Daniel's looking at a PA speaker. It's got an RCA input, a 3.5mm input, and an XLR input that's meant for a microphone. And he wants to know how far he can run each one before it stops sounding right.
Which is a much harder question than it sounds.
That's the thing. He wrote in that even audio engineers he's met don't have an intuitive grasp of how long you can run RCA before you're rolling the dice on degradation. And he wants the balanced versus unbalanced question answered properly. What each of those options actually offers at different run lengths. He's framing it against the IEC and display cable conversations we've done before, the same way of thinking about where a cable stops being a cable and starts being a component.
And the honest answer is that the numbers people quote are all over the map. You'll see ten feet for RCA from one source and a hundred and fifty feet from another. That's not a rounding error. That's two completely different claims about how the world works.
So let's dig into the physics and the practical guidance, and why the two often don't match.
Start with what the wire actually is. Unbalanced cable, so RCA, quarter-inch TS, 3.5mm, that's two conductors. You've got the center signal conductor and you've got the shield, and the shield is doing double duty as the return path. That means anything the shield picks up gets injected straight into your signal. There's no second copy of the signal to compare against.
Whereas balanced.
Three conductors. Hot, cold, ground. The hot and cold carry the same signal in opposite polarity. Any interference that hits the cable hits both conductors essentially equally, and at the receiving end the input stage subtracts the two. What was common to both disappears. That's common-mode rejection. It's not that balanced cable is magic wire. It's that the receiving circuit gets to throw away everything the two conductors agree on.
And the shield in a balanced cable isn't carrying signal at all.
Right. It's grounded, it's doing shielding, and it's not part of the audio path. That's the structural difference. Unbalanced has two jobs crammed onto one conductor.
So that's the noise story. But Daniel's question is about length, and length is a different mechanism.
That's where it gets interesting. The length limit isn't really about the cable rejecting interference. It's about capacitance. Every cable has capacitance between its conductors, distributed along the whole run. Typical figure is somewhere between a hundred and twenty and three hundred picofarads per meter. Cheap PVC-insulated cable sits up around three hundred. Polyethylene gets you down around a hundred and fifty.
And capacitance against what?
Against the output impedance of whatever's driving the cable. You've got a resistor on one end and a capacitor along the wire, and that's a low-pass filter. The cutoff frequency is one over two pi times source impedance times total capacitance. Total capacitance being capacitance per meter times length. So the cable doesn't have a frequency response on its own. It has a frequency response in the context of whatever is feeding it.
Which means "how long can I run this cable" is not a property of the cable.
Not even slightly. Change the source and you change the answer, with the same wire.
Give me the actual numbers, because I want to know how bad this really is.
Worst-case simulation that Tonestack published, and this is the pessimistic end. Ten meters of cable, three hundred picofarads per meter, driven by a six hundred ohm source. Attenuation at twenty kilohertz is zero point two one decibels. That's it. Twenty kilohertz is above what most adults can hear anyway.
So the ten-foot rule is not about treble loss.
Not remotely. And the table makes it clearer. With that same three hundred picofarad cable, a six hundred ohm source hits zero point one decibels of loss at seven meters, which is twenty-three feet. Half a decibel at sixteen meters, fifty-two feet. Now swap to a hundred ohm source, and you get zero point one decibels at forty meters, a hundred and thirty-one feet. Half a decibel at ninety-three meters. Three hundred and five feet.
Same cable.
Six times the length for the same loss, because the source impedance dropped by a factor of six. And if you use hundred and fifty picofarad cable instead of three hundred, you double all of those lengths again.
So a hundred-ohm source on good cable gets you past six hundred feet before you've lost half a decibel at twenty kilohertz.
Which is why the ten-foot rule is nonsense as a general statement. But here's the counterintuitive part, and this is where most people's intuition is actually backwards. Line level is harder to drive long than mic level.
Say that again.
