Well, “The Taming of the Shure” (Recording, May 2006) seems to have started some discussion about microphone loading. As a follow-up, the Editor asked for a more general article about what happens when a microphone feeds the input of a preamp or console, and how that input’s characteristics affect the sound of the microphone
This is the short version, not the extended dance mix. Real engineers will groan at my oversimplifications, but if I put in the whole story, it’d take eight pages, and eyes would glaze from here to Simferopol.
How microphones and preamps affect one another varies with the type of microphone. I’ll start with the easy ones.
These have become ubiquitous in recent years, from cheap (e.g. Oktava MC012) to pricey (Neumann TLM series). In these mics, what talks to the preamp is a simple electronic circuit, typically one or two transistors and support components (resistors, coupling capacitors, etc.).
The circuit’s behavior is governed by Ohm’s Law, one of the foundation stones of electricity:
Current = Voltage / Impedance
A quick look at the equation tells you that the lower the impedance, the more current you’ll draw for a given voltage. Most transistors produce more distortion when they produce more current, so a lower impedance load means higher distortion.
In the real world, this means lower impedance loads are usually bad for transformerless condenser mics, and in fact most manufacturers specify a minimum load, below which you should not go. Typically, this is 1000–1200 ohms. Okay? Okay. We’re done with the easy part.
Moving-coil dynamics—the setup
These are a whole different kettle of worms; to understand them we need to take a short detour into the murky swamp of Filter Theory. Don’t panic! There will be no fancy math here, just a couple of pictures and a bunch of words.
To understand the high-frequency behavior of a moving-coil mic (that’s what the preamp/ console affects most noticeably), it’s useful to think of it as a lowpass filter system. This means lower-frequency sounds will be converted to electrical signals, just like a mic is supposed to do. Above a certain frequency (the cutoff frequency) the electrical signal will be progressively rolled off; the microphone becomes less efficient at converting sound to electricity at high frequencies, until the signal eventually trails off to near-nothingness.
In a lowpass filter, how the system behaves around the cutoff frequency is determined by the damping of the system, abbreviated as “d”. We’ll talk about the physical meaning of that in a moment. (Real engineers, instead of talking about damping, instead use a mysterious quantity called “Q”. Q is nothing more or less than the reciprocal of damping, or 1/d; they prefer Q because it makes the algebra neater. I’ll stick with “damping”, because it’s easier to visualize in physical terms.)
Bears and springs
Take a look at Figure 1, which I call the Three Bears diagram. This shows the frequency responses of three lowpass filter systems; they all have the same theoretical cutoff frequency, but each has a different amount of damping. The dotted line (Papa Bear) corresponds to very low damping (d = 0.5), and you’ll notice the frequency response rises to a peak before starting its inexorable way downward. The dashed line (Mama Bear) represents high damping (d = 2.0); it has a “droopy” response, which starts declining well before the cutoff frequency. And the solid line (Baby Bear) represents an intermediate level of damping (d = 1.414); its response is neither peaky nor droopy, but flat up to nearly the cutoff frequency, after which it goes down like the others. (See the sidebar “Magic Numbers” on page 40.)
What does this mean in the real world? First of all, it’s important to know that this description of filter behavior applies equally to mechanical devices, electrical circuits, and things which combine both, like moving-coil dynamic microphones. (It was, in fact, a giant breakthrough in engineering when scientists realized that mechanical and electrical systems could be analyzed using the same mathematical tools.)
So let’s look at a strictly mechanical system, a weight dangling from a spring (Figure 2). Gravity pulls the weight downwards (says Newton), the spring pulls it upwards. At a certain point, the forces balance, and the weight stays there; we call that the rest position. The forces exerted by gravity and the spring are called restoring forces, because they act to restore the weight to its rest position when it’s pulled away.
Let’s disturb the system: move the weight up a few inches, then let go. The spring’s force is lessened (a spring’s force is directly proportional to its degree of extension), so gravity is pulling harder than the spring, and the weight will be pulled downward toward its rest position.
