What Phantom Power Is For, and Why Your Microphone May Not Need It

This page contains affiliate links. If you buy through one we may earn a commission, at no cost to you.

It is not as high as 48 volts by the time it arrives, and that is the first thing worth fixing. The supply is fed through two resistors of 6.8 kilohms each, which is 3.4 kilohms in parallel, and a microphone drawing the 10 milliamps the standard allows drops 34 volts across them and runs on 14. The most power that can ever cross that network is 169 milliwatts. So the headline voltage is not a supply rail, it is the open circuit end of a deliberately current limited feed, and the reason it is 48 rather than 12 has nothing to do with needing more power. A 24 volt version of the same standard delivers more.

And the answer to the second question is no. A dynamic microphone has no electronics in it at all, so there is nothing for the power to power, and the common experience of plugging a professional microphone into a phone and hearing nothing is almost never a quality problem. It is a scheme mismatch, and there are at least four different schemes in circulation.

Why a condenser needs power at all, in one calculation

A condenser capsule is a capacitor. One plate is a thin stretched diaphragm and the other is a fixed backplate a few tens of microns behind it, with a charge held across the gap. Sound moves the diaphragm, which changes the spacing, which changes the capacitance, and if the charge is held constant the voltage across the capacitor changes in proportion. That is the whole transduction and it is beautifully direct.

The trouble is the impedance. A capsule of that size is perhaps 20 to 50 picofarads. Take 30, and the source impedance of a capacitor is one over two pi times the frequency times the capacitance: at 100 hertz that is one over 2 pi times 100 times 30 times ten to the minus twelve, which is 53 megohms. At 20 hertz it is 265 megohms.

Now try to send that down a cable. Five metres of microphone cable has perhaps 500 picofarads of capacitance between its conductors. A 30 picofarad source feeding a 500 picofarad load is a capacitive divider, and the fraction that survives is 30 divided by 530, which is 5.7 percent. That is a loss of 24.9 decibels, before the cable has touched anything else, and any stray electrical field near a 50 megohm node will arrive as hum at full strength.

So a condenser capsule cannot drive a wire. There has to be an amplifier inside the microphone body, within millimetres of the capsule, to convert that enormous impedance down to something a cable can carry. The amplifier needs power, the capsule needs its charge, and the power has to get there somehow. That is the entire problem phantom power solves.

One important distinction that decides almost everything downstream. A true condenser capsule needs an external polarising voltage to put the charge on the backplate, typically somewhere between 48 and 60 volts, and that is why 48 is the number: it is set by the capsule, not by the amplifier. An electret capsule has a permanent charge trapped in a dielectric layer during manufacture, so it needs no polarising voltage at all and only needs power for its internal impedance converter. That is why an electret runs happily on a volt and a half from a single cell and a true condenser does not, and it is the whole reason the low voltage schemes exist.

The trick itself, and why it is called phantom

A balanced microphone connection has three conductors. Pin 1 is the shield, which goes to ground. Pins 2 and 3 carry the audio as a difference between them: the microphone drives them in opposite directions and the input at the far end responds only to the difference. The point of that arrangement is interference rejection, because any electrical field the cable passes through induces the same voltage on both conductors, and the same voltage on both means no difference, so it vanishes.

Phantom power exploits exactly the same property in reverse. Feed the direct voltage onto pin 2 and pin 3 through two equal resistors, and the DC appears identically on both conductors. The input responds to the difference, the difference is zero, and the audio rides on top of it undisturbed. The microphone, which sits between the two conductors, sees the voltage and takes its power from it. Hence the name: the supply is invisible to the thing it is sharing a path with.

Two consequences worth knowing. The resistors have to be matched, and the standard specifies 0.4 percent, because a mismatch is precisely a failure of the equal voltage condition: whatever ripple sits on the supply rail arrives at the input attenuated only by the mismatch fraction. At 0.4 percent, one millivolt of supply ripple becomes four microvolts of differential signal, which against a typical microphone output around ten millivolts is 68 decibels down, and that is the design target. Poor matching is audible hum, and it is audible as a property of the input rather than the microphone.

