Digital Microphone Cable: AES3, Dante and USB Compared

A studio buys a new AES3 interface and the first cable out of the drawer is a 10 m mic cable. It locks, the meters move, and everyone assumes the job is done. Two weeks later a second converter is added, the sample rate goes to 96 kHz, and the same cable starts producing occasional clicks that no one can trace. The cable did not change. The signal did. Digital microphone audio is carried on at least three different physical formats, and each one has its own impedance, connector and distance rule. Treating them as interchangeable is how intermittent faults get built into a system.

The 110 Ohm Requirement Behind AES3

An AES3 line is specified at 110 ohm, and that number comes from the transmitter and receiver design, not from marketing. When the cable impedance departs from it, part of each transition reflects back down the run and arrives at the receiver out of position. At 48 kHz there is enough timing margin to absorb small reflections. At 96 kHz or 192 kHz the same cable sees the reflection closer to the sampling instant, and the error rate climbs. Digital cable reaches 110 ohm through conductor spacing, dielectric material and shield geometry; a microphone cable with the same XLR shell may sit anywhere from 60 to 120 ohm because nobody specified it. The connector is identical, the electrical part is not.

How Far AES3 Runs and What Cable It Needs

Published AES3 distance is about 100 m over 110 ohm cable, and that figure assumes a compliant cable, correctly terminated connectors and a receiver inside tolerance. Every patch point adds a discontinuity, so an install with a wall plate and two patch panels should plan well under that ceiling. Sample rate matters too: a run that is solid at 48 kHz can become marginal at 192 kHz because attenuation has not changed while the timing window has halved. Balance the plan against the highest rate the system will ever run, not the rate it runs on installation day, and keep a spare channel for later growth. A patch bay in the middle of a run costs more margin than the last twenty meters.

Why a Mic Cable Can Pass AES3 at 3 m

A two meter mic cable between two converters often works, and that is the trap. At short length the reflections are small and land early, so the receiver recovers the clock and reports no errors. Nothing about the cable has become correct; the run is simply too short to expose the mismatch. Add one connector, double the sample rate or move to a longer run and the margin disappears. Systems assembled this way fail in ways that look random and get blamed on the converter, the driver or the computer.

Seventy-Five Ohm Coax, BNC and Baluns

AES3id sends the same data unbalanced on 75 ohm coax with BNC connectors and reaches several hundred meters on proper video-grade cable. Converting between the two formats requires a balun or an active converter that transforms 110 ohm balanced to 75 ohm unbalanced and keeps the shield reference intact. A passive adapter that solders two coax conductors to pins 2 and 3 does not do that, and it usually produces a link that works until the cable is moved. Keep the conversion at one labeled point in the rack so a later fault can be isolated without guessing which format is present.

Networked Audio Over Cat5e and Cat6

Dante and AES67 carry audio as packets over Ethernet, so the physical layer is structured cabling rather than a dedicated audio pair. Cat5e or Cat6, RJ45 plugs and a switch that can handle the packet rate are the requirements, and the 100 m limit applies per link, not across the entire path. Because audio is packetized, a damaged cable usually produces latency spikes, a renegotiated link speed or a channel that drops and returns, not a quiet hiss. That changes fault finding: read the port error counters first, then look at the cable. A link that has never shown errors is probably not the cause of a clock problem. Keep a spare patch lead and a spare switch port in the rack.

Cat5e or Cat6 for Dante

Cat5e carries gigabit Ethernet at 100 m and handles most Dante systems at 48 kHz. Cat6 tightens crosstalk margin and is the better choice in dense racks or where cable shares a tray with power. Both are 100 ohm nominal, so the interface does not care which you buy; the difference is headroom against interference. Termination quality matters more than category: a poorly punched pair or a solid-core patch lead that gets flexed every day will cause more dropouts than the category choice ever will.

What a Switch Adds to the Path

A switch is not a cable, but it changes how a cable fault presents. On a Dante network the clock is derived from the audio stream, so one marginal link can show up as instability across several devices rather than one dead channel. Keep audio traffic on its own switch or VLAN where the budget allows, count every patch point between device and switch, and read the port error counters before replacing hardware. If errors are zero at both ends, the cable is doing its job.

USB Microphones and Their Data Cable

A USB microphone carries audio and power on one cable, so the conductor gauge is part of the power budget. USB 2.0 allows a 5 m passive cable, and a bus-powered mic is already close to what a single port supplies. Thin conductors raise the voltage drop; long or poorly shielded cable raises both drop and bit errors. The failure does not sound like degradation, it sounds like a device that will not enumerate, drops out when the buffer fills, or clicks in time with disk activity. Choose a cable rated for the current, keep it short, and do not extend it with a passive coupler. The cable that ships with the mic is usually matched to its current draw.

Dropouts, Clocking and Fault Finding

Digital faults are counted before they are heard. A receiver that cannot lock either mutes, repeats a block or flags a CRC error, so the audible result is a click or a dropped channel, never a gradual loss of treble. Start by confirming the format and the cable impedance, then change one variable at a time: cable, connector, port, sample rate. If the fault follows the cable, replace it. If it follows the device or the rate, copper was never the answer. Keeping one known-good cable of each format in the rack turns this from an afternoon into ten minutes. Write down what changed before the fault appeared, because digital failures usually start with a format or a sample rate.

Digital formats and the cable each one needs

Real-World Application Scenarios

Three deployments show where the decision lands. A post-production room wiring two converters over AES3 needs 110 ohm cable and a patch bay plan, because the run crosses a rack and adds two connectors. A live venue distributing 32 channels to amplifiers uses Dante on Cat6 with a dedicated switch, and the work is termination and labeling. A podcast studio with four USB microphones needs short shielded cables and a powered hub. In each case the connector looks familiar and the electrical requirement is not, which is why the format belongs on the purchase order.

Related Guides on This Site

Related reading covers impedance conversion and connector-level detail: BNC to XLR digital audio balun guide, analog XLR audio cable construction, and USB-C to XLR audio adapters. Work through the connector side and the impedance question before specifying a format for a new rack or a run between rooms.

Frequently Asked Questions (FAQ)