Mic level is tiny. Minus sixty to minus forty dBu. A few thousandths of a volt. Line level is around a volt, roughly a thousand times stronger. Everyone assumes a bigger signal travels better. It doesn't, because the limiting factor for a line-level signal isn't voltage, it's current. You have to charge and discharge that cable capacitance every cycle, and the higher your signal voltage, the more current that takes.
Slew rate.
EAW worked the numbers. A device outputting plus twenty-four dBu, which is about twelve point three volts RMS, driving three hundred feet of thirty-two picofarads per foot cable at thirty kilohertz, needs thirty-one milliamps to charge the cable. The device only supplies twenty-nine milliamps into a six hundred ohm load. So it can't. It slew limits. And slew limiting doesn't just roll off the top end, it generates high-frequency intermodulation distortion.
So the loud signal is the one that runs out of current.
With a ten kiloohm load you get about two hundred and sixty feet of usable length. With four ten kiloohm loads in parallel, two thousand five hundred ohms, it drops to two hundred and thirteen feet. Derate that by twenty percent for safety and you're at a hundred and seventy feet.
And the mic signal?
A hundred ohm mic on five hundred feet of thirty-two picofarad cable, minus three decibels at a hundred kilohertz. Comfortable. Swap in a six hundred ohm distribution amp output and the same five hundred feet collapses to seventeen kilohertz. That's audible.
Because the source impedance changed.
Same cable, same length, same signal level. Seventeen kilohertz versus a hundred kilohertz.
One more myth while you're at it, because I've heard people talk about audio cables as transmission lines.
They're not. Transmission-line effects kick in when the cable length approaches a quarter wavelength. At twenty kilohertz, a quarter wavelength is about two point three miles. Three point seven kilometers. Unless you're running a cable across a small town, you're not dealing with transmission-line behavior. You're dealing with an RC filter and a noise antenna.
Which is the part I keep coming back to. Daniel asked specifically about RCA and how long you can run it, and the answer isn't a number.
The frequency response of the cable on its own is meaningless. That's the direct quote from Tonestack, and it's correct. If you tell me a cable is ten meters long, I can't tell you anything useful. I need the source impedance, the capacitance per meter, the length, the load at the other end, and what's in the room around it.
Five variables.
And that's before we get to the practical guidance, which is a mess.
So we've seen the physics. Now let's look at what the practical guidance actually says, and why it's all over the map.
Kettner Creative says XLR is theoretically good for a thousand feet, three hundred meters, and RCA shouldn't be run more than ten feet.
Ten feet.
Three meters. AudioCalcs gives you a different set entirely. XLR around a hundred meters. Balanced TRS about twenty-five. Quarter-inch TS about six. 3.5mm TRS two to three meters. RCA five to six meters. HomeRecordingSetup says unbalanced starts losing highs past about twenty feet. Home Studio Guys says under fifteen to twenty feet for unbalanced, over a hundred feet for balanced is routine. Hosa says keep unbalanced under fifteen to twenty feet. And then Calendar-UK, in the same page, says both fifteen to twenty feet maximum and, quote, "Most RCA audio cables that are built reasonably well with shielding can be run effectively from one hundred to two hundred feet."
In the same article.
That's not a source with a view. That's a source that copied two paragraphs from two different places and never noticed they disagree.
So for RCA the published answers range from three meters to two hundred feet. That's a sixty-to-one spread.
And the reason is that there's no single dB-per-meter figure, because each cable class fails by a different mechanism. Unbalanced long runs fail by picking up noise and by creating ground loops. Long runs from a high-impedance source fail by treble rolloff. Long runs from a high-level source fail by slew limiting. Those are three different failure paths and they kick in at wildly different lengths.
So the limit is not a property of the connector.
The limit is a property of the system. Source impedance, cable capacitance, length, load, and the electromagnetic environment. Five variables. Nobody hands you five variables in a table because a table can't hold them.