You’d think the weight would simply move to the rest position and stop, as the two forces balanced, but it doesn’t do that. The weight is moving and it has momentum; it wants to keep going. So it does, overshooting the rest position by a few inches until the restoring force of the spring brings it to a halt and starts it upward again. You can guess what happens next: instead of stopping at the rest position, it overshoots.
This would go on forever, or until the sun expanded into a red giant and vaporized the apparatus, except that there is one more force to contend with: the internal friction of molecules inside the spring, which converts some of the system’s energy into heat as the weight boings up and down. (There’s also a bit of friction from the air as the weight moves through it.) The friction siphons energy from the weight-and-spring assembly, and the weight moves less and less with each boing, eventually stopping at the rest position.
Very well, but what if you don’t want the spring to keep boinging up and down for that long? What if you want the weight to return to its rest position with only a little boinging?
What stopped the boinging in the first experiment was the internal friction of the spring; if you increase the friction, it should bring the boinging to an end more quickly. One way to do that: stick the whole apparatus into a bucket of water. Now the additional friction as the weight splooshes up and down in the water will slow its motion down quickly, with only a small overshoot and a couple of boings.
Congratulations; you have just applied damping to the system. Literally.
Carry that one step farther by using a thicker liquid like hydraulic fluid. Now the friction will be so great that the weight won’t overshoot at all; instead, it will slow down as it gets to the resting position and stay there. No boings. (Rearrange things a bit, put the whole thing into a metal cylinder, and you’ll have invented the automotive shock absorber. Just thought I’d mention that.)
Moving-coil dynamics—the punchline
Okay, enough with the analogies. In a moving-coil microphone, the diaphragm and the coil attached to it, with the diaphragm supported at its edge by a springy surround, act like our spring-and-weight model in the previous section. Without some kind of damping, the whole thing will boing back and forth—it will ring—at a particular frequency whenever you hit it with a short-duration sound (a transient). (This is, as usual, a simplification. Real moving-coil microphones actually contain a jumble of components, interacting with each other; the manufacturer has juggled their various resonances to produce an overall response that will—he hopes—please the customer.)
To keep the system from ringing, various forms of damping are applied to the moving parts of the microphone. Most of this damping is mechanical: it’s designed into the moving parts by the manufacturer. But it’s also possible to apply damping electrically.
Here’s where you’ll have to take something on faith: the lower the load impedance into which a moving-coil microphone works, the more its internal motion is damped.
This means that a lower load impedance acts in two ways. First, it changes the mic’s frequency response, lowering the peak in the high frequencies. Second, it damps down the ringing of the mechanical system, which can make for a kinder, gentler sound.
That’s the main reason for the recent popularity of adjustable input impedances on microphone preamps: they can control high-frequency ringing by increasing the mechanical damping on the moving system.
Does this only affect moving-coil microphones? No. Read on.
The transformerless condenser mic is a fairly recent development; classic condensers of the past included head amplifiers (made with vacuum tubes or, later, field-effect transistors or FETs) which were coupled to the outside world by transformers. The mechanical systems of these microphones are isolated by the head amplifiers; any ringing in the moving parts must be controlled by the manufacturer’s mechanical and electrical design.
The transformers, however, are another story. Remember how I said that electrical and mechanical systems can be analyzed using the same mathematical tools? Well, a transformer can be analyzed as a lowpass filter system (it also includes a highpass filter, an item we’ll save for the dance mix); the concept of damping applies as much to a microphone’s output transformer as it does to a mechanical system.
Can you see where this is heading? Transformers, like moving-coil microphones, care about what load they face, and loading a transformer with a too-high impedance can make it ring as thoroughly as an insufficiently-damped moving-coil mic. Figure 1 applies too; insufficient damping causes the mic to exhibit a peak in the high frequencies. Peaks and ringing go together—they are, in fact, all aspects of one big picture (another topic for the dance mix).
Well and good; lower load impedances damp down ringing in the transformers of the condenser mics that use them. There is, however, a tradeoff involved. The electronic circuit of the mic’s head amp sees the load impedance (reflected through the transformer), and just as in the transformerless condenser mic, Ohm’s Law rules. A lower load impedance will draw more current, and the head amplifier will produce more distortion.
