Microphone Engineering
A microphone has one job: convert acoustic pressure waves into an electrical voltage that mirrors them. Three transducer mechanisms have survived a century of competitive engineering, and they have survived because each one is genuinely better than the others at something. A Shure SM58 will beat a Neumann U87 on a screaming vocalist in a loud room every time, not because the SM58 is better but because its moving-coil transducer cannot be overloaded by SPL that would send the U87’s FET preamp into clipping. A Royer R-121 ribbon will make a Marshall stack sound like a record while a small-diaphragm condenser on the same cabinet sounds like an ice pick. Choosing a mic is not really about price, brand, or fidelity in the abstract; it is about matching the physics of the transducer to the physics of the source and room. This post walks the three transducer families, the polar patterns that decide what each mic actually hears, the trade-offs that make a hundred-dollar workhorse outperform a thousand-dollar boutique on certain sources, and a practical framework for choosing.
The fundamental problem
Sound in air is a longitudinal pressure wave between roughly 20 Hz and 20 kHz, with peak pressure variations that are tiny in absolute terms. A conversation at one meter produces around 0.02 Pa peak; a loud snare at six inches can hit 10 Pa. The diaphragm has to be light enough to track sub-millisecond transients but stiff enough to survive years of abuse, sensitive enough that quiet sources rise above the noise floor but not so sensitive that loud sources distort. The transducer has to deliver a voltage at an impedance downstream preamps can use. Every design choice trades one of these against another.
The three mechanisms that won the century are moving-coil dynamics, capacitor (condenser) capsules, and ribbon transducers. They make those trade-offs differently enough that nobody has been able to retire any of them.
Dynamic microphones: moving coil in a magnetic field
A dynamic mic is a loudspeaker run in reverse. A thin Mylar diaphragm, around 25 microns thick, has a coil of fine copper wire glued to its underside. The coil sits in the gap of a permanent magnet. When sound moves the diaphragm, the coil moves through the field, and Faraday’s law gives you an EMF proportional to the coil’s velocity. That voltage, on the order of a millivolt at speech levels, is the signal.
The numbers tell you what dynamics are for. A Shure SM58 has a sensitivity of -56 dBV/Pa (about 1.6 mV at 94 dB SPL), rated impedance 150 ohms (actual 300), and frequency response 50 Hz to 15 kHz with the famous mid-presence peak around 4-6 kHz that makes vocals cut. There is no maximum SPL spec because in practical terms there is no maximum; the assembly is robust enough to live inside a kick drum or six inches from a guitar cab at 130 dB SPL without complaint.
The drawback is moving mass. A diaphragm plus glued-on coil has perhaps 50-100 times the mass of a condenser diaphragm. That inertia smears fast transients; the SM58 does not hear the leading edge of a hi-hat the way a small-diaphragm condenser does. It also does not extend cleanly above 15 kHz, and the low output forces the preamp to deliver 50-60 dB of clean gain.
Dynamics earn their keep on loud sources where overload immunity matters more than transient detail. The Shure SM57 (same cartridge as the SM58, different grille) is the default snare and guitar-cab mic in every studio. The Shure Beta 52A is purpose-built for kick drums with a response that scoops boxy 400 Hz and lifts the 4 kHz click. The Sennheiser MD421 owns toms and shows up on brass and guitar cabs; its five-position bass roll-off compensates for proximity effect at different distances. The sE Electronics V7 competes with the SM58 on live vocals with a slightly more open top.
Condenser microphones: variable capacitance
A condenser capsule is a charged capacitor whose capacitance changes with sound pressure. The diaphragm is a very thin gold-sputtered Mylar membrane, typically 3-6 microns thick, stretched across a metallic backplate with a 25-micron air gap. The two surfaces form a capacitor of 30-100 picofarads. A DC polarizing voltage charges it; when the diaphragm moves, the capacitance changes, and because the charge is held constant, the voltage across the capacitor varies in proportion to displacement.
That voltage is tiny and the source impedance is enormous (gigaohms, because the capacitor is small), so a condenser cannot drive a normal mic cable directly. Every condenser includes an impedance-converting amplifier at the capsule, usually a JFET or in classic designs a vacuum tube. This is why condensers need power: 48 V phantom delivered through pins 2 and 3 of the XLR, an internal battery, or in tube mics a dedicated PSU. The phantom requirement and the active electronics are inseparable from the condenser principle.
The payoff is responsiveness. A 4-micron diaphragm with no coil attached has so little mass that it tracks transients faithfully past 20 kHz. A Neumann TLM 103 has a sensitivity of 23 mV/Pa (about -33 dBV/Pa), self-noise of 7 dB-A, and handles 138 dB SPL at 0.5% THD. That is roughly fifteen times the output of an SM58 with detail extending to 20 kHz the SM58 simply does not capture. The flagship U87 Ai offers three switchable polar patterns and a sound that has defined “professional studio vocal” for two generations.
