When 12 Gauge Wire is not 12 Gauge!

Apr 3, 2010
16,022
0
0
Seattle, WA
#1
I know a lot of members here use premium speaker wires. But I thought I share this testing I did on common low cost 12 gauge wires. As always, comments and questions are welcome.

Introduction
Why test 12 gauge (AWG) wire? 12 AWG speaker wire is a “safe bet” from performance point of view because anything thinner may interact with the low impedance of your speakers and cause the frequency response to vary beyond threshold of hearing (-0.5 dB). That change can “color” the sound.

Once you get to 12 AWG and in reasonable (shorter) lengths, you should be good. This conclusion however only holds if the wire you buy is actually 12 gauge wire and has the nominal resistance that is used in the computation of dB drop. For this reason, it is useful to see if the wires that one can readily buy online or from local sources in US complies with the nominal values for 12 AWG. The measurement in question is “DC resistance” where we measure the resistance of the wire when it is being fed direct current (DC). This is the most basic parameter for cables.

I plan to keep updating this data. So if you like your favorite speaker wire measured, PM me for address and be ready to mail a 4 foot/1.5 meter section of your wire and I will measure and add its value to the measurement table below.

Test Methodology
The purpose of a good speaker cable is to transfer energy with very little drop. That very characteristic makes it very hard to measure the resistance by definition, that is a very small value. Typical (DC) resistance of speaker wire is in the area of 0.0015 ohm/1.5 milliohm per foot. This small resistance puts typical multimeters out of business since they become inaccurate in single digit ohm let alone in thousands of an ohm.

There are different solutions to this problem. The one I opted for is the so called 4-wire or Kelvin measurement invented by Lord Kelvin 100+ years ago. At high level, the 4-wire system separates the leads that provide power to the load (i.e. our speaker wire) from the leads that measure the voltage drop across it (which given the current, tells us the resistance using Ohm’s law). Because the leads are separated, we are now free to provide much more current and as a result, create a larger voltage drop. Not only that, but we also eliminate the effect of meter probe because the current that is going through them is a fraction of what we are feeding the load. Please look online if you like to have more details about its operation.

Unfortunately high-end milliohm meters are quite expensive, some going for as much as $5,000. I don’t do enough of this work to justify investing in them. The unit I have used here is a portable unit which has resolution down to 0.1 milliohm and accuracy of 1% (plus 5d). Its output current is rather low at 200 milliamps since it runs on batteries. That however, is still 200+ times more than standard multimeters which use 1 microamp to 1 milliamp typically for resistance measurements.

In my test fixture, I am using a common, 2-probe system. There are still 4 leads going to the load but they attach to the load in pairs. This removes one of the sources of mistakes (using the wrong probes for high current and voltage measurements) and makes it much faster to test multiple items. Accuracy is still quite high.

As is always the case, the reality and theory are different. A milliohm meter is a very sensitive device. This means that it will actually measure the contact resistance of its own probes. This is easy to back out however by zeroing out its lead resistance first which is what I did. What is not so easy is to guarantee that you put the same contact pressure on the wires in question. There is an easy solution to this which is to use a much longer length of cable and hence, have its resistance swamp the connection load. Problem with that is the wire will coil every which way and won’t match from sample to sample. I wanted a predictable setup where every wire was tested the same way which meant straight and flat.

My solution to the problem was to use a ~3 foot segment of wire that I could hold flat on my desk but then short out one end and measure the resistance as seen from the other end. This does create a new problem in that the twisted end again has certain resistance and variability. To counter that, I put “clamping” load on it in the form of a beefy paper clip. I tested that fix by pushing hard on the connection while the clip was holding it and the difference was negligible. Without that clip, there would be considerable change when I put force on it.

Here is what the final fixture and my test setup looks like:



Sample Wires Tested:
Here are the samples that I managed to acquire during a two week or so period:

Monoprice 12 AWG Speaker Wire: I bought a 50 foot spool through Amazon third-party service. I paid $25.35 and shipping was “free” (Prime). Monoprice’s own price is lower but you have to pay shipping and I prefer to not create accounts online any more than I have to.



The cable itself has a blue stripe on one of the wires which is useful in identifying which wire is which. The reel was cardboard and the overall impression screamed budget/low-end. The stripped wire did not hold well together due to many soft strands. Stripping it resulted in loosing fair number of strands.

RadioShack 12 AWG “AUVIO” speaker wire: I bought a 50 foot spool this on sale for $39. With tax it came up to $43 or so. Since I picked it up locally there was no shipping.