And the scheme requires a balanced connection to exist at all. Two conductors carrying a difference, plus a separate ground, is the minimum. A two conductor unbalanced connection has no second signal pin to put the other resistor on, so phantom power cannot be done over it at any voltage.

Why not a higher voltage, and why 24 volts wins

The standard defines three phantom levels and the numbers do not run the way you would guess. P48 is 48 volts through two 6.8 kilohm resistors, P24 is 24 volts through two 1.2 kilohm, and P12 is 12 volts through two 680 ohm.

Work out what each can actually deliver. Maximum power transfer across a source resistance is the supply voltage squared divided by four times that resistance. For P48 the parallel pair is 3.4 kilohms, so 48 squared over 13,600, which is 169 milliwatts. For P24 the pair is 600 ohms, so 576 over 2,400, which is 240 milliwatts. For P12 the pair is 340 ohms, so 144 over 1,360, which is 106.

So P24 delivers 42 percent more power than P48 does, from half the voltage, because its feed resistance is nearly six times lower. Which settles the question in the original title: 48 volts is not high in order to deliver power. It is high because a true condenser capsule needs a polarising voltage in that region, and once you have committed to supplying it through a resistor network the current ceiling falls out of the resistor values rather than the other way round. If power were the goal the standard would look completely different, and in fact the connector that does have power as a goal looks completely different.

Four schemes in our own catalogue, and what each listing says

Start with the one that publishes its figure, because it is the only one that does. A Rode VideoMicro compact on camera microphone states in its own feature list that no battery is required, being powered by camera plug in power at a minimum of 3 volts. That is a published electrical specification and it is the only one in nine microphone products.

Plug in power is the fourth scheme and the smallest. It is nominally 5 volts fed through 2.2 kilohms on the tip of a 3.5 millimetre jack, which by the same arithmetic delivers at most 2.8 milliwatts. That is 60 times less than P48 and it exists for one purpose, which is to run the field effect transistor impedance converter inside an electret capsule. It cannot polarise a true condenser and it is not trying to.

Read the published minimum of 3 volts against that network and you can see how tight the margin is. If the microphone needs 3 volts at its terminals and the source is 5 volts through 2.2 kilohms, the available current is 2 volts over 2.2 kilohms, which is 0.91 milliamps, or 2.7 milliwatts. There is almost no slack, which is exactly why the record lists as its main weakness that it requires camera plug in power and will not work with devices lacking that feature, and as a drawback that it requires plug in power from the camera. Both are correct and both are about a 2.8 milliwatt budget. Full record in our Rode VideoMicro review.

Then the one with the best powered connector in the catalogue, which nobody frames that way. A Rode VideoMic GO II-H shotgun microphone bundle publishes dual outputs, a 3.5 millimetre jack and a USB-C port appearing as a class compliant USB audio device, with no batteries and bus power.

USB bus power is 5 volts at up to 500 milliamps on a basic port, which is 2,500 milliwatts. That is 15 times the best phantom figure and 880 times plug in power, and it is why a microphone with a USB-C output can carry a converter, a preamplifier and a digital signal processor and still be the size of a cigar. The same microphone through its 3.5 millimetre output has 2.8 milliwatts to play with. One microphone, two sockets, a thousandfold difference in the power budget, and the record describes it as dual outputs for camera and desktop use, which is true and is about convenience rather than about the real asymmetry. Details in our Rode VideoMic GO II-H review.

Then the silence, and it is in the product that most needs to break it. A Zoom H1essential handy recorder bundle ships, among its stated contents, a lavalier condenser microphone, along with a 32 gigabyte card, a windscreen and batteries. A condenser needs power. The record does not say where that power comes from, does not name the recorder's input connector, and does not say whether the recorder supplies anything at all.