Condensers subdivide along several axes. Large-diaphragm capsules (over about 25 mm) like the U87, TLM 103, AKG C414, and AT4040 produce a fuller, warmer sound with more proximity effect; they are the default for studio vocals. Small-diaphragm capsules (12-15 mm) like the Neumann KM 184 and Schoeps CMC are more transient-accurate and preferred for acoustic guitar, drum overheads, and orchestral work. Tube circuits add soft harmonic content at the cost of more noise and an external PSU; solid-state designs are cleaner and lower maintenance. Multi-pattern designs like the U87 and C414 mount two cardioid capsules back to back and select the pattern by varying the bias on the rear diaphragm. The AT2020 ($100) and AT4040 ($300) cover budget and mid-range; the C414 XLII ($1,200) is the swiss-army-knife; the TLM 103 ($1,400) and U87 Ai (~$3,600) cover serious vocal work. The gap from $100 to $400 buys much more useful improvement than the gap from $1,400 to $3,600.
Ribbon microphones: foil in a field
A ribbon mic is conceptually a dynamic with no coil. A single corrugated aluminium foil, typically 1.8-4 microns thick and 2-6 mm wide, is suspended in a strong magnetic field. Sound moves the ribbon directly, and the ribbon itself acts as a one-turn conductor through the field. Raw output is even lower than a dynamic (often -50 to -60 dBV/Pa), and a step-up transformer immediately after the ribbon multiplies it up.
Because the ribbon is open to air on both sides, the transducer responds to the pressure gradient across it rather than absolute pressure on one side. That means a bare ribbon is inherently figure-of-eight: equally sensitive front and back, with deep nulls at 90 and 270 degrees. The natural figure-of-eight pattern is one of the defining acoustic features of ribbons and the basis for the Blumlein stereo technique.
The Royer R-121, introduced in 1998, is the modern industry standard: -50 dBV/Pa sensitivity, frequency response 30 Hz to 15 kHz with a gentle high-frequency roll-off, 135 dB max SPL at 1 kHz. The active R-122 MkII adds a phantom-powered head amp that lifts output to about -36 dBV/Pa and eliminates the need for an exotic preamp. The Coles 4038 is a 1953 BBC design still made by hand in England, with a famously smooth midrange that has stayed in the drum-overhead and brass orbit for seventy years. The AEA R84 sits between Royer and Coles in price and character.
Ribbons exist because the smooth, slightly dark, naturally figure-of-eight character flatters certain sources in ways condensers cannot. Brass through a ribbon loses the bright bite that grates on the ear at high SPL; an overdriven guitar cab through an R-121 sounds musical where a small-diaphragm condenser on the same cab sounds spiky. The cost is fragility: a passive ribbon needs 60-70 dB of clean preamp gain, and a stretched ribbon means a trip to the re-ribboning service. Vintage RCA ribbons can be destroyed by phantom power; modern ribbons (R-121, Coles 4038) are phantom-safe with correct wiring but vulnerable to hot-patching. Active ribbons solve both gain and phantom-safety at the cost of the simpler all-passive path.
Comparing the three families
| Property | Dynamic (SM58) | Condenser (TLM 103) | Ribbon (R-121) |
|---|---|---|---|
| Transducer | Coil moving in magnet | Variable capacitor with FET | Foil ribbon in magnet, transformer-coupled |
| Sensitivity | -54 to -58 dBV/Pa | -30 to -40 dBV/Pa | -50 dBV/Pa (passive); -36 dBV/Pa (active) |
| Self-noise | Negligible | 5-15 dB-A | Negligible (thermal of transformer) |
| Max SPL | Effectively unlimited | 117-140 dB | 130-138 dB |
| HF extension | Rolls off above ~15 kHz | Flat to 20 kHz+ | Gentle roll-off above ~12 kHz |
| Transient response | Slow (heavy diaphragm) | Fast | Fast |
| Power required | None | 48 V phantom or tube PSU | None (passive) or phantom (active) |
| Native pattern | Cardioid | Anything (dual capsule) | Figure-of-eight |
| Fragility | Tank-like | Moderate | Very fragile |
| Typical cost | $100-$400 | $100-$10,000+ | $700-$5,000 |
Polar patterns and what they hear
The polar pattern describes how sensitivity changes with the angle of arriving sound. It is determined by the acoustic design of the capsule (how the back of the diaphragm is exposed or sealed to rear-arriving waves), not by the transducer mechanism, though some patterns are easier to achieve with some transducers.
Omnidirectional Cardioid Figure-of-Eight
0 0 0
. . .
