I’ve always had an ambivalent relationship with the Shure SM57 microphone. On the one hand, it’s ubiquitous, probably the best-selling microphone in the world (along with its sibling, the SM58). It’s rugged, reliable, and affordable; and it has become a standard in studios around the world, pro or home, with applications ranging from guitar amps to snare drums. It has fueled the performances of bar bands for decades, and for at least the past quarter century, when a President of the United States has told a lie, he’s done it through an SM57.
Call me a contrarian, but this microphone has never worked for me. The problem is the high end: there are a couple of sharp high-frequency peaks in the response curve that combine with what sounds like intermodulation distortion to produce a sound that’s just too bright, harsh and “wiry” for my tastes. Yes, it’s a sound that “cuts through” a dense mix, or a loud crowd, but it’s not natural, it’s not clean, and it’s not what I want.
On the other hand, a lot of people whose ears I respect have praised the SM57 as a good “Swiss Army mic”. One of them is our own Scott Dorsey, who has asserted (both on the rec.audio.pro newsgroup and in this magazine) that the SM57 is a useful mic with fully professional sound—when it is properly loaded.
Carry that load
Loading is a hot topic at the moment, as several preamp manufacturers have introduced models with variable input impedances. How come?
For the last few decades, microphone inputs (on consoles and standalone preamps) have standardized on an impedance of 1500–2500 ohms, which is optimal for most condenser microphones. As it turns out, though, one size does not fit all. Old-fashioned ribbon microphones (RCAs and their progeny) were designed with the assumption of a relatively high loading impedance; the preamps of that era typically connected the secondary of a high-ratio input transformer directly to the grid of the first amplifier tube, with no terminating resistor. This presented an extremely light load to the microphone, which was designed accordingly.
More germane to our story, the required load on many dynamic (moving-coil) microphones turns out to be surprisingly low. The motor of a moving-coil microphone is essentially a loudspeaker in reverse; the diaphragm (suspended by an elastic surround) moves the coil through the field of a magnet, generating an electrical signal which goes out into the world, often through a transformer inside the mic body. Because there is no electrical isolation of the capsule from its load, the characteristics of the generator, in particular its damping, are affected by that load.
Here’s a way to grasp what’s going on. (Warning: I am about to oversimplify. If you’re a mechanical engineer, feel free to scream—to yourself.) Go over to one of your monitor speakers (unpowered), unplug the wires and gently push the woofer cone in and out, with your fingers splayed across it so you’re pushing uniformly. If it’s a vented-box monitor, the woofer will move quite freely; if it’s a sealed-box design, the woofer will be more constrained in its motion, but you will still be able to push it in and out with ease.
Now plug the speaker wires back in and turn on the amplifier. Try pushing the woofer—do you feel how reluctant it is to move? The woofer’s motion is being damped by the low impedance of your power amp, typically a fraction of an ohm. In effect, the woofer is no longer free to move around, but will only move when and where the amplifier tells it to.
Something similar happens when a microphone is operating. With a light (higher-impedance) load, the microphone’s motor is operating with little constraint. It follows the changes in air pressure that constitute sound waves, yes, but it’s also free to rattle around some, and it can slightly overshoot the mark, particularly at high frequencies, adding ringing, intermodulation, or general “hashiness” to the sound. If there’s a transformer in the mic, it can also exhibit high-frequency ringing when lightly loaded.
When the load impedance is lower, howeve r, it acts as an electromechanical damper on the microphone’s motor. Like the plugged-in woofer, the diaphragm is operating under greater constraint, and is less capable of overshooting, ringing and other varieties of nasty experience. Likewise, the transformer (if present) has any high-frequency electrical ringing damped down by the heavier load, producing an output that is (one hopes) truer to the sound striking the diaphragm. It is worthy of note that, according to Scott, most of Shure’s mic preamp and mixer designs at the time the SM57 and its siblings (including the ancestral Unidynes) were introduced had input impedances close to 600 ohms, so it makes sense to assume the microphone was designed with lower impedances in mind.
Such is the theory. What happens when it’s tested in the crucible of life itself?