This is one good looking cable and spool! It oozes quality. The spool is blue and substantial. Likewise the wire looks thick and beautifully wound. I weighed the spool and wire and it was 4.51 pounds. In comparison, the Monoprice was 2.68 pounds. If I put the two next to each other and put the price tag on them, I am pretty sure most people would go for the RadioShack wire. The visual difference is unmistakable.

Parts Express Wired Home SKRL-12-50: I bought a 50 foot spool again through Amazon for $24.20. Shipping was an additional $7.69 for a total of $31.89. I went to their site and it was similar in price with shipping so I bought it from Amazon. It took over 7 days to get this wire. They shipped it quickly but they used economy service to send it to me. Being spoiled by free Prime shipping from Amazon, it was quite annoying to pay nearly $8 and have to wait a week.



As the listing indicates, this is from a company called Wired Home. It came in a nice blue plastic spool. It was not nearly as substantial as the RadioShack wire but definitely a step above lower end stuff.

Belden 5000UP 12 AWG: This is an in-wall speaker cable. As one of the largest cable suppliers in the world, and set of measured specifications, I thought this would provide a nice baseline to compare others. I could not find 50 foot spool of this wire on Amazon. All that was available through third-parties was 100+. Parts Express sold it however by foot though so I ordered 20 feet. The cost for that was $19.60 and shipping was $14.00 for express delivery.



The outer wrapping in this cable is thick and substantial. Not to the level of RadioShack wire but still above average. The individual wires inside strip easily and hold their form strongly. It is the closest thing to electrical wire.

Fry’s 12 AWG Wire: It was hard finding this wire at Fry’s Electronic as it was not with the rest of the speaker wires in the AV department. This is what it looks like:



Price was a reasonable $15.99 for the 50 foot spool. The spool is very light and the wire pretty flexible.

Electrical Wire: This is your typical stranded 12 gauge electrical wire that I had bought from Home Depot. It is a single conductor wire so not very suitable as speaker wire. But I thought I include it as a reference since I had it in my drawer of electrical parts. I don’t have the label handy but it is similar to this:



It strips easy because it matches the gradations in the a typical wire stripper but is very stiff.

Colman in-wall 12 AWG: I have a few hundred feet of this in my house. My then contractor (before I started Madrona Digital) selected it without my involvement. I told the contractor to pick “good quality cable” and this is what he bought. The application is non-critical (background music in the kitchen and feeding power to other devices). This is what it kind of looks like:



As you can see, it is a typical in-wall (CL3) cable with outer insulation and Belden like inside wires.

ICE 12 AWG Speaker Cable: ICE is one of the “go to” brands of cables for custom AV installers. We use a number of different speaker cables at Madrona and I found this left over reel in the shop and thought I should test it:



BestBuy 12 AWG Cable: As with Fry’s, the speaker area had a bunch of wires from Monster and their own house brand but nothing that went up to 12 AWG. I remembered that the automotive section often has heavier gauge wires and that was the case. They had a non-descript 20 foot spool. One conductor is copper colored and the other “silver.” I suspect it is actually aluminum wire.

Canare 4s11: This is a premium in-wall cable. It has four 14 gauge conductors. You can use two of the 14 gauge wires together if you only need one speaker feed which is the way I tested it. Alternatively you can use it as redundancy in case during construction a nail or screw went into it.



Coat Hanger: No, you don’t new glasses; I did say coat hanger! :D There is an online fish story that says someone performed blind testing of coat hanger against monster cable and nobody could tell the difference. There are other issues with that story but here, I thought I focus on the DC resistance.



The specimen I used has no brand or label. It is awfully thick though and was very hard to unwind into a straight “wire.” To combat contact resistance, I zeroed my meter by putting the probes next to each other and using that as the new “zero.” In a real situation that contact resistance would also be part of the equation.

Measurements
OK, enough rambling; let’s get into the measurements as shown in the table below. The first column is the length of the wire I was testing. I was not anal about keeping the length exactly 3 foot. So instead, I measured the actual segment and used that in the computation. In some cases I had a fixed length already and I used that.

The second column is our key data, the measured resistance in milliohms. Since this would vary based on the length of the wire being tested, I divided its value by the length and arrived at the industry standard milliohms/ft.

Next is the claimed DCR if available. Yes, there are discrepancies between my measurement and theirs. Since these are stranded wires, it is hard to get the exact number the resistance is supposed to be. Likely there are differences between my fixture and the one cable manufacturer used. So the best use of the measurements is as a relative value to compare one wire against another, rather than attempting to match it to any published spec. To that end, I used the measured DC Resistance of Belden cable as the baseline and used that to create a ratio in the next column (“Ratio to Standard”).

