How to Wire a Small Studio Without the Hum

You have a desk, an interface, two pairs of speakers and a hum. It has been there long enough that you only notice it between tracks, and somebody has told you to buy a ground lift, which is a three dollar adapter that removes the third pin from a power plug. It is the one fix on this page that reliably works and that you should still not use, because the pin it removes is what keeps a chassis from going live when something inside fails.

A studio that hums, or drifts off center when you turn it down, or cannot honestly tell you which of two pairs of speakers is better, has a wiring problem far more often than an equipment problem. Wiring is also the part nobody photographs, which is why it gets skipped, and it is the cheapest thing in the room to fix.

What the Hum Is Telling You

Three different faults produce a noise you would describe the same way. A loop: two boxes joined by both a cable and the building's earth, with current circulating around the ring. Induction: a signal cable lying alongside a power cable, a wall wart, or the back of a computer. Or a connection, meaning a plug half in or a design fault inside one box.

Each responds to something different, so you can separate them in two minutes. If the hum appears only when two particular devices are connected to each other, it is a loop. If it changes when you move a cable, induction. If it crackles when you wiggle a connector, a connection. If it is there with nothing plugged into the input at all, the noise is inside that box and none of the rest of this applies.

The mistake worth naming is treating a hum as a gear fault, which is appealing because a gear fault is actionable and it is somebody else's problem. The replacement interface goes back into a room wired exactly as before, and hums too.

Plus Four and Minus Ten, and Where Your Gain Goes

Professional gear runs at a nominal level of +4 dBu and consumer gear at -10 dBV. Almost every explanation stops there, which is why nobody can use it. The gap between them is about 12 dB, not the 14 most posts quote, because 0 dBu is 0.775 volts and 0 dBV is 1 volt.

What that does in your room is the useful part. Feed a -10 dBV source into a +4 dBu input and the signal arrives about 12 dB quieter than the input expects. You make it up downstream, and whatever you turn up to make it up turns up the noise sitting under it. Go the other way and you are 12 dB hot into a stage with less headroom above nominal to begin with, so peaks square off before the meters have told you anything. Hence the rule that governs the rest: put the gain early and the attenuation late.

The small studio usually does the opposite. The speakers are loud, so the software fader comes down to a third. Now the converter delivers a fraction of the voltage it was built to produce while the amplifier inside each speaker still runs at full gain, hissing exactly as much as it always did. The music got quieter and the noise did not, which is the whole mechanism behind a desk that sounds clean loud and grainy quiet.

The fix is a setup step rather than a purchase. Software fader at unity, interface output at nominal, sensitivity trims on the back of the speakers turned down. Then set the controller where you would call it comfortably loud and bring those trims up until it is. Your normal level now sits in the upper half of the controller's travel instead of the first few degrees of it.

What you connected	What arrives	What it costs you
A -10 dBV source into a +4 dBu input	About 12 dB quieter than the input expects	You make it up downstream, and whatever you turn up to make it up turns up the noise sitting under it
A +4 dBu source into a -10 dBV input	About 12 dB hot, into a stage with less headroom above nominal to begin with	Peaks square off before the meters have told you anything

What Balanced Actually Buys

A balanced connection carries the signal on two conductors of opposite polarity, with the shield doing nothing but shielding. The input at the far end subtracts one from the other, so anything that got into both equally, which is most interference, subtracts to nothing. That is common mode rejection, and it is the entire trick.

Two consequences fall out and both get missed. The rejection happens at the receiving input, not in the cable, so a balanced cable into an unbalanced jack is an unbalanced connection with a nicer plug on it. That is why better cable for a consumer source achieves nothing, and the argument at length is in What Cables Can and Can't Do.

And on an unbalanced connection the shield is a signal conductor. It carries the audio's return current because nothing else will, so any other current in that shield, including current circulating in a ground loop, arrives at the input indistinguishable from music. On a balanced connection nothing in the shield is signal, and the same current passes through unheard. That is why a room wired balanced usually does not hum, and it is a better reason to go balanced than any claim about sound quality.

• Balanced: three pin XLR, and TRS quarter inch, where tip is hot, ring is cold and sleeve is shield. TRS and TS plugs look alike at arm's length, so count the rings on the shaft.
• Unbalanced: TS quarter inch and RCA. Fine over short runs in a quiet electrical environment, and where hum problems live.
• Impedance balanced: a common design where the cold pin is not driven but is tied to ground through the same impedance as the hot. Half the output voltage, all of the noise rejection, and not a corner being cut on you.