Bang The Drum Slowly
I realized last fall, when I began teaching a course in the recording program at Webster University, that I had all the ingredients at hand to test the assertion that a lower load would do good things for an SM57. They included:
• An SM57.
• A preamp with two channels having adjustable input impedance (the Universal Audio 2-610).
• A studio with two good recording systems (a Studer A-807 analog recorder and an iZ Technology RADAR 24 digital recorder).
• A class of recording students with good ears and promising abilities to talk about what they heard.
• A student who owned a set of drums and knew how to play them.
For our experiment, during the Fall 2005 semester, we set up a crash cymbal and an SM57 about 18" above it. I realize that it’s unlikely anyone will use an SM57 on cymbals—it just isn’t done—but I wanted something that would stress the high frequencies as severely as possible. The microphone was connected alternately to the two channels of the preamp; one was set for an input impedance of 2000 ohms, the other for 500 ohms, using the switches on the front panel.
To do any useful comparisons, though, levels need to be matched very closely. The human ear is a funny thing; a volume difference of a few tenths of a dB is audible to most listeners. Often the difference is not perceived as an overt level difference when it’s small, but as changes in timbre, dynamics, “depth”, and other qualities.
So we needed to match levels. After we got a basic level on the cymbal, using the 2k load, we moved the mic in front of a handy guitar amp, fed it with a 700 Hz tone generator, and left the room. (Why did we leave the room? Aside from the obnoxiousness of the tone, the level changed noticeably when a person walked around the studio, even though the mic was only a few inches from the loudspeaker—such is the power of the acoustic environment.)
We used the Studer recorder as a glorified voltmeter, adjusting its input control until the meter read exactly 0 VU. We then switched the microphone cable and the cable running to the Studer over to the other channel of the 2-610, which was set to 500 ohms input impedance. Without touching the controls on the recorder, the students “talked me in” as I adjusted the gain control on this second channel until the meter was again at 0 VU. The two channels, with their differing loads, were now matched in gain within a pointer-width, about 0.1 dB. (Let’s hear it for real analog meters; on a job like this, an LED just can’t touch ‘em.)
The levels having been matched, we replaced the mic on the cymbals and gestured to our uki, Matt Sims, to start playing. He alternated between light rhythmic taps and smashing crashes à la Lars Ulrich, while we recorded on a single track of the Studer (to preserve repeatability), alternating between the two channels of the preamp. We repeated the tests on a single channel of RADAR, then settled in for listening; I record here a combination of my own impressions and the students’ comments.
• The differences were clearly audible on both systems, but somewhat more so on RADAR. At 2k the sound was bright and ill-differentiated, almost like bursts of pure white noise, and quite unpleasant. At 500 ohms, however, it was like we’d plugged in a different microphone. While the sound was still brightish, it had much more detail, and the wash of hiss was replaced by the sheen characteristic of real cymbals.
• Along with the sheen, there was a much clearer sense of someone, well, tapping a real piece of metal, with a clearly defined “clang” along with the cymbal tizz. There was space around the sound, too, whereas the sound at 2k bloated into the space and crowded it out.
In short, the recordings at 500 ohms confounded everything I had ever experienced with SM57s.
Just for grins, we repeated the experiment with a singer. The uki in this case was Tom McArthur, and while we appreciated his efforts, the results were more equivocal. In retrospect, it was probably a mistake to use “A Little Help from My Friends” as the test song, as it contains few sibilants; if we tried it again, I would sing something like “Swim, Sam, Swim” instead.
For further grins, we repeated the cymbal experiment with a Shure SM81 condenser mic. In this case, everyone preferred the results with a 2k load; that channel sounded cleaner and clearer in every way. Shure recommends a load of 1500 ohms or higher with this microphone, and it looks like they’re right.
Mulling it over
So what had we learned? It looked like Scott’s assertion was confirmed: the SM57 in fact performed far better with a lower-impedance load. But as I pondered the results over winter break, I realized there was a fly in the ointment.
On the 2-610 preamp we used, the input impedance is changed by switching taps on the input transformer. That means the load is always a complex one, incorporating the various inductances and capacitances that come with a transformer input. The Groove Tubes VIPRE preamp does the same thing.