The inference is available and should be read as an inference. A lavalier packaged for a pocket recorder will be an electret on a 3.5 millimetre plug, which means the recorder has a 3.5 millimetre input supplying plug in power, which means it is not supplying phantom power and cannot accept a microphone that needs it. If that is right, then the one thing a buyer with a studio condenser needs to know about this recorder is absent from its listing, and the one thing a buyer of the bundle needs to know about the included microphone is absent too. Record in our Zoom H1essential review, and the companion argument about what its 32 bit float claim is really doing is in why 24 bit matters and 96 kHz does not.

And the fifth scheme, which is where this category actually went. A tongyu J3 wireless lavalier microphone publishes a 2.4 gigahertz plug and play link, a noise reduction chip, real time ear monitoring and up to 8 hours of battery. There is no powering question here at all, because the capsule, its converter and its radio all sit on one side of a battery and nothing electrical crosses to the receiver. The listing publishes a battery life and not a voltage, which is the correct thing to publish for this architecture, and it is why the whole phantom power problem has quietly disappeared from the consumer market. Record in our tongyu J3 review.

A dynamic microphone, and the one case where the voltage does harm

A dynamic microphone is a generator. A coil of wire is attached to the diaphragm and sits in a magnet's field, and moving a conductor through a magnetic field induces a voltage in it. There is no charge to maintain, no impedance to convert and no amplifier, because the coil's own impedance is a few hundred ohms and it can drive a cable directly. Nothing in it can use power and nothing in it needs any.

On a correctly wired balanced connection, leaving phantom power switched on does no harm to it, and the reason is worth stating because it is the same reason the trick works at all. Both ends of the coil are connected to pins 2 and 3, both of which sit at the same DC potential, so there is no voltage across the coil and no current flows through it. That is why a mixing desk can leave the supply on across a whole input bank regardless of what is plugged in.

Where it does harm is where the equal voltage condition fails. An unbalanced cable, a faulty cable with a shorted or swapped conductor, or hot plugging while the supply is on all put a real voltage difference across the coil, and 48 volts across 3.4 kilohms is 14 milliamps. A moving coil microphone will usually survive that. A ribbon microphone, whose moving element is a foil a few microns thick, may not. So the practical rule is not that phantom power is dangerous, it is that connecting and disconnecting things with it live is, and that the thing it damages is a cable fault you did not know you had.

And the mismatch that produces the common complaint. A phantom powered studio condenser plugged into a 3.5 millimetre jack through an adapter fails three ways at once: there is no 48 volts, the jack is unbalanced so the scheme cannot exist on it, and the 2.8 milliwatts available cannot run the electronics even if the capsule were polarised. The result is silence, or a faint and noisy whisper if the capsule is an electret after all. The reverse direction is the destructive one: an electret designed for 5 volts, fed 48 through 3.4 kilohms on an XLR input with phantom power on, is being given ten times its design voltage.

The one product on this site that publishes the mechanism

There is exactly one place in our catalogue where the scheme described in the drawing above appears as a specification, and it is not in a microphone listing. Our guide to signal generators includes a five in one audio tester whose specification row for its phantom check reads, in full, "48V on pins 2 and 3".

That is the whole mechanism in six characters, and it is more than every microphone listing on the site says between them. Pins 2 and 3, not pin 2, because the equal voltage on both is the entire point. The same product publishes output pads at minus 10, minus 20 and minus 40 decibels and a 1 kilohertz sine and pink noise generator, and it is a tester rather than a microphone, which is probably why it is the only listing that had any reason to describe the thing it measures. Worth knowing that such instruments exist, because the question of whether an input is actually supplying phantom power is otherwise unanswerable without taking the lid off something.

Our shotgun microphones guide is the nearest any of our own pages gets on the microphone side. It says that some of its picks draw plug in power straight from the camera, so there is nothing to charge and no level to monitor, while others add onboard controls that need internal power. Both halves are correct and the page is right about the practical consequence. What it does not say is that plug in power is a different standard from phantom power rather than a weaker version of it, running at a tenth of the voltage into a tenth of the impedance on a connector that cannot carry the phantom scheme at all. That distinction is the one that decides whether two pieces of equipment will talk, and it is the gap this page fills.