One thing to watch when you cross between them: a balanced output usually delivers about 6 dB more into a balanced input than the same box's unbalanced output does, so levels change when connector types do.

Where the Loop Comes From

Two boxes, each with a three pin power cord, are both tied to the building's safety earth. Join them with a shielded audio cable and you have built a closed ring: out through one power cord, through the panel, back through the other, and home along the shield. The loop exists the moment that second cable goes in, and nothing is broken.

Two things push current around it. Different outlets sit at slightly different potentials, sometimes only millivolts and more so on different circuits, and that difference drives current down the path of least resistance, which is your audio shield. And the ring encloses an area, so any alternating magnetic field crossing it induces a current. That is why an identical rig hums in one room and is silent in another.

A third cause behaves like a loop but is a fault in one box. Well designed gear lands the shield on the chassis where the connector enters, so shield current goes straight to metalwork. Badly designed gear lands it on the internal audio ground, so that current runs through the circuit's own ground reference on the way out. That is the pin 1 problem, and it means one box in an otherwise quiet rack will hum on a balanced connection, where balanced is supposed to save you.

Computers bring more paths than anyone counts: USB to the interface, HDMI to a display, ethernet to the router, a charger. Each is another conductor that can close a ring. The charger has a habit of its own, though. A two pin switch mode supply leaks a small current onto the chassis, which shows up as a buzz that appears when you plug the laptop in, vanishes on battery, and is not a loop at all.

The Order to Diagnose It In

The order is the value here. Most people start at step six, buy something, and end up no wiser.

1. Get a baseline. Interface to controller to one pair of speakers and nothing else. Pull the USB from the interface as well as the wall, and if the interface is bus powered, put it on its own supply for this test. If that hums, the fault is in those three cables or in one of those three boxes, and the sections above tell you how to separate them: unplug the input and see whether the hum survives, which finds a noisy box, then watch for the box that only hums once something else is connected to it, which is pin 1. If it is quiet, the noise arrives with something you are about to add back.
2. Add one source at a time and listen after each. Turntable, television, computer, second speaker pair. Whichever brings the noise is the one to work on. This sounds slow and takes four minutes.
3. Put the whole desk on one circuit. It does not remove the loop. It shrinks the potential difference driving current around it, which is usually enough, and it is the most effective free change in most rooms.
4. Check whether the offending link is balanced at both ends. If it is unbalanced you have your answer and can stop diagnosing.
5. Move the cable before you buy anything. Off the power cords, away from wall warts and dimmers, crossing mains at right angles rather than running alongside it. If that changes the noise it is induction, and no isolator will fix it.
6. Now fix it. First choice, convert the link to balanced at both ends if the gear allows, because that removes the mechanism rather than treating it. Otherwise insert a one to one isolating transformer on that link: it passes the audio and breaks the path the shield current needs. The Ebtech Hum Eliminator is the one most people can justify and the Jensen ISO-MAX is the better transformer, and either is the right answer for a turntable or a television that will never be balanced. If the hum arrives with a cable or antenna feed, put a coaxial isolator on that feed instead.

Two things not to do, and for opposite reasons. Never pull the safety pin: it works, and it works by opening the loop at the one point in the system that exists to keep you alive. A ground lift switch on a DI box or an interface is a different thing, disconnecting pin 1 of an audio connector rather than the mains earth, and it is safe to flip. And do not buy a power conditioner to cure a hum, which fails the other way round: it is perfectly safe and it mostly does not work. One built around isolation transformers can help, at many times the price of a line isolator on the link that needs it, and a strip with a surge light does nothing for hum at all.

Where the Monitor Controller Sits

After the converter, before the speakers, with nothing between it and the speakers but cable. That is the gain rule from two sections ago applied to a box: attenuate any earlier and the level control sits on the wrong side of the things that make noise. The controller belongs after everything that generates noise and immediately before the thing that makes sound, so when the music gets quieter, so does the floor underneath it.

The position also makes the controller the star point of the room. Every source and every speaker pair meets at one box, which is a better topology than a chain because a star has fewer rings in it. It is also why a hum sometimes arrives on the day the controller does: the box that gathered everything into one place closed a loop that used to be open, and the instinct is to decide the new box is faulty. Run the diagnosis above before believing that.