. . . . . .
. . . . . ***** .
. ********* . . ********* . . ******** .
. ***************** . ****** . . . *************** .
. ***** ***** . ****** . . . ***** *****.
. ***** MIC ***** . ***** MIC . . ***** MIC *****.
. ***** ***** . ****** . . . ***** *****.
. ***************** . ****** . . . *************** .
. ********* . . ********* . . ******** .
. . . . . ***** .
. . . . . .
. . .
180 180 180
0 = on-axis (front) 180 = off-axis (rear) * = sensitivity region
Omnidirectional capsules respond to absolute air pressure on one side of a sealed diaphragm. Sensitivity is roughly equal in all directions, there is no proximity effect, and the response is typically the flattest a given mic can achieve. Omnis are wonderful for the natural ambience of a good room and for lavaliers where the source moves. In a bad or noisy room, they capture everything you wish they would not.
Cardioid (“heart-shaped”) is the most common pattern. Acoustic delay paths from the rear are designed so rear-arriving sound reaches both sides of the diaphragm in phase and cancels. Rear pickup is 20-25 dB down, side pickup about 6 dB down. The SM58, SM57, TLM 103, and U87 in cardioid mode all live here. Cardioids forgive imperfect rooms by rejecting rear reflections, and they exhibit proximity effect: close to the source, the pressure-gradient component grows large relative to the pressure component and bass rises by 6 dB or more in the 100-300 Hz region. This is the “radio voice” sound. Singers exploit it; engineers compensate with high-pass filters when they do not want it.
Supercardioid and hypercardioid patterns tighten the front lobe at the cost of a small rear lobe. Supercardioid (Shure Beta 87A, Sennheiser e945) has its null at 125 degrees off-axis, which puts maximum rejection where stage monitors typically sit. Hypercardioid (Beta 87C, AKG C535) tightens further with the null at 110 degrees.
Figure-of-eight (bidirectional) picks up equally front and back with deep side nulls. Native to ribbons, selectable on multi-pattern condensers. Two figure-of-eights at 90 degrees give a Blumlein stereo image; one paired with a cardioid gives Mid/Side with continuously variable width in post.
| Pattern | Rear pickup | Side pickup | Proximity effect | Typical use |
|---|---|---|---|---|
| Omni | Full | Full | None | Room mics, lavaliers, classical |
| Cardioid | -20 to -25 dB | -6 dB | Strong | Vocals, instruments, general |
| Supercardioid | -10 dB | -8 to -12 dB | Stronger | Live vocals, broadcast |
| Hypercardioid | -6 dB | -12 dB | Strongest | Film boom, broadcast |
| Figure-of-eight | Full | Deep null | Strongest | Ribbons, Blumlein, M/S |
Why a $100 SM58 sometimes beats a $1000 condenser
Newcomers often assume more expensive equals better and that any decent condenser will outperform any decent dynamic. This is not true and the reasons are physical.
The first reason is overload. A condenser with a 138 dB max SPL is fine for a singer at 12 inches. Put it three inches from a snare during a rim shot and you are past 145 dB peak; the internal FET clips and you get distortion. The SM57 in the same position has no electronics to overload and effectively no excursion limit; the worst case is mechanical compression that engineers actually like.
The second reason is bleed rejection. In a tracking situation with a full kit, isolation matters. A condenser at -33 dBV/Pa picks up every nearby drum loudly; a dynamic at -56 dBV/Pa picks them up 23 dB quieter for the same input pressure. Combined with a tight cardioid pattern and a directional null aimed at the next-loudest source, dynamics give you the isolation you need to mix later. Beta 52A inside the kick, SM57 on snare, MD421s on toms is the rock-drum standard not because condensers are bad but because dynamics are right.
The third reason is environment. A condenser’s thin diaphragm hates moisture, dust, plosives, and rough handling. The SM58 has been thrown from stages, dunked in beer, and stepped on by drummers, and still works the next day. Live engineers and field reporters use dynamics for the same reason.
The fourth reason is colouration. The mid-presence peak Shure designed into the SM58 sits exactly where consonants and vocal intelligibility live. It is not flat and it is not trying to be; it is voiced for the source. A flat-response condenser captures the room and the source faithfully; an SM58 captures a vocal that already sounds like it belongs in a mix.
The fifth and most underappreciated reason is the support chain. A $1,000 condenser into a $200 interface preamp with poor noise figure and a stiff input impedance does not deliver anything close to its potential. A $100 SM57 into a great preamp (a Neve 1073, an API 512c, even a clean modern Grace m101) is a very good signal. If you have $1,500 to spend on a vocal chain and put $1,400 of it on a TLM 103, you have made a mistake.
A practical decision framework
When picking a mic, walk through these in order.