Let's make that concrete, because I think people hear "five variables" and it stays abstract. Walk me through one specific setup.
Take a laptop with a 3.5mm output. Output impedance on those is often somewhere between fifty and a few hundred ohms, honestly it varies wildly and nobody publishes it. Say it's two hundred ohms. You run thirty feet of typical consumer RCA, call it two hundred and fifty picofarads per meter, so about nine meters, total capacitance around two and a quarter nanofarads. Cutoff frequency comes out somewhere north of three hundred kilohertz. You will never hear that. The treble loss is not the problem.
So what is the problem?
The problem is that the laptop output is unbalanced, the shield is your return, and you've just laid thirty feet of antenna next to a power strip and a monitor. If there's a ground potential difference between the laptop and the speaker, that difference rides on the shield and shows up as hum. If there's a switching supply nearby, you get whine. None of that is on the capacitance table.
And if you swap the laptop for a cheap USB turntable with a high-impedance output?
Then the capacitance matters. High output impedance, say ten kiloohms, same cable, and now your cutoff is down in the audible range. You get the dull, closed-in sound people describe. Same thirty feet, same connector, completely different failure mode, because the source changed.
So the cable was never the variable that mattered.
The cable was one of five. The source is the one that flips the answer.
That brings us to the PA speaker scenario. What happens when you try to send line-level audio into a mic-level XLR input?
That XLR input on your PA speaker is almost certainly wired for a microphone. Mic level, minus sixty to minus forty dBu. And you want to feed it a signal from a phone or a laptop, which is line level, about a volt. That's roughly a thousand times stronger. Shure's guidance on this is blunt. Connecting a line-level source to a mic-level input produces sound that's loud and distorted, because the input simply isn't built to accept that much signal.
So it'll work in the sense that sound comes out.
Sound comes out. It just comes out clipped and ugly, and if the speaker has any gain staging on that input you may also be driving the preamp into clipping before the amplifier even sees it. The fix is a pad, an attenuator, a DI box, or a cable with resistors built into it. Something that knocks the level down before it hits the input.
And there's a second problem there, which is impedance.
There is. Best practice is that the input impedance should be seven to ten times the source impedance. A mic input is typically around fifteen hundred ohms. If your source is a guitar pickup at twenty to forty kiloohms, you're way outside that ratio, and what you get is a rolloff at both ends. Thin and dull. That's the classic problem a DI box exists to solve, because it presents a high impedance to the instrument and a low impedance to the mixer.
So for Daniel's speaker, the XLR input is the balanced path, which is what you want for a long run, but it's the wrong level.
Both things are true at once. The XLR input gives you common-mode rejection and it's the right electrical interface for distance. It's just expecting a signal a thousand times smaller than what you're sending.
So what's the actual recommendation?
If you're going more than a few meters, run a balanced XLR line and put a DI or an attenuator at the speaker end. If you're staying short, the RCA or the 3.5mm is fine, you're not going to hear any treble loss over three meters, and you avoid the level mismatch entirely.
The "right" answer depends on which failure you're more worried about.
Short unbalanced run, your risk is the environment, so keep it away from power cables and lighting rigs. Long balanced run through a DI, your risk is the source impedance and the current demand, so check what's driving it. Different problem, different solution.
Which gets us to the meta-question Daniel actually asked, which is why engineers don't have intuition here.
Because the guidance is environment-dependent and it's repeated without the math attached to it. Somebody reads "keep RCA under ten feet," repeats it, and it becomes folklore. Meanwhile the same person ran a fifty-foot RCA from a CD player to an amp for fifteen years and never noticed a problem, because a modern CD player has a low output impedance and there was nothing in the room radiating interference.
The folklore and the experience don't match, so neither one builds intuition.
Right. EAW's line on this is the most honest thing in the whole literature. "There are no shortcuts or rules of thumb. You need to do the math." And then, "Foolproof? Plug-and-play? Definitely not."