A lot of recently-introduced preamps, however, do the job differently. Instead of using multi-tapped transformers, they change input impedances by switching additional resistors in parallel with the input, loading the microphone with an essentially resistive load. (The new PreSonus ADL600 works that way—watch for a review soon.) Would such a load be equally effective in taming the SM57’s response? If so, that would open up some interesting possibilities.
It was time for a second round of tests.
As the new semester was beginning, I built a Gizmo. It was made from two XLR connectors (male and female), a foot of cable, and a 698 ohm resistor; the cable connected the male to the female plug in the usual fashion, while the resistor was soldered between pins 2 and 3 of the male.
With the fresh ears of the spring students, we repeated the experimental setup using the same procedures, with one difference. Instead of switching one channel of the preamp to 500 ohms, I left both switches in the 2k position, but connected the Gizmo in line with the mic cable on channel 2. This placed the resistor in parallel with the preamp’s internal impedance, for a net load of 517 ohms, close enough for folk music.
When we repeated the experiment using the Gizmo—thanks to our uki, Caey Oliver—the results were the same. All of the students thought the lower-impedance input sounded more natural, less hashy, and more like a real cymbal. Just to have a fair comparison using the new set of ears, we re-ran the previous semester’s experiment using the internal impedance setting (changing transformer taps), and again the results were the same. Some students thought the differences were more subtle with the Gizmo, others with the internal load switch.
As it happened, I had my Martin guitar along, the veteran of many a mic test, so we decided to try that as a signal source as an alternative to the cymbals. For the first round, one of the students, Jonathan Fournier volunteered as uki. The results were equivocal; the differences between loads weren’t particularly audible.
How come? Jonathan’s a very clean player with a smooth style, who fingerpicks with his fingertips rather than nails or picks—not a lot of fast transients to make an SM57 misbehave. I decided that clean and smooth were not what was called for, so I picked up the M artin and cranked out a couple of raunchy blues choruses, complete with plenty of pick noise, rattles and thumb thumps, while the students switched from the plain channel to the Gizmoed one. That should give the SM57 something to contend with.
It did. One student commented that the difference between the two loads was like night and day, as though—and he recognized the cliché—a veil had been taken off the sound. Another said the side with the Gizmo sounded like we’d substituted a different and better microphone. I’d have to agree; everything was better defined, the strings were less rattly, and the frets buzzed less. Even the thumb thumped more clearly, with tighter bass and a better sense of pitch.
Most important, all of us heard the differences using both methods of changing impedance. Some thought the Gizmo was actually more effective than the internal switch, others found them equal. None thought it was less effective.
Go thou and do likewise
The fact that we were able to effect significant improvements in sound using the Gizmo meant several things. First, of course, it validates the idea of internal resistive loading as a way for a commercial preamp to alter its input impedance.
It also, however, opens the possibility that you, the reader, can do the same thing we did, and for an astonishingly small outlay of cash.
Our Gizmo used three parts, bought from a local surplus store and costing $8.47 including sales tax, plus a length of 3-conductor shielded cable from my basement stash. You can do the same, or you could get equivalent results by soldering the resistor across pins 2 and 3 of the male XLR connector on a pre-existing mic cable (be sure you label it!) or a short one purchased especially for this use; Markertek (www.markertek.com) sells a nice 18-incher for $11.95. Do the wiring as shown in Figure 1 (note that the XLR inserts are shown from the back side, the side where you solder).
The resistors can be found at Mouser Electronics (www.mouser.com) for $1.10 per package of 10; order stock # 270-698. (If you use a different resistor from 698 ohms—see the sidebar—substitute that value for the last three digits in the stock number.) They don’t have a minimum order, but you’ll probably pay about the same as the resistors’ cost for shipping.
Perhaps the simplest and least expensive route is to wire up a “barrel” connector, again using the hookup shown in Figure 1. Barrel connectors are simply short metal cylinders with a female XLR connector at one hand and a male at the other; they’re often used to house balanced mic pads and other passive devices intended for inline use. The Neutrik NA3FM barrel connector is available from Markertek and other outlets; expect to pay about $7.50 apiece. Whether you use a barrel, make a short cable or use an existing cable, I suggest you solder the resistor across the male insert rather than the female; some brands of female XLR connectors (including Switchcraft) use small semicircles of metal rather than cups as the solder pins, and it’s tough to connect more than a single wire to them.
