Last month I talked about room resonances—things that ring and rattle when you make noise in their presence. This time I’m going to address things that make noise on their own—mysterious buzzes, clicks and thumps; voices from the ether that appear in our recordings unbidden; omnipresent hash that fills the room with weird shrieks. In short, I’m going to talk about electromagnetic interference
I’m not going into ground loops and noise in the power lines, and I won’t get into the special problems of electric guitars and amps; they’re worth an article all their own. I will cover the problems endemic to complex recording systems in an electromagnetic jungle.
An RFI Primer
Broadly speaking, we define Radio Frequency Interference (RFI) as any electromagnetic wave from a few kilohertz up to several gigahertz that creeps in where it’s not wanted.
At radio frequencies, components behave peculiarly. Balanced transformerless inputs, which reject common mode garbage nicely at audio frequencies, lose their balance and act as open windows to RFI. Normally clean op-amps (especially those with bipolar inputs) turn into rectifiers that demodulate radio signals all too well. Capacitors begin behaving like inductors. And RFI leaks into every open port in a piece of equipment: inputs, outputs, and connections to the AC line or wall wart.
Where does it come from?
Broadly speaking, any time electrons are in motion, electromagnetic energy is generated. RFI, for example, can be generated in large quantities by a spark. (Early radio transmitters used electric arcs—in effect, continuous sparks—as a source of radio frequency radiation, which they filtered and amplified at the carrier frequency of choice. Radio operators on shipboard inevitably carried the nickname of ‘Sparks’ for this reason.) If you shuffle across the carpet and touch the doorknob, the resulting spark can be heard as a click in the output of all too many pieces of audio equipment.
Semiconductors turning on and off rapidly can generate RFI. Digital circuits, for example, generate pulses of RFI as each gate turns off and on. Diodes in the power supplies of most electronic equipment generate a broad spectrum of RFI that radiates throughout the chassis. (Audiophile manufacturers are beginning to use diodes that generate less garbage, and report significant decreases in the sound’s harshness.) The triacs and SCRs used in light dimmers and motor speed controls also produce RFI prodigiously.
Then there are deliberate sources of RF: radio and television stations, CBs and ham transmitters, and radio-controlled gadgets of every sort. They become RFI when they sneak into audio circuits.
What does it do?
RFI’s effects can range from blatant to insidious. At worst, obvious problems are overlaid on audio: 60Hz hum and buzz, CB calls appearing on PA systems or tapes, radio stations playing through a guitarist’s stomp boxes. (Oh, it’s happened to you, too?)
RFI can also have more subtle effects. Several years ago, it was shown by experiment that RFI can combine with audio signals in amplifier circuits. If the circuit is nonlinear (as most bipolar transistors are at radio frequencies), the RFI and the audio will intermodulate to create new audio signals not found in the original program. In other words, distortion.
This distortion usually shows up as hardness and harshness in the high frequencies. Sibilants splash, guitar pick noise is exaggerated, strings become steely. Sound familiar? This is the rap solid-state gear has had to endure from the beginning, and digital equipment still lives under its cloud. Much of the problem may be traceable to RFI intermodulation.
Luckily, it can be fixed.
Getting a buzz on
Thomas Edison worked for years to perfect an electric lighting system that was clean, efficient, and bright. For some reason, modern Americans feel compelled to make it dim again. There are several ways to do that, but most people use cheap dimmers built into floor lamps or installed in switch boxes.
Cheap dimmers switch the line current on and off at high speed using triacs or silicon-controlled rectifiers (SCRs). Since they’re operating directly on the AC line, they inject large quantities of switching noise into the circuit. This noise, amplitude modulated at 60Hz, is a prime cause of RFI in audio equipment.
The first solution to the problem, of course, is to stop using dimmers. If you need dimmer lights, why not use a smaller bulb? Or a lampshade? Or turn off some lamps? At the very least, you can leave the dimmers all the way on or off while recording (most cheap dimmers produce the worst noise when they’re halfway up).