The Relative Difference column takes out the value of the Belden cable giving us a “pure” percentage of how much higher or lower the DCR is relative to Belden. In that regard, Belden gets a reference of 0. Negative numbers now mean a wire has higher resistance than Belden and positive numbers the other way around.

Weight KG indicated the weight of the sample wire I tested in kilograms. It is in metric because my scale seemed to give better accuracy there. I then normalized that per industry standards to weight/1000 meters.

Next is the thickness of the conductors. This is hard to measure as some of the stranded wires flatten when you try to squeeze them with micrometer. But I did my best anyway to give you a sense of how thick the conductors are.

While this report is mostly focused on DC resistance, I thought I include a couple of AC parameters such as Capacitance and Inductance per foot. These are not a factor in short speaker runs in the room but can be if you have long runs behind the walls.

The bottom row in green is the Geometric Mean (Geomean) of the column of data above it. Geomean is an average of a set of numbers that doesn’t get thrown off badly by one or more samples being way off the scale. Since that is what I am dealing with here, it makes for a better value than simple average. The number then provides a statistic average of the samples I tested.



And the Relative Difference charted as bar graphs:



I have color coded the underperforming wires in orange. As you can see, Monoprice, Fry’s, Bestbuy and Coat Hanger fall in this bucket. The coat hanger actually went past the bottom of the graph so whatever story there is on how it sounds is quite suspect.

Fry’s and Bestbuy wires must be aluminum cored to have such high resistance. They are not thin enough for the difference to be due to that. I would certainly avoid using both in any high fidelity application.

Of the none-in-wall wires, the RadioShack by far leads the pack on both subjective quality and measured DC Resistance. It managed to slightly outperformed our Belden reference. At $40, that is not much of a premium cost wise considering that you can pick it up in person and be able to instantly use it.

The Monoprice’s resistance is almost twice as high as RadioShack wire but not nearly as bad as the BestBuy and Fry’s no name wires. But being least bad doesn’t translate to good in my book :). My recommendation is that if you want to go the mail order route, go with the Parts Express Wired Home cable. It outperforms Monoprice both subjectively and in measured resistance (50% lower than Monoprice).

Conclusions
So there it is. Clearly 12 AWG wire is not 12 AWG when you buy a no-name brands. The notion then that you should buy any old wire that says 12 AWG and shopping purely based on price is not a wise one. Where you can, buy branded cable that comes with proper specification.
 
Dec 12, 2013
935
0
16
NYC , USA
#4
Nice post a very comprehensive. It is amazing how companies make claims that are just false . Great read

Al
 
Oct 1, 2010
910
0
16
Cleveland Ohio
#5
May 30, 2010
13,938
19
38
Portugal
#6
Amir,

Thanks for these interesting measurements. However the use of the Belden wire as a reference introduced a bias in the data analysis and perception of the results. IMHO, the values should be compared with the typical copper resistance values used by reference sites and manufacturers for 12 AWG copper wire 5.211mOhm/m or 1.588mOhm /foot - a large difference from your reference 1.15mOhm/foot.

Also your choice of the star quad Canare 4s11 (AWG14 as you say) is not very clear - in order to duplicate the area section we should decrease 3 points - a double run of AWG 15 is equivalent to AWG12.
 
Apr 3, 2010
16,022
0
0
Seattle, WA
#8
Amir,

Thanks for these interesting measurements. However the use of the Belden wire as a reference introduced a bias in the data analysis and perception of the results. IMHO, the values should be compared with the typical copper resistance values used by reference sites and manufacturers for 12 AWG copper wire 5.211mOhm/m or 1.588mOhm /foot - a large difference from your reference 1.15mOhm/foot.
There is no standard for 12 AWG stranded wire. Depending on the number of strands and their dimensions you get different values. Also, my fixture is not the same as the one manufacturer used so I can't use Belden's data or any other manufacturer. The best way to use the data is comparative. You can ignore the normalized reference to Belden and still look at how others compare with each other.

Also your choice of the star quad Canare 4s11 (AWG14 as you say) is not very clear - in order to duplicate the area section we should decrease 3 points - a double run of AWG 15 is equivalent to AWG12.
The choice was determined by the wires I had on hand. The idea was to test a large sampling of wires so that we can get a decent overall view of the cables as a whole. I bought all the cables I could and augmented it with these others.
 
Apr 3, 2010
16,022
0
0
Seattle, WA
#10
And thanks for the kind words everyone :). This was one of those projects which seemed too trivial to do but it sort of got a life of its own and became something of substance! I have already one offer for a cable to test and have a Monster and Transparent speaker cable to add to the table.
 
