other conductor. The

intentional signals carried by cables are differential mode signals – the audio or video signal in a

home audio system, Ethernet signals on CAT5/6 cable, and the RF signal carried by the feedline

connecting our antennas to our transceivers.

A common mode signal is one that places equal voltage on all conductors – that is, the voltage be-

tween the two ends of the cable are different, but there is no voltage between the conductors. An-

tenna action produces a common mode voltage and current along a cable. The antenna current

induced on audio and video wiring is a common mode signal. That is, with "ideal" cable, there is no

differential voltage between the signal conductors as a result of this antenna action. If the cable is

shielded, nearly all of this current flows on the shield (and skin effect causes it to flow on the out-

side of the shield). If the shield is ideal (that is, if the current is distributed with perfect uniformity

around it), the field inside the shield will be zero, and thus none of this antenna current will flow

inside the cable. Conversely, when a cable shield is carrying differential mode current, as in the

case of coax, skin effect will cause that differential mode current to flow on the inside of the shield.

The real world is not ideal, so most interfering signals will simultaneously exist in both common

mode and differential mode, but in most real world conditions, one or the other mode dominates.

Several cable defects (essentially manufacturing tolerances) certainly can and do convert this

"common mode" antenna current to a differential signal (that is, a voltage between the signal con-

ductors), but that is rarely the most powerful coupling mechanism. One common defect that af-

fects both balanced and unbalanced cables is imperfect construction of cable shields. In even the

best "real world" balanced twisted pair cables, there are imbalances in the capacitance between

"red" and "black" conductors to the shield on the order of 5%. [B. Whitlock, JAES, June 1995] In

balanced paired cables that use "foil/drain" shields, there is even more imbalance in the inductive

coupling between each conductor and the shield. Noise (or RFI) coupled by this mechanism is

called "shield-current-induced noise," or SCIN. [N. Muncy, JAES, June 1995] All three of these

mechanisms convert shield current to a differential signal at system input and output terminals.

If the cable is an unshielded pair (loudspeaker cable, for example), RF will be induced approxi-

mately equally on both conductors (but, depending what the input circuit of the equipment looks

like at RF, current flow into the equipment may not be equal on both conductors). This can also

produce a differential voltage at the input (or output) terminals.

Output Wiring is Important Too! It is well known, for example, that RF interference is often cou-

pled into the output stage of audio equipment – for example, the power amplifiers that feed loud-

speakers or headphones. There is always feedback around that output stage, so RF present at the

output will follow the feedback network to the input of a gain stage, where it will be detected and

amplified. This problem is made much worse when parallel wire cable (zip cord) is used to feed

the loudspeakers or headphones, and can usually be solved simply by replacing the zip cord with a

twisted pair of POC (plain ordinary copper). [Pseudo-scientific advertising hype for exotic cables

notwithstanding, it was shown nearly 30 years ago that #12 copper twisted pair (or #10 for very

long runs) is a nearly ideal loudspeaker cable.] [R. A. Greiner, "Amplifier Loudspeaker Interfacing,”

JAES Vol 28 Nr 5, May 1980] As we will discuss later, the twisting of a pair greatly reduces the

level of RF that the wiring couples to circuitry.

Power Supply and Control Wiring can also act as antennas. When I bought the house I owned in

Understanding and Solving RF Interference Problems

Page 3

Chicago, I upgraded the electrical wiring and put all of it in steel conduit (EMT). This shielded the

wiring, so that only the short power cords between equipment and the wall outlets could act as

antennas. The house I recently bought in California is wired with no conduit, using unshielded

parallel conductors. Thanks to its length, and the fact that it is not shielded, this wiring acts as an

effective receiving antenna for the RF I am transmitting, and an effective transmitting antenna for

the RF trash generated by computer equipment, power supplies for low voltage lighting fixtures,

and even battery chargers.

Current Returns to its Source Current flows in a complete circuit that includes the source of the

current. The circuit will couple noise inductively, and also by antenna action. The cause of many

RFI and noise problems, as well as the solution to them, lies in identifying and controlling these

circuits. Always ask, "Where does the noise (or RFI) current flow?"

Loop Area One of the most fundamental laws of electrical circuits is that the current that is mag-

netically induced between two circuits is proportional to the loop area of each circuit. Making the

loop area small also minimizes the extent to which the wiring can act as an antenna. When we use

a closely coupled pair of conductors to form a transmission line, we are reducing the loop area,

which reduces the total current induced in the loop by an interfering signal, and the total magnetic

field produced by current in that loop. The transmission line, of course, has other useful proper-

ties. More on this later on.