The second solution is to buy high quality theatrical dimmers. These incorporate filters that eliminate most of the switching noise; unfortunately, they cost a lot of money. So, alas, do large rheostats and variable output transformers (Variacs); the latter can also generate mechanical hum. I’ll leave discussion of those to Ethan’s article, where he’ll talk about how to install them to the best effect if you want to get seriously dim.
Get off the line
The first step in dealing with dimmers is to get as far away from them as possible, electrically speaking. Your studio should have its own electrical circuit for audio equipment that is not shared with anything else—especially dimmers. If possible, this circuit should be on the opposite electrical phase from dimmers and other sources of RFI (see below). In most buildings, the left and right sides of the breaker box correspond to the two phases. (Unless you’re experienced in house wiring, have an electrician install any new circuits. You could burn your house down.)
Isolating your gear can alleviate dimmer RFI problems, but it may not solve them; those short pulses can radiate throughout a building’s electrical system. In fact, I’ve heard of dimmers that were so bad, their garbage leaked into audio gear in other houses fed by the same utility pole!
So the next step is to filter out the garbage. Most electronic supply houses and surplus stores sell RFI filters intended for AC lines (Corcom is a popular brand). These are available in a wide range of shapes and current ratings, including devices intended to replace IEC power-cord sockets in equipment. I use 20A units to power several low-powered devices when necessary, building the filters and 3-prong outlets into small mini-boxes.
Most of the power strips claiming to incorporate filtering aren’t terribly effective. On the other hand, several audiophile and pro audio manufacturers sell high quality power conditioner boxes that filter out RFI effectively without compromising the performance of the equipment, especially power amplifiers. If you can afford these devices, I recommend them. (I’ve listed a couple of manufacturers at the end of the article.)
By the way, many of the problems associated with dimmers also apply to fluorescent light fixtures. If possible, turn off every fluorescent light in the building when you’re recording.
Turning your radio off
Everyone’s on the air these days. We have broadcast band radio (AM & FM), shortwave, hams and CBers, VHF and UHF television, pagers, dispatchers, cellular and cordless phones, and a bewildering array of cordless gadgets. We’re wading through an electromagnetic soup, and it affects our recordings.
Several years ago, synthesizer pioneer Wendy Carlos (Switched-On Bach) built a new studio in the heart of New York City. She had money, so she enclosed it in a “Faraday Cage,” completely surrounding the room with copper sheets and mesh. She probably has the most RFI-free location in Manhattan, and she said in an interview that when she began working in her new room, she found the noise level in the circuits had dropped by about 10dB. What she had previously thought was intrinsic equipment noise was, in fact, RFI.
You probably can’t afford a Faraday cage (I know I can’t), but you can take steps to make your gear less junk-sensitive. Most of this process involves equipment modification, so please be sure you know what you’re doing before you start—when in doubt, let a professional tech do the work. (This goes double for tube gear, which has high voltages that can kill you quickly and very painfully.) Also, you should assume that modifications will void the warranty. Since most equipment these days has a pitifully short warranty period, this may not matter.
None of the Above: a case study
Susan Urban runs None of the Above, a classic coffeehouse located in the upstairs loft of a Chicago church. Like most small coffeehouses, their budget is tiny and their equipment is cheap. In their case, it hummed. Badly.
Susan asked me to fix it. At first, I looked for a ground loop—the hum was at 60Hz instead of 120Hz, typical of a grounding problem. But I changed my mind when I discovered that the hum changed as I moved the mic cables across the floor, and changed again when I shifted the AC power cord.
The main problem was RFI, in this case from a UHF TV station whose carrier was heavily amplitude modulated at (just under) 30Hz and its harmonics (60Hz, 90Hz, etc.) by the vertical sync pulse. Because it was UHF (ultra-high frequency), its signal could slip in almost anywhere.
And it faced few barriers. The PA head was completely unshielded, with no provision for RFI filtering. It was tailor-made for problems, and it didn’t disappoint.