The position also sets a limit. Anything downstream of the controller is outside its control: a subwoofer with its own level knob, a headphone amp fed from a spare speaker output, an EQ patched into one pair only. Each adds gain the controller cannot see, which quietly puts your two pairs back out of match. Matching them was the point of all of this.

Matching Two Pairs to the Same Level

Louder sounds better. It is the most reliable finding in this whole field and it does not care that you know about it. A decibel is enough to decide a comparison, so if pair A is 2 dB louder than pair B you will prefer pair A, you will have reasons, and every reason will be a story you told yourself about loudness.

1. Generate pink noise at minus 20 dBFS, which is the studio convention and which any DAW will do.
2. Get a meter, set to C weighting and slow response. A phone app is fine here, because you are comparing two readings taken the same way rather than establishing an absolute, and relative accuracy is the easy part.
3. Put the meter where your head goes, at ear height, and leave it there for every reading. On a stand if you can, because you are a large soft reflector and you change the answer.
4. One speaker at a time. Mute the other. Two speakers playing the same noise interact, and the sum shifts with a few centimeters of meter position, so a pair measured together tells you about the room rather than the speakers.
5. Fix the controller at one position and write it down. It does not move again until you are finished.
6. Read all four and adjust until they agree, within half a decibel, which is about as repeatable as a reading off a phone is anyway.
7. Check the match again much quieter. If it holds, you are done. If it does not, the attenuation is not tracking evenly between channels, which is a different problem from the one you just solved.

This is where per output trim stops being a line on a spec sheet. Without it, matching means turning the recessed pot on the back of each speaker, which some speakers do not have, which nobody can read reliably, and which one guest leaning on a cabinet can undo. With it the match lives in the controller and takes ten minutes to redo when a pair changes.

So that is the line to look for on a monitor controller spec sheet, and it is not the one the marketing leads with. An independent level trim on each speaker output, in numbered steps you can write down rather than a recessed pot that needs a screwdriver and a flashlight. The PreSonus Monitor Station V2 puts a trim on each of three speaker outputs for around three hundred dollars, which is the cheapest way to own the feature. The Drawmer MC2.1 does it for two pairs in the four to five hundred range, with a better attenuator and fewer buttons on the front. At the top the Grace Design m905 calibrates each of its three outputs in half decibel steps and holds a reference level you can return to, which is the difference between matching your speakers once and being able to prove a year later that they still are. It runs to a few thousand dollars, and I was employee number one at Grace Design in 1997, so weigh that with my history attached to it.

While the noise is running, settle the center image, the other symptom that gets blamed on the wrong part. An image that sits left at low volume and centers as you turn up is usually channel balance rather than a speaker. Swap the left and right cables at the controller's outputs: if the image follows the cables the fault is upstream, and if it stays put it is the speaker or the room. Then check polarity by playing the noise in mono, where it should appear as a hard, small point hanging between the cabinets. If it is wide and vague, one channel is out of phase and no amount of placement will fix it. For the placement half, How to Set Up Speakers in a Real Room covers the rest.

The Short Version

Wire the monitor path balanced end to end, XLR or TRS at both ends, and accept unbalanced only where the source gives you no choice. Put the whole desk on one circuit. Set the gain early and attenuate once, at the controller: software fader at unity, interface output at nominal, speaker trims set so your normal listening level lands in the top half of the controller's travel.

When something hums, diagnose in order. Baseline with nothing connected, one source at a time, everything onto one circuit, check whether the offending link is balanced, move the cable, and only then buy something. The something is a one to one isolating transformer on the link that cannot be balanced. It is never the safety pin.

Then match the two pairs with pink noise and a meter, one speaker at a time, from the seat, within half a decibel, before you compare anything through them again. That hour turns a second pair from a decoration into the most useful check in the room, and it is the reason per output trim is worth paying for. If you have money for exactly one thing here, make it a set of balanced cables and an afternoon with a meter.

---

Or Bring the Rig In

The listening room at Free River Studios, 215 E. Lewis St #301 in Livingston, is on the third floor, and it is a working studio as well as a room to listen in, so all of the above has been argued out in it once already. Bring your own controller and a handful of cables and we will put a meter on it. It is by appointment.

How I Listen sets out what happens before anything here gets recommended.