There is a switch on the front of your interface marked 48V, and somewhere on a forum there is a warning that pressing it will kill a ribbon microphone. Both halves of that sentence are true often enough to be repeated forever, and the connection between them is not the one most people assume. Phantom power does not travel down a separate wire. It rides on the two conductors that carry your signal, which is why a fault in the cable and a fault in the microphone are the same event.
Why it is called phantom
A balanced microphone cable has three conductors: two signal legs and a shield. The standard for 48-volt phantom power puts the supply on both signal legs at once, through two resistors of 6.81 kilohms, with the shield as the return path. The supply itself sits at 48 V, plus or minus 4, and is limited to 10 mA per microphone.
The two resistors have to be matched to within 0.1 per cent, and that requirement is the whole design. A microphone reads the difference between the two legs; if both sit at the same voltage, the difference is zero and the microphone sees no supply at all — hence the name. A condenser taps the supply against the shield deliberately, because it needs power for its impedance converter. A dynamic ignores it. A passive ribbon, which is a dynamic microphone with an unusually delicate moving part, also ignores it — as long as the two legs really are equal.
What is actually hanging in the field
The moving part is worth picturing before discussing what damages it. A ribbon element is a strip of corrugated aluminium foil suspended between the poles of a magnet. Published thicknesses across three well-known designs: 0.6 microns for the classic Coles 4038, 1.8 microns for AEA's ribbons, and 2.5 or 4 microns for Royer's. On one AEA stereo model the strip measures about 60 by 4.7 millimetres at 1.8 microns thick.
For scale: kitchen foil is roughly 16 microns. The thinnest of these ribbons is about one twenty-fifth of that, cut into a strip narrower than a pencil and hung in a magnetic gap so that sound alone can move it. That fragility is why it sounds the way it does, and why it does not tolerate a direct current.
Why a correct cable is a non-event
Both manufacturers most associated with modern ribbons say so plainly. Royer's position is that their ribbons are not usually affected by phantom power, and that switching it on or off with the microphone connected should pose no problem. AEA's manual for its passive stereo model agrees, with the exception stated precisely: the danger is a cable in which the shield is shorted or miswired to a signal leg, or a hot patch through a patchbay.
The reason follows from the arithmetic. With both legs at 48 V, the voltage across the ribbon is zero and no current flows through it. Break that symmetry and the picture changes at once. Short one signal leg to the shield and the supply drives current through a single 6.81-kilohm resistor: 48 divided by 6810, which is 7.05 mA. That current has to return somewhere, and in a miswired cable the return path runs through the ribbon.
Seven milliamps sounds small. Against a ribbon it is not. The strip generates a few tens of microvolts as it moves, and the current that produces in normal use is smaller than the fault current by three orders of magnitude or more, on our reckoning of typical ribbon impedance. A direct current of that size in a magnetic gap does not make the ribbon sing: it pushes the foil one way and holds it there while heating it, which either stretches it out of its corrugation or tears it.
The moment of connection is the dangerous one
Steady state, then, is safe with a correct cable and lethal with a faulty one. The third case is neither: it is the instant a connector is inserted or removed while the supply is live.
Inside every phantom-capable input sit coupling capacitors, charged to something near the phantom voltage while the supply is on. When a connector makes and breaks contact during insertion, they can be shorted to ground through whatever pins touch first — a discharge the 6.81-kilohm resistors do not limit, because it does not come through them. Design literature on these input stages treats the transient as large enough to require clamp diodes rated for momentary peaks around an amp, a figure from a different world than the standard's tidy 10 mA.
A three-pin XLR is shaped to reduce this: the shield pin is longer and mates first. A quarter-inch jack in a patchbay does the opposite — as the plug slides in, its tip and sleeve sweep across contacts they will not rest on, and for a few milliseconds the supply lands in the wrong places. The patchbay is not dangerous because it is a patchbay; it is dangerous because a sliding contact guarantees the transient a mating connector avoids.
The rules that follow
All the sensible practice falls out of the three cases above, none of it requiring the belief that 48 volts is inherently hostile to ribbons.
- Connect and disconnect with the supply off, and give the input a few seconds afterwards. The coupling capacitors need time to discharge; switching the button off and pulling the plug in the same second is the transient you were avoiding.
- Test the cables you use with ribbons. A continuity tester costs less than a dinner and finds exactly the fault that matters — shield shorted to a signal leg, or the two signal legs swapped at one end only.
- Keep passive ribbons out of quarter-inch patchbays carrying phantom power. If the routing has to exist, patch with the supply globally off.
- Do not run a passive ribbon through a cheap splitter or adaptor chain on a live supply — every extra contact is a chance for the two legs to stop being equal.
- Check whether your ribbon is active. Active ribbons contain an internal preamplifier and require the supply — switching it off does not protect them, it silences them.
The recommendation, including the one against buying
If your signal path contains a quarter-inch patchbay, a splitter feeding two destinations, or cables of unknown origin, do not buy a passive ribbon yet. Buy the cable tester first, or buy an active ribbon, which tolerates the supply because it expects it. The failure being avoided is not gradual: a stretched ribbon loses output and gains rattle in one event, and re-ribboning costs a meaningful fraction of the microphone.
And if your path is a straight run from microphone to interface with a cable you have tested, then the folklore does not apply to you. Leave the ribbon plugged in, leave the supply off for the channel it sits on, and stop worrying about a switch that cannot reach it.
The check that takes a minute
Take the cable you use with the ribbon and a continuity tester. Confirm that each pin connects only to its own number at the far end and to nothing else. That single test separates the cable that makes phantom power a non-event from the cable that makes it a repair bill.
How this was put together
Five independent sources sit under the figures above: the international standard for phantom powering, which specifies 48 V plus or minus 4, two 6.81-kilohm feed resistors matched to 0.1 per cent, and a 10 mA per-microphone limit; Royer Labs' published position that its ribbons are unaffected by phantom power in normal use; AEA's owner documentation for a passive stereo ribbon, naming shorted, miswired and hot-patched connections as the actual risks; published ribbon thicknesses for the Coles, AEA and Royer designs, and the ribbon dimensions of the AEA model; and design literature on 48-volt input stages describing coupling-capacitor discharge during connection and the clamp diodes specified to survive it.
The derived figures are ours: the 7.05 mA that a single feed resistor delivers into a shield-to-leg short, the comparison of that current against the microamp-scale current a ribbon generates acoustically, and the reading of why a sliding quarter-inch contact produces the transient that a mating XLR is shaped to avoid.