The scheme that did not go away, it just moved

One more piece of history is worth a paragraph, because its descendant is sitting inside three products we sell.

Analogue wireless microphone systems had a problem phantom power could not help with. A frequency modulated radio link offered maybe 55 decibels of signal to noise, and speech with any dynamics in it wants 80 or more. The solution was companding: compress the dynamic range before transmission, typically two to one, and expand it by the inverse after reception. A matched pair of curves, and if the match is exact the output is the input and the noise floor has been pushed down by the expansion.

The artefact when it is not exact is called breathing, and it is unmistakable once you know it: the background noise rises and falls with the speech, because the expander is moving the gain and taking the room with it. Modern 2.4 gigahertz digital links do not compand, because they send numbers and numbers do not accumulate noise in transit.

But the technique has reappeared without its other half. Our three lavalier listings publish, variously, a noise reduction chip, adaptive noise cancellation, AI noise reduction with voice enhancement, and automatic voice levelling. Automatic levelling is a compressor. It is applied at the transmitter, before anything is recorded, and unlike the analogue compander there is no matching expander at the other end, because the system is not trying to undo it. So the gain movement is permanent and the breathing is baked into the file. To their credit, all three records carry the right caveat in their own drawbacks, saying variously that noise cancellation reduces rather than eliminates background noise, that it is an aid and not magic in extreme noise, and that the system is not studio grade. None says the processing cannot be removed afterwards, which is the part that matters.

Problems in our own records

Three worth reporting.

The first is the near total absence of electrical specification. Of nine microphone and recorder products, exactly one publishes a supply requirement, and it publishes it as a minimum voltage inside a marketing bullet rather than as a specification. None names a powering standard. None states a current draw. The Zoom bundle includes a condenser microphone and says nothing anywhere about how it is powered. So a buyer cannot determine from any listing on this site whether a given microphone and a given input will work together, which is the single most common failure in this category and the one with the most alarming symptom.

The second is the specification corruption reported elsewhere in this batch, and all four products named here sit inside it. Every one of them falls in the product identifier range 3500 to 3900, where a detector for the signature finds 286 of 401 published products affected and none at all outside the range. Their specification labels read "Feature", or are cut off in the middle of a word, and their advantage lists are stored as one unbroken run of text. Every figure on this page was read from the summary and the advantage and drawback lists instead.

The third is a small one about the recorder bundle. Its stated contents include a lavalier condenser microphone, a windscreen and batteries, and its own drawbacks note that the accessories are entry level and that the batteries are not rechargeable. Fair enough. But the listing never says whether the batteries power the recorder only, which is what AAA cells in a pocket recorder normally do, or whether anything in the bundle needs power of its own. For a bundle whose whole proposition is that it is complete, the power path through it is not described at any point.

The honest limit, and the free technique that beats the purchase

The specifications that would settle this are two lines long and nobody prints either: what the microphone needs, and what the input supplies. Until they do, the three rules below cover almost every real case and cost nothing.

Match the connector first and the brand never. A three pin balanced connector can carry phantom power and a 3.5 millimetre jack cannot, so a microphone on an XLR and an input on a 3.5 millimetre socket are not a compatibility question, they are two different systems, and an adapter that fits mechanically changes none of that. The adapter is the trap: it makes a connection that cannot work.

If a microphone has no battery and no USB socket, it is drawing power from the input, and you need to know which kind. If it came with an XLR plug it wants phantom power. If it came with a 3.5 millimetre plug it wants plug in power. Those two facts are almost always inferable from the box even when the specification is silent, which it usually is.

And the free diagnosis, which is where this lands. If a professional microphone sounds like nothing, do not reach for gain, because gain will raise a noise floor and nothing else. Test the chain instead with something that needs no power at all: any dynamic microphone, or the same microphone on a different input that you know works. If the dynamic microphone works on that input and the condenser does not, the input is not supplying what the condenser needs, and no setting on either device will change that. That single substitution separates a powering fault from every other fault in the chain in about thirty seconds, and it is the test the whole of this page exists to justify.