The first step was to clean up the power. (See Figure 1.) I installed .01uF 1000V ceramic disc capacitors on all three legs of the power cord. I also threaded ferrite beads onto the power leads going to the power transformer. These increase the inductance of the wires, raising its impedance at radio frequencies. In effect, the ferrite beads made it harder for RFI to pass into the circuit, while the capacitors were making it easier for RFI to detour into the grounded chassis. (Ceramic discs are the only capacitor type to use here, by the way—they have low inductance, which means they conduct RFI more readily. Keep the leads short and be sure to use at least 1000V units.)
In parallel with the capacitors, I connected 150V metal-oxide varistors (MOVs). These devices have little effect on RFI, but they help absorb surges (see below).
While I was at it, I connected more .01uF 1000V ceramic discs in parallel with the main power supply capacitors, helping to filter out any RFI that was radiated by the diodes.
Next, I looked at the audio input circuit (see Figure 2). The microphone inputs were badly designed, using a differential amplifier made from a noisy (and slow) bipolar chip, the LM-324. As I mentioned before, bipolar chips have a tendency to act as demodulators when presented with high levels of RFI, not a good idea. Since the LM-324’s noise is comparable to that of a decent biFET amplifier (the TL074), I could substitute the latter chip without making the situation any worse. (This was a lucky break; most of the time, FET-input op-amps are too noisy for mic inputs. But the board’s design made the input so noisy already that a quieter chip wouldn’t have helped.)
But why substitute? It turns out that bipolar transistors are the worst devices in high-RFI situations, tubes are somewhat better, and FETs are the best (as I discovered when designing a phono preamp for my parents, who live one block from an antenna array on Chicago’s John Hancock building). I’ve developed a prejudice that any IC accepting input from the outside world should use FETs instead of bipolars. If this had been happening a few years later, I would have used an AD-713 chip instead of the TL074, but those hadn’t come out yet.
Anyway, I wasn’t through. I installed more ceramic disc capacitors on the power supply traces near the op-amps. The new amplifiers are faster, and I didn’t want them taking off into oscillation, but I also wanted to provide more filtering; remember, the chassis had no shielding. I also installed ferrite beads on the input resistors and bypassed pin 1 of each input jack (the ground pin) to the chassis with still more .01uF 1000V ceramic discs. (I clamped the capacitors’ leads to the chassis using the nuts holding the input jacks in place.)
I did the same for the output jacks, connecting the sleeve of each 1/4" output jack to the chassis with a ceramic disc. Finally, I cleaned all the input and output jacks, as well as some internal pin connections, with DeOxit, my favorite contact cleaner, and treated them all with PreservIt, my favorite contact preservative. These removed the oxide from the connectors (metal oxides are great, nonlinear rectifiers). Incidentally, while I said I wouldn’t say much about guitar electronics here, I will note that the DeOxit/ PreservIt treatment works wonders at getting radio stations out of stomp boxes.
The result? Still the same not-great mixer, but without a trace of hum or buzz. The inputs were still hissy, but the sound was noticeably cleaner, at least in part due to reduced RFI-induced distortion. (I lowered the hiss level slightly by substituting metal film resistors for the original carbon composition units.)
Go Thou and Do Likewise
I deliberately picked an example of very bad equipment to show how RFI-proofing is done, but the process applies to gear on every level. A surprising amount of equipment aimed at the pro-audio and project studio markets is designed with little or no resistance to RFI, and the results (nasty voices intoning “That’s right, ol’ buddy” in the midst of a take) are infuriating.
The basic steps are: clean up the power inputs (ceramic discs, MOVs, and ferrite beads), clean up the audio inputs and outputs (discs and beads), substitute FET-input chips for bipolars when possible (see my article ‘Clean Up Your Gear, Part 3,’ May 1996), bypass the internal power supply rails with small polyester caps or, in severe cases, more ceramic discs, and use contact cleaner and preservative on all the jacks. Use it on all your cords, too.