Which is a strange thing for a company selling audio equipment to write.
It's an unusually candid document. And there's a good reason it reads that way. The engineering answer is "it depends on five things," and the marketing answer wants to be "run it a hundred feet, it's fine." Those two positions can't both survive.
Daniel's instinct that this is confusing across subject areas is correct. It's not that the audio world doesn't know the answer. It's that the answer isn't a number.
Hilbert: I agree with all of that.
Good.
Hilbert: I had a boss at a small touring outfit who would not use an RCA cable longer than ten feet. He had a whole theory about it. Said anything longer would suck the life out of the highs. So what he did instead was chain four or five ten-foot cables together with barrel connectors whenever he needed to reach the PA. Which is thirty, forty feet of the same cable, just with more connections in it.
Did you tell him?
Hilbert: I was nineteen and he owned the van. I did what I was told and I coiled the cables at the end of the night.
The physics we just described never got tested, because he found a way to obey the rule and break it at the same time.
Hilbert: He was not a stupid man. He was a careful one. Someone had told him ten feet and he never had a reason to ask why.
What actually went wrong on those gigs?
Hilbert: Hum, mostly. We played a lot of places with bad power. One night the whole rig picked up a local AM station during soundcheck. Very faint, but you could hear a voice in the mains. We spent an hour swapping cables before we figured out it was the long run from the mixing desk. The shield was acting like an antenna and feeding the radio straight into the audio path.
Which is the mechanism the physics section underweights. Capacitance tells you about treble. It says nothing about what the cable picks up along the way.
Hilbert: The ten-foot rule isn't really about the treble. It's a rule about not having a long piece of wire lying around near lights and power amps.
That reframes the whole thing. The number everyone quotes is a horse-trading approximation of an environment risk, not a filter calculation.
Hilbert: We switched to a longer run with a transformer on it after that. Not because the treble improved. Because the radio went away.
The limit for unbalanced cable is often set by the room you're in, not by the wire.
Hilbert: You can have a hundred-foot run that's dead silent and a fifteen-foot one that hums. Depends what it's lying next to.
Which is exactly why the tables don't work. The number that matters is the one on the page, and the page is wrong half the time because it can't see the room.
Hilbert: I've still got the barrel connectors in a box somewhere. I don't know why I kept those.
If the limit is system-dependent, how's a listener supposed to evaluate their own setup? Nobody's going to sit down and measure their source output impedance on a Sunday afternoon.
Start with the source. If it's a modern phone, laptop, or a dedicated audio interface, the output impedance is probably low enough that treble loss over a reasonable run is a non-issue. Then ask what's near the cable. If it's running alongside mains wiring or a lighting dimmer, you've got a noise problem, not a capacitance problem, and the fix is routing or a balanced connection, not shorter cable.
If you can't move the cable?
Then you buy a DI. Which is cheaper than replacing a hundred feet of cable you've already run.
There's a version of this where the answer becomes a niche skill, though. Phones with headphone jacks are disappearing. Everything's going wireless or digital or over a network. Does any of this survive?
The physics doesn't care about fashion. Capacitance is capacitance. But the need to know it drops. If your speaker takes a Bluetooth or a network connection, the analog run is zero feet and the whole question evaporates.
Which means the people who still need to know this are the ones doing high-fidelity work, live sound, installs. The people who care about the last half a decibel at twenty kilohertz.
Those people are exactly the people who were never going to trust a rule of thumb in the first place.
Which is Hilbert's point, in a way. The mentor who chained ten-foot cables together wasn't wrong about wanting short runs. He was wrong about why.
He had the right instinct and the wrong model. Which is most of audio.
If you got something out of this one, a review wherever you listen helps other people find the show. Our producer is Hilbert Flumingtop. This has been My Weird Prompts, the human-AI collaboration podcast.
Email us at show at my weird prompts dot com. We'll be back soon.