- What is the maximum SPL the source will produce at the intended distance? Above 135 dB SPL, a passive dynamic is the safe choice. Above 140 dB peak, it is the only choice unless your condenser has a heavy pad. Kick drums, snares, cranked electric guitar cabinets, brass bells, screaming vocalists are dynamic territory.
- How fast are the transients? Cymbals, hi-hats, finger-picked acoustic, brushed snare, classical strings are condenser territory, probably small-diaphragm. A dynamic on a hi-hat will sound dull.
- How does the source sit in the high mids and top? Bright, edgy, or harsh sources (overdriven amps, hot brass, bright cymbals) often benefit from a ribbon’s natural high-frequency roll-off. A Royer R-121 on a Marshall cabinet has been the default for thirty years for this reason.
- What does the room sound like? A good room rewards an omni or wider cardioid; a bad room punishes both. In a bad room, get closer, use a tighter pattern, and pick a mic with strong off-axis rejection.
- What is the rest of the chain? A great mic into a poor preamp is wasted; a humble mic into a great preamp can sound spectacular. The preamp’s noise figure matters most when the mic’s output is low (passive ribbons, dynamics). Cloud lifters and similar inline boosters add 20-25 dB of clean gain ahead of a noisy preamp specifically to make low-output mics usable.
USB condenser mics like the Audio-Technica AT2020 USB+, Shure MV7, and Blue Yeti deserve an honest note. For podcasting and content creation they are fine. They are weaker for serious music recording because the integrated USB interface forecloses on the most important variable in the chain: the preamp. You get what the manufacturer chose to give you. See our notes on signal chains in live broadcast for the broader system view.
The signal chain and other practical traps
A microphone is the start of a chain whose weakest link sets the ceiling. The preamp matters: a dynamic or passive ribbon needs 55-70 dB of clean gain, more than cheap interface preamps provide without audible hiss. The conversion stage matters: 24-bit at 48 or 96 kHz with a quiet converter is affordable now. Balanced XLR with quality connectors is a baseline. The impedance and noise considerations from how a transistor actually works at the device level reappear at the system level here.
Phantom power deserves a warning. IEC 61938 delivers 48 V DC through pins 2 and 3 of an XLR with a small current limit, and is safe for almost any modern mic. Active ribbons (R-122 MkII, R-84A) need phantom. Modern passive ribbons (R-121, Coles 4038) are safe when the cable is correctly wired but can be damaged by hot-patching or mis-wired cables that present DC unequally. Vintage RCA ribbons should never see phantom. Safe practice: turn phantom off before patching, turn it on once the cable is settled, and do not hot-patch.
Impedance loading is the trap engineers forget. A mic has a source impedance (150-300 ohms for dynamics, 50-200 ohms for active condensers, 200-700 ohms for ribbons after the step-up transformer). The preamp should present 5-10x higher to avoid loading the mic and shifting its frequency response. Some ribbons sound noticeably different into a 1 kohm input versus 10 kohm; Royer recommends at least 1.5 kohm for the R-121, with most enthusiasts running 10 kohm or higher. Preamps with switchable input impedance (Grace, Millennia, AEA RPQ) let you tune this deliberately.
Monitoring matters as much as capture. Closed-back headphones for tracking (to prevent bleed) and open-back for mixing is non-negotiable for serious work; the physics of how those headphones reproduce sound mirrors how microphones receive it, as covered in how noise-cancelling headphones work.
Verdict
There is no best microphone, only a best mic for a source, a room, a chain, and a workflow. Dynamics win on loud, abusive, high-bleed, rough-environment sources where their robust simplicity and natural rejection of distant sound are irreplaceable; the SM57, SM58, Beta 52A, and MD421 have held the rock workhorse positions for fifty years for substantive physical reasons. Condensers win on quiet, transient-rich, detail-dependent sources in good environments; the U87, TLM 103, C414, and AT4040 cover most of what serious studios actually use. Ribbons win on bright, harsh, or aggressive sources where their natural roll-off and figure-of-eight pattern produce a smoother result; the R-121, R-122 MkII, and Coles 4038 are not nostalgia, they are the right physics for those sources.
If you want one mic that will record almost anything acceptably, buy an SM57 and a great preamp. The second mic is a large-diaphragm condenser around $300-500 (AT4040, Rode NT1, AKG C214). The third is a small-diaphragm condenser pair for overheads and acoustic guitar. A ribbon is a fourth purchase, not a first, and it should follow a specific source that fights every other mic you own.
The fastest way to make better recordings is not to buy a more expensive mic. It is to put the mic you already have in a better position, in a better room, into a better preamp, with better gain staging. Microphones are physics; once you know which physics you need, the choice gets simpler.
Comments