With transformer-coupled condenser mics, there’s usually a happy medium: a load impedance which is low enough to damp transformer ringing properly, but high enough to avoid generating distortion in the head amp. Some microphone manufacturers will tell you what the ideal load is (or at least give a recommended minimum). For the rest, it’s probably worth experimenting on your own, if you have a preamp or console with adjustable input impedance (or are willing to make some Gizmos—see “The Taming of the Shure”). Many condenser mics seem to have a loading “sweet spot” around 1500 ohms, which is a common input impedance on modern preamps and consoles, particularly transformer-coupled designs.
Ribbon microphones have gotten popular again, after decades of eclipse; a lot of recordists have recognized that the slight brittleness of many digital recording systems often meshes better with the smoothness of ribbons than the zingy quality of condensers. Ribbons fall into several categories: modern designs like the Beyer M160 and M260; classics like the RCA 44 and 77 series (or their clones from AEA), and mics like the Royer and newer AEA products, which are hybrids of new and old ideas. (Most of the recent spate of Chinese ribbon microphones seem to be copied from AEA designs.)
I’ve done little work with loading on modern ribbons like the Beyers, but my experiment with an M260 and a mandolin (described in my review of the Presonus ADL 600 preamp in this issue) suggests that they might profit from lower impedance loads in the same way as moving-coil dynamics. (Beyer recommends an input impedance of 1000 ohms or more for their ribbon mics, but there’s no law requiring you to follow their suggestion.)
Classic ribbons and their modern offspring are a different story. In the old days, when ribbon mics ruled American recording and broadcast studios, they were typically used into a preamp input transformer with the secondary feeding directly into the grid of a vacuum tube. There was no resistor hanging on the secondary, as is standard practice with today’s transformer-input devices.
Since a tube’s grid constitutes a very high impedance (at least at audio frequencies), this meant the microphone on the input side of the transformer was also seeing a high impedance, and in fact classic ribbons were designed to work best into these higher impedance loads. Their mechanical systems included enough damping that additional electrical damping was not necessary. In fact, it would be detrimental, as the excessive damping would create the kind of droopy response represented by the dashed line in Figure 1.
The folks at Royer, recognizing that many microphone preamps don’t have a high enough input impedance for ribbons to be happy, have introduced microphones with their own built-in head amps, providing internally the higher impedance load the ribbon mechanism wants to see. These active ribbon mics are meant to have the tonality of a ribbon but be as load-independent as a transformerless condenser.
Ribbons can pack some other surprises. In a 1955 JAES paper (cited by Wes Dooley on his AEA website) Richard Werner described his experiment pairing an RCA 77D microphone with a stock RCA preamp. At mid-frequencies the mic’s impedance was 250 ohms and the preamp’s was 1500 ohms. At the microphone’s 50 Hz bass resonance, though, the mic’s impedance rose to 1300 ohms, not uncommon in classic ribbons. Unfortunately, the preamp transformer’s impedance dipped to 600 ohms at 50 Hz; the net result was that the interaction took a bite of nearly 9 dB out of the microphone’s response at that frequency!
Mic preamp choice is clearly critical for ribbons; along with the low noise and high gain required by the mics’ low output, it’s important that the preamp’s input impedance not be too low. Royer suggests 1500 ohms, and I’d consider that a bare minimum; AEA’s mic preamp, designed expressly for ribbons by Fred Forssell, has an input impedance of 18000 ohms.
Maps and journeys
This article gives a quick overview of what happens when a microphone meets an input. I’ve left out a lot of things—dynamic mics, for example, often include transformers of their own, and microphones also function as highpass filters whose bass response can be affected by loading (see Richard Werner’s ribbon for an example). I hope, though, that I’ve included enough information to allow readers to experiment with input impedance controls in an informed way, or to make good use of Gizmos. Remember that theory provides the road map, but you take the trip. Enjoy!