August 2026 • sapi_44530a

Cable runs, ground loops and the hum I chased across three visits

The call came in on a Tuesday: “There’s a hum in the left channel. Only the left. It comes and goes.” I’d installed that system eighteen months earlier. Two subs, a five-channel amp, a media PC feeding a DAC, and a turntable the owner swore he used. The hum was new. Or the hum was always there and he’d just noticed. You never know which until you’re standing in the room.

Three visits. Two of them were my fault for not measuring first. What follows is the actual sequence, because ground loop troubleshooting is not a checklist. It’s a process of elimination where half the eliminations are wrong.

Visit one: I assumed it was the turntable

Left channel only is a clue, not a diagnosis. My first guess was the phono stage. Moving-magnet cartridges are sensitive to everything, the tonearm cable is a long unshielded run by design, and the turntable sat on a shelf next to a dimmer switch. I disconnected the phono stage from the preamp entirely. The hum stayed.

That’s the first lesson, and it cost me an hour of driving. Disconnect at the source, not at the destination. Pulling the cable from the preamp input tells you the preamp is quiet. Pulling it from the phono stage tells you the cable and the preamp together are quiet. Different answers. Do it in the right order and you skip a trip.

With the phono stage out, the hum was still in the left channel of the main amp, with the media PC running and the DAC connected by USB. So it wasn’t analog source related at all, which I should have heard from the character. A 60 Hz tone with a buzzy edge is ground. A 60 Hz tone with a cleaner sine shape and harmonics at 120, 180 is more often a transformer or a cable picking up field. This one had the buzz. Ground.

What “ground loop” actually means here

Two pieces of gear, both plugged into safety earth, both connected by a signal cable whose shield is tied to chassis at both ends. You’ve just built a loop. Current from the building’s wiring, or from a potential difference between outlets on different circuits, flows through that loop and the shield becomes an antenna that’s also a conductor. The signal cable is now carrying the noise it was supposed to reject.

In this room the media PC was on a different circuit from the amp. Same panel, different breaker. That’s enough. I ran an extension cord from the amp’s outlet to the PC. Hum dropped by maybe 10 dB but didn’t disappear. So the circuit split was part of it, not all of it.

Visit two: the cable run was the stupid part

I’d run the line-level cable from the DAC to the preamp along the same path as a bundle of AC wiring. Twelve feet of it, parallel to the Romex for most of the run. That’s a textbook mistake and I made it because the wall was already open and the path was convenient. Convenience is how you get hum.

I re-ran it. Separated by six inches minimum, crossed the AC at 90 degrees where I couldn’t avoid it, and swapped the cheap RCA for a shielded pair with a proper braided screen. Hum mostly gone. Buzzy edge gone entirely. What remained was a low, steady 60 Hz with almost no harmonics, audible only with your ear at the woofer, and only on the left.

That residual was not a ground loop. That was the left channel input stage picking up field from the amp’s own transformer, because the left input sits closer to it than the right. Some amplifiers are laid out better than others. This one wasn’t. I could have chased it further. I moved the input cable two inches to the right and it dropped below the noise floor of the room. Two inches.

Visit three: the thing I should have checked first

Two weeks later the owner called again. Hum was back, louder, and now in both channels. Same system, nothing changed on his end. “Nothing changed” is always a lie, and it’s never a malicious one. He’d added a network switch and a NAS in the same rack. Both had switching power supplies. Both were plugged into the same power strip as the preamp.

Switching supplies are not the same animal as a linear transformer. They leak high-frequency noise back onto the ground and the AC line, and a preamp with a modest power supply rejection ratio will happily amplify it. I moved the network gear to a different strip, on a different circuit, and put a ferrite clamp on the NAS supply cable. Hum gone. Hasn’t come back in eight months.

Here’s my honest verdict on the whole exercise: I solved this three times because I kept fixing the symptom in front of me instead of mapping the grounds first. A twenty minute audit at the start would have found the circuit split, the parallel cable run, and the switching supplies in one pass. Instead I found them one visit at a time.

The audit I now do before touching anything

  • Draw the system. Every box, every cable, every power connection. Mark which outlet, which circuit, which panel.
  • Identify every path to earth. Safety ground on each power cord, shield connections on each signal cable. Any two earths joined by a signal cable is a loop.
  • Note every switching supply. Routers, NAS boxes, phone chargers, USB hubs, LED drivers. They all belong on a circuit away from the preamp if you can manage it.
  • Check the line-level runs against the AC runs. Six inches of separation, 90 degree crossings, shielded cable. No exceptions for short runs.

None of that requires test gear. It requires a notepad and the discipline to finish the map before you start pulling cables.

What I use to confirm, not to diagnose

A cheap multimeter with an AC millivolt range will tell you if there’s a potential difference between chassis grounds. If you read more than a few millivolts between two pieces of gear that are also connected by a signal cable, that’s your loop. A clamp meter around the signal cable reading current is the other confirmation. If you don’t have either, a pair of clip leads and your ears will do, but you’ll be slower and you’ll guess more.

What I do not recommend is the ground lift adapter on the power cord. It works, in the sense that the hum stops. It also removes safety earth from a chassis that was designed to have it. I’ve seen it done in installed systems and I’ve told the owners to stop. Fix the loop at the signal end or the circuit end. Don’t unhook the safety ground.

One more thing worth saying plainly. A hum you can hear at the listening seat and a hum you can only hear with your ear against the grille are different problems. The second one is often not worth solving. The first one always is, and it’s almost never the cable the owner suspects. It’s the one nobody drew on the diagram.