The equipment designer can also use multilayer printed circuit techniques to place a "ground" (ref-

erence) plane next to all signal wiring, turning each circuit trace into an unbalanced transmission

line, where the return current is carried on the reference plane under wiring. A single reference

plane makes a very large reduction in the ability of that circuit trace to receive interference; sand-

wiching it between two such planes virtually eliminates it. These techniques, called microstrip (one

plane), or stripline (the sandwich), are widely used by better designers. They reject noise coupling

both inside and outside the equipment by drastically reducing the loop area of the current path

(and have the additional benefit of making high speed data circuits behave better because they are

transmission lines).

Loaded Words That Cause Misunderstandings One of the most overused and misunderstood

words in electronics is "ground" (or "earth" in British English). There are several important and

common uses of the words. One meaning is an actual connection to mother earth. Some common

earth connections include the steel structure of a building, a buried conductive water pipe, a con-

crete encased grounding electrode (called a Ufer, after its inventor, Herbert Ufer), and, of course,

one or more conductive rods driven into the earth. [Concrete mixes vary widely in their conductiv-

ity – most we are likely to encounter are highly conductive, but some are effective insulators.] The

primary function of this earth connection is lightning protection.

A second common use of the word "ground" (or "earth" in British English) is a third conductor that

is part of the power system wiring that should never carry current (except in the case of a fault) but

connects the conductive enclosures of equipment to a common point within the power system.

This "green wire" or third pin in the outlet in North American power systems, is called the "equip-

ment ground" (or "protective earth" in British English). The green wire is required to be connected

to all exposed conductive parts of electrical equipment "that might be energized" in the event of

equipment failure. The purpose of this connection is to provide a sufficiently robust current path

that a fuse will blow or circuit breaker will trip in the event of equipment or wiring failure that

causes the chassis to be "hot," thus protecting people from electrical shock and preventing fires.

A third common use of the word "ground" (or "earth" in British English) is to describe "circuit com-

mon" or "circuit reference" within equipment. Circuit common should nearly always be connected

to the power supply reference, and to the shielding enclosure of the equipment. If the source of

noise is within equipment, circuit common is reference for the noise voltage (and current), and it is

the point to which that noise current wants to return.

A fourth common use of the word "ground" is as the "return" for an unbalanced antenna like a ver-

tical or long wire. In this application, the antenna needs some conductor to be a low impedance

"sink" for the antenna current. The radials for an elevated or ground-mounted vertical antenna

Understanding and Solving RF Interference Problems

Page 4

serve this function. There is an excellent discussion of this in http://w2du.com/Chapter05.pdf

Ground Wiring Some hams like to think of the earth as if it were somehow a "sink" into which all

noise can be poured, never to bother us again. Indeed, you'll find lots of bad advice to solve RFI

problems with "a better ground." In fact, nothing could be further from the truth. An earth connec-

tion is rarely part of a solution to RF or noise problems. Rather, we need a better understanding of

the four common meanings of “ground,” that “ground” is not a single point, and that connections

between them change current paths. [There are exceptions to every "rule," and this no exception.

See "Shunting Common Mode RF to Earth" later in this tutorial.]

Consider a noise filter hanging between some piece of noisy equipment and the power line, with

capacitors from the "hot" and "neutral" to "ground." What is that "ground?" It is circuit common and

the shielding enclosure of the equipment, the green wire in the power cord, which is connected to

the equipment ground in the power system, which goes to the breaker panel, which is in turn

bonded to neutral and a real earth connection at the service entrance to our building (and, if we've

done it right, there should be a bond between the power system ground and any grounds we've

added for our radio equipment). In most systems, the green wire follows a rather long path – typi-

cally a quarter wave on 80 meters, and perhaps even on 160 meters. That current path is an an-

tenna , and any RF current flowing on that conductor will radiate! In fact, the connection to earth

may increase current flow. Like any other radiated RF signal, our receiving antennas will hear it. All

of those "ground" connections must be present to have a safe installation, but it is the combination

of the high series impedance of the filter's choke and the connection between the filter's "ground"

and the shielding enclosure of the equipment (and it must be very short) that suppresses the noise.

The earth connection provides lightning safety .

This basic scientific fact has major implications in the design of filters intended to prevent noise

coupling from noisy equipment to our ham stations. If we add a filter to wiring that enters or

leaves a piece of noisy equipment, it is the shielding enclosure for that equipment to which any

"ground" of our filter should return (and, of course, circuit common should also be connected to

that shielding enclosure). All connections between the filter and the noisy equipment should be as

short as possible (what my old EE professors liked to call "zero length" to emphasize the impor-

tance of making them short). Why? First, to minimize the loop area, and thus the inductance.