Speaking of cords, not all audio cables are created equal. I’ve had particularly good experiences with two brands, Gotham and Canare, designed for minimum RFI. (I’ve heard similarly good things about Mogami cable, but haven’t tried it.)
Gotham cables use a conductive plastic outer layer for improved shielding, while Canare uses “star-quad” construction, with four wires carrying the signal instead of two. These wires are twisted together in a way that provides improved hum and RFI rejection. Like most cables, the Canare carries the ground in the external braided shield, while the Gotham has a separate internal ground wire, which I prefer. (However, the Gotham cable’s grounding doesn’t work with certain mic preamps. The Rolls/Bellari RP220, for example, buzzed fiercely when I hooked up my Gotham cables, but was quiet as a lamb with Canare and other conventionally-grounded cables.)
TDK has recently introduced a product that sounds like a winner: ferrite rings that clamp on the outside of cables to improve RFI rejection. I haven’t tried them yet, but the idea makes sense, especially in dealing with digital garbage (I’ll get to that shortly).
The ultimate RFI protection for input and output circuits is, unfortunately, the most expensive: transformer coupling. Talking about transformers is a great way to start a fight (I can see Scott Dorsey warming his flamethrower up already), but they are unsurpassed thus far at rejecting RFI in a balanced circuit. Good line-level transformers like the Jensen JT-11P-1 run about $65 apiece.
Incidentally, all the measures I’ve outlined in this section also help keep out dimmer buzz.
I mentioned early in the article that digital circuits generate RFI when their gates turn on and off—which they do many times every second. This hash, if it isn’t dealt with, radiates into the room and causes interference with radio and TV receivers, which is why the FCC requires type-approval of all digital audio equipment. Unfortunately, it can also radiate through the digital piece’s AC cord and travel through the house circuit to your other audio gear.
We all use a lot of digital equipment these days: DATs and digital multi-tracks, reverb units and effects generators, samplers and synths, and of course our computer-driven workstations. This dumps large quantities of hash onto the AC line, to the detriment of the analog gear that shares the juice.
I’ve gotten concerned enough about this that I now power all my digital stuff through a separate Corcom line filter. This filter is symmetrical, so it works in both directions, and I’m pretty sure I hear the difference when it’s in the circuit. (Most of my analog gear is home-built, with extra attention paid to RFI-proofing, so I’m already a step ahead of the game.)
A certain amount of digital garbage is also radiated down the audio output line by some gear; the TDK clamps mentioned above were expressly designed to help filter out this hash. Several audiophile friends have testified that they do, indeed, help—and that their digital playback sounds sweeter with the clamps in place.
Pop goes the furnace
When large inductors like electric motors are connected to the AC line, they draw a sudden jolt of current. (They draw another one when they’re disconnected.) This jolt causes a voltage dip that travels down the AC line to your gear (and dims the lights); as a result, when the furnace, AC, or fridge turns on or off, many pieces of recording equipment go “pop.” This is not acceptable, unless you plan to turn off these essential items every time you record. In most climates, this isn’t practical.
Isolating the studio circuit(s) from the motor-laden circuits can help, but often the jolt travels through all the wiring in the building and out to the power pole. The RFI-protection methods mentioned above (beads, discs, and MOVs) help minimize the effects of motor-induced jolts. (The relays that turn the motors on and off can also spark. More RFI.)
An electrician can often install a snubbing device on the offending motor that minimizes the jolt, and can bypass a sparking relay with a similar device. (Don’t, don’t, don’t do this yourself. Please. Get a professional.) Computer-type surge arresters have a reputation for working poorly in audio systems, but the dedicated audio power conditioners mentioned earlier include surge-arresting circuits that do a good job. Manufacturers of power conditioners with good reputations for audio work include Adcom (908-390-1130) and Audio Power Industries (714-545-9495), makers of Power Wedges.
Paul J. Stamler is a producer/engineer and session player in the St. Louis area.