When you’re done, triple-check to make sure that each pin is connected to the same numbered pin on the other connector (remember that the inserts are mirror-images of one another), and that none of them are shorted to one another or to the case. If you have a volt-ohmmeter or digital multimeter, you should measure about 698 ohms between pins 2 & 3, and infinite impedance between either of those and pin 1.
(By the way, I chose 698 ohms as a good compromise resistor value that gives a total load of about 500 ohms (±10%) with common available preamp impedances from 1500–2400 ohms. If you want to tailor the resistor to your preamp or board’s actual value, see the sidebar.)
Applications and conclusions
Will the Gizmo tame other microphones besides the SM57? It’ll certainly work on mics that share the SM57’s basic motor, such as the SM58, although it won’t help the latter’s drastic rolloff in the top octave (that’s a function of the head design). It’s probably worth trying on other microphones with an unbridled high register; I wonder, for example, whether it might gentle down the Sennheiser MD421 Mark II, a real screamer in comparison with the Mark I.
Is there any down side to using the Gizmo on a dynamic mic? Only a very small one: the lower impedance load will drop the microphone’s level between 1 and 2 dB compared with the stock load, which you’ll need to make up when setting the preamp’s gain. That’s usually not a big deal, unless you’re recording something like a quiet mountain dulcimer—and if you’re doing that with an SM57, you get no sympathy from me.
The Gizmo is almost certainly not going to improve the performance of transformerless condenser mics. If they’re bright, the brightness comes from the capsule design, which is isolated from the output by the microphone’s head amplifier, so the only result of lower-impedance loading is likely to be increased amplifier distortion, hardly desirable.
With transformer-coupled condenser mics there might be some improvement in damping down ringing in a poor-quality output transformer (endemic to some of the cheaper Chinese specimens), but that would have to be balanced against the likely increase in head-amplifier distortion as the circuit works harder—witness our poor results when the class tried the lower impedance on an SM81.
My feeling is that lower-impedance loading will prove to be a useful tool for getting better sound out of dynamic mics, particularly the ubiquitous SM57 family; by showing that the improvement is attainable through purely resistive loading, we’ve opened up the possibility of big improvements for very small expenditures, always a pleasant prospect. Try a Gizmo on your SM57; I think you’ll be pleased—and if not, you’re only out a very small sum of money.
Paul J. Stamler is a recording engineer, audio enthusiast, and instructor in recording technology, in St. Louis. He may be reached via firstname.lastname@example.org
Paul offers deep thanks to the members of his Analog Recording Technology classes at Webster University for lending their ears and instruments to these experiments. Fall semester 2005: Sean Beach, Thomas Carpenter, Regina Goh, Tom McArthur, Larry Morris, Thad Salender, Matthew Sims, Sahila Topon, and Shawn Yates. Spring semester 2006: John Bolduan, Chris Collum, Marc DuPain, Jonathan Fournier, Ben Majchrzak, Matt Nichols, Casey Oliver, Daniel Ruder, Daniel Williams. Thanks also to Scott Dorsey for starting the discussion in the first place. You folks make this fun.
SIDEBAR: Custom-Sizing Resistors
If you want to calculate a load resistor that’s tailored to your preamp/board’s actual resistance, it’s easy to do. The formula is:
1/Rg = 1/Zd – 1/Za
where Rg is the resistor to be used in the Gizmo, Zd is the desired total load impedance, and Za is the actual load impedance of the input.
So, for example, if you want to create a load impedance of 500 ohms, as we did in the experiments, you would first look up the specified input impedance of your board. A Mackie board with XDR Pro preamps has an input impedance of 1300 ohms; plugging the numbers into the equation, we get:
1/Rg = 1/500 – 1/1300 = 0.002 – 0.000769 = 0.00123
Pushing the “1/x” button on the calculator, the answer is:
Rg = 812.5 ohms
The nearest value in the 1% tolerance series of resistors is 806 ohms, so that would be the one to use.—PJS