Second, to minimize antenna action. More about this when we discuss specific filter designs.

Insufficient Input and Output Filtering As hams, we know that equipment needs good input and

output filtering to prevent RF from coming in on input and output wiring. Beginning in the 1950's,

hams operating the HF bands were deluged with TVI complaints because television manufacturers

failed to include high pass filters in their sets. Likewise, audio equipment needs good low pass fil-

tering to reject our signals. Many myopic designers of "high futility" audio gear (and even some

professional gear) don't include low pass filters because they don't want to degrade the phase re-

sponse of the audio path. While good phase response is certainly important, so is RF rejection.

Good engineering can satisfy both needs without compromise. Ever since those early days, hams

have always assumed that a good low pass filter will kill RFI in audio systems, and a good high

pass filter will kill interference to FM and TV. Unfortunately, while good filtering is important, other

mechanisms are far more important in most real world situations.

Shield Resistance adds hum and buzz to unbalanced wiring (audio, video, and data (RS232). The

"green wire" at every AC outlet is at a different potential, thanks to leakage current of equipment

plugged into that outlet, as well as other leakage current flowing on the green wires. When that

equipment is interconnected with unbalanced wiring, the difference in potential (60 Hz and its

harmonics, plus noise) causes current flow on the shield, and the IR drop is added to the signal. A

"beefy" shield (big copper) minimizes R – that's why the best video cables use heavy copper

shields! Audio transformers eliminate the hum/buzz by breaking the current path at DC and audio

frequencies, but most hum and buzz in your ham station can be solved without a transformer –

simply power all interconnected gear from the same outlet, bond equipment all chassis’ together,

and use coax with beefy copper shields for audio. An unshielded transformer can make matters

worse, coupling noise from a power transformer into its unshielded windings. See Chapter 8 for

more.

Understanding and Solving RF Interference Problems

Page 5

The Pin 1 Problem: The most common way

that hum, buzz, and RF interference enters

equipment is via a design defect first widely

understood by the pro audio community

thanks to the work of Neil Muncy, (ex-W3WJE).

He named it "the pin 1 problem," because it is a

mis-wiring of the shield of audio cables – pin 1

in the XL connector commonly used for pro

audio, but it is just as much a problem in un-

balanced interfaces of all types, as shown in Fig

Fig 2 – The Pin 1 Problem

The proper connection for a cable shield to equipment is the shielding enclosure (chassis), but

products with a "pin 1 problem" connect the shield to the circuit board instead. Nearly all con-

sumer equipment, including even the most expensive "high futility" gear, is built with pin 1 prob-

lems. Virtually all computer sound cards have pin 1 problems. So do most RS-232 interfaces and

nearly all ham equipment – indeed, almost all RFI problems we describe as "RF in the shack" have

pin 1 problems as their root cause!

Fig 2 illustrates both right and wrong connection of the shield. The trouble-free connection on the

right goes straight to the shielding enclosure, so shield current flows harmlessly out the safety

ground on the power cord. Any noise (or RF) on the cable shield stays "outside the box."

The connection on the left, however, is a pin 1 problem. Current flowing on the shield bypasses

the shielding enclosure and is forced onto the "ground bus" – that is, "signal common." To get to

the power system ground, noise current must follow that "ground bus" around the circuit board –

what Henry Ott calls "the invisible schematic hiding behind the ground symbol." The wires and cir-

cuit traces that make up that invisible schematic have resistance and inductance by virtue of their

length, and the IZ voltage drops across those R's and L's are coupled into each "gain stage" that

connects to the ground bus! Once that happens, every semiconductor junction that "sees" the RF

will detect it, and succeeding gain stages will amplify the detected RF.

What if there is no "shielding enclosure?" Fig 3a and 3b shows how to avoid pin 1 problems with

unshielded or partially shield equipment. (Of course, unshielded equipment has other potential

problems, which we'll talk about later.)

Fig 3a – 120VAC power

Fig 3b – 12VDC power

Why is equipment built with pin 1 problems? Two reasons. First, "fuzzy thinking" on the part of

engineers, who have lost track of where noise current flows. Second, the construction techniques

used in modern equipment, and the connectors built to support those techniques, make it more

likely that pin 1 problems will happen. In "the old days," we mounted an RCA connector or phone

jack by drilling a hole and screwing it down to the chassis. Today, those connectors come with

solder tabs for mounting directly to a printed circuit board, which is then built, tested, and fitted

into an enclosure. Screwing those connectors down to the enclosure increases cost significantly!

And FIXING pin 1 problems in equipment having this kind of construction can be quite difficult.

RFI and Pin 1 Problems There are three ways to cure RFI coupled by pin 1 problems. The first two

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: