A recent speaker cable test I conducted for Geoff Morrison at The Wirecutter got me thinking once again about this highly contentious subject -- just when I thought I'd banished it from my mind forever.
Read the follow-up to this article here.
Later, I mentioned to Allan Devantier, manager of acoustic research at Harman International (makers of Harman Kardon receivers, JBL and Infinity speakers, and numerous other audio brands), what we were up to. He and I got into an in-depth discussion of whether it'd be possible to demonstrate from a technical standpoint that, at least in fairly extreme circumstances, speaker cables can make a detectable difference in the sound of your system.
First, a disclaimer: I don't have a strong opinion about speaker cables. I've done blind tests for Home Theater magazine in which panelists developed consistent preferences for certain cables over others. Yet I rarely concern myself with it.
I've felt dismayed by both sides of the argument: the belligerent insistence in some audio publications (notably Stereo Review, which has now morphed into Sound & Vision, and the barely alive The Audio Critic) that speaker cables make no difference, and on the other side, some high-end audio reviewers' long-winded, elaborate, effusive descriptions of the differences in the "sound" of speaker cables. It seems to me that both sides are defending entrenched positions rather than engaging in an honest, open-minded effort to seek the truth.
Just in case you're wondering what I use personally: some pro speaker cables made by Canare, some generic in-wall 14-gauge, four-conductor cables for longer runs, and a few other cables I have sitting around. I should add that in more than 20 years of speaker reviewing, testing speakers from <$50/pair to >$20,000/pair, I've only ever had one manufacturer express concern about what speaker cables I was using.
What got Devantier interested was when I began talking about how a speaker cable might, in theory, change the frequency response of a speaker. Every speaker is basically an electrical filter -- a combination of resistance, capacitance and inductance tuned (one hopes) to deliver the best possible sound quality. If you add extra resistance, capacitance or inductance, you change the filter values and, thus, the sound of the speaker.
Normal speaker cable doesn't have significant capacitance or inductance, but its resistance does vary somewhat -- especially with thinner cables, because all other things being equal, the thinner the wire the greater the resistance.
I'll let Devantier take it from here. He begins by citing research by Floyd Toole and Sean Olive, his respective former and current colleagues at Harman, who were at the time working at Canada's National Research Council:
"In 1986 Floyd Toole and Sean Olive published research on the audibility of resonances. They found that listeners are particularly sensitive to low-Q [high-bandwidth] resonances. Midrange peaks of just 0.3 decibels (dB) were audible under the correct conditions. Since loudspeaker impedance varies with frequency, the DC resistance of the cable becomes very important. The following chart shows the maximum allowable cable length to ensure that amplitude response variations caused by cable resistance are kept below 0.3 dB. This chart assumes a minimum speaker impedance of 4 ohms and a maximum speaker impedance of 40 ohms and that cable resistance is the only factor; it does not include inductance and capacitance, which can only make things less predictable.
"It should be clear from this table that under some circumstances the cable and the loudspeaker can interact to cause audible resonance."
cable resistance length
gauge ohms/foot for 0.3 dB ripple
(AWG) (both conductors) (feet)
12 0.0032 47.23
14 0.0051 29.70
16 0.0080 18.68
18 0.0128 11.75
20 0.0203 7.39
22 0.0323 4.65
24 0.0513 2.92
"You know, you could measure this," Allan said to me, pointing his finger in a way that implied a command more than a suggestion.
I've been doing frequency response measurements of speakers since 1997, but I've always just used a nice, big, fat speaker cable to connect the speaker under test to the amp -- something I knew wouldn't affect the accuracy of the measurement. But what if I substituted a crummy, cheap little generic speaker cable? Would I be able to measure a difference? And would it be the kind of difference that would be audible?
1) the 12-ga Linn cable I've been using for speaker measurements for the last five years or so
2) a cheap 12-ga Monoprice cable
3) a cheap 24-ga RCA cable
To minimize environmental noise, I ran the measurements indoors. Neither the microphone nor the speaker nor anything else in the room was moved. I used an extra-long FireWire cable so the computer and I could both be out of the room entirely. I also repeated each measurement a few times so I could be sure environmental noise wasn't appreciably affecting the measurement. Why so careful? Because I knew I'd be measuring subtle differences -- if I could measure anything at all.
I then took the response with the Linn cable and divided it by the response of the Monoprice and RCA cables. This gave me a graph showing the differences in the frequency response caused by the cables. I then applied 1/3-octave smoothing to help ensure no residual environment noise snuck through.
Turns out Devantier was right -- I could measure this. As you can see in the chart, the results with the two 12-ga cables were only subtly different. The biggest change was a boost of maximum +0.4 dB between 4.3 and 6.8 kHz.
Is this audible? Maybe. Would you care? Probably not. To put it in perspective, that's about 20 to 30 percent of the change I typically measure when I test a speaker with and without its grille.
But switching to the 24-ga cable had a huge effect. For starters, it reduced the level, requiring me to normalize the measured response curve by boosting it +2.04 dB so I could compare it with the curve from the Linn cable. The 24-ga cable's resistance also had obvious effects on frequency response. For example, it cut bass between 50 and 230 Hz by a maximum -1.5 dB at 95 Hz, cut midrange between 2.2 and 4.7 kHz by a maximum -1.7 dB at 3.1 kHz, and reduced treble between 6 and 20 kHz by a maximum of -1.4 dB at 13.3 kHz.
Is this audible? Yeah. Would you care? Yeah. Would you like the sound better with the skinny cable or one of the fat ones? I don't know. Regardless, our past recommendation to use 12- or 14-ga cables is looking pretty wise.
This is a fairly extreme example. While there might be a few exotic high-resistance speaker cables out there, almost all speaker cable of at least 14 ga or so has low-enough resistance that any sonic anomalies it introduces should be at least minimal, and probably inaudible. But it's important to note that I measured slight and repeatable response differences even with two cables close in size and structure. Also, note that the Revel F208 has an average impedance of 5 ohms (as measured by me). The effects would be more pronounced with a 4-ohm speaker, and less pronounced with 8-ohm speakers, which are by far the most common.
So what's the lesson to take away from this? Mainly, don't use skinny cables in any system where you care about the sound quality. Also, maybe don't be so quick to judge those who say they hear differences among speaker cables. Sure, many of them are obviously exaggerating these effects, and the ads from high-end cable companies often grossly exaggerate these effects. But the calculations and experiments we did here suggest it's possible they really are hearing something.
Need more info on buying stereo systems? Check out our Basic Guide.