May 30, 2010
13,938
19
38
Portugal
#11
There is no standard for 12 AWG stranded wire. Depending on the number of strands and their dimensions you get different values. Also, my fixture is not the same as the one manufacturer used so I can't use Belden's data or any other manufacturer. The best way to use the data is comparative. You can ignore the normalized reference to Belden and still look at how others compare with each other.


The choice was determined by the wires I had on hand. The idea was to test a large sampling of wires so that we can get a decent overall view of the cables as a whole. I bought all the cables I could and augmented it with these others.
Amir,

Happily there are standards for stranded wire. From Wikipedia - The AWG of a stranded wire is determined by the total cross-sectional area of the conductor, which determines its current-carrying capacity and electrical resistance. Because there are also small gaps between the strands, a stranded wire will always have a slightly larger overall diameter than a solid wire with the same AWG.

IMHO, an objective study of AWG wires should compare exactly the crosssection areas of the cables - it is what you paying and is being advertised. As you have taken the care to weight the wires it would be nice to show the ratio between weight and measured resistance - it is a good indicator for the material type, although we are now in obscure zones - material wire should not be relevant to sound quality per se.:) BTW, the resistivity of aluminum is only 70% larger than copper - it can not explain alone the poor performance of the cheap cables.
 
Jul 1, 2010
8,713
0
0
#12
Excellent work Amir. My approach, back in the days when wire was an issue for me, was buy Belden and be done with it. I think your choice of Belden as the benchmark makes all the sense in the world.

Tim
 
Oct 1, 2010
910
0
16
Cleveland Ohio
#13
In my old 1981 Alpha Wire and 1985 Belden Cable catalogs, they both have charts for stranded wire. The Ohms per 1000 feet varies depending on stranding (from 1.45, 1.58, 1.70 to 1.75). Newer charts have all standings at 1.71.
 
Jan 13, 2012
429
0
0
Oslo, Norway
#14
Would using AC change your measurements, given that it's AC-current from amplifier to speakers, and not DC? (Just for my edification).

And while I would never attempt to use coat-hangers myself, I do remember that gizmodo-article, and that they twisted and soldered four coat hangers to create a speaker cable. Maybe they realized they would need more than one.

EDIT: Found the article, I remembered right.

In the other, four coat hangers twisted and soldered into a speaker cable.
 

ddk

Industry Expert
May 19, 2013
3,447
6
38
Utah
#16
Amir, Do you think that flat woven wires of the same gauge and brand would behave differently than their twisted counterparts? I have some to send you if you like.

david
 
Apr 3, 2010
16,022
0
0
Seattle, WA
#17
Amir,

Happily there are standards for stranded wire. From Wikipedia - The AWG of a stranded wire is determined by the total cross-sectional area of the conductor, which determines its current-carrying capacity and electrical resistance. Because there are also small gaps between the strands, a stranded wire will always have a slightly larger overall diameter than a solid wire with the same AWG.
I have read all of that and many other sites. What you quote is not a "standard." It simply says that there can be variations which there are. Here are two references on stranded wire gauges. The first one is from Calmont wire and the other from George Washington University:



DC Resistance is the last two columns:



As you see, there is wide variation here and no one value has been anointed as the 'standard.'

IMHO, an objective study of AWG wires should compare exactly the crosssection areas of the cables - it is what you paying and is being advertised. As you have taken the care to weight the wires it would be nice to show the ratio between weight and measured resistance - it is a good indicator for the material type, although we are now in obscure zones - material wire should not be relevant to sound quality per se.:) BTW, the resistivity of aluminum is only 70% larger than copper - it can not explain alone the poor performance of the cheap cables.
First, this is not a "study." It is a set of ad-hoc tests. Second, the crossection cannot be measured accurately with the tools that I have. My micrometer compresses the fibers and different wires compress differently than each other depending on the make up of the strands. That data by the way was in the table as best as I could measure. Note that the weight included the jacket/insulation. I originally intended to strip the wires and measure just the conductor but it was too tedious and for now, I thought the data as is would be helpful to some.

And yes, being aluminum does not fully explain the high resistance of some wires. The purpose of my testing was not to explain the "why" but the "what."

I am open to someone else creating a test fixture that is better and report their results. I have yet to see an analysis of such values from multiple manufacturers. So it would be a useful endeavor. From my point of view, I like to include more wires in the table so that we get a good feel of their relative characteristics which I think is fairly solid.
 
May 30, 2010
13,938
19
38
Portugal
#18
I have read all of that and many other sites. What you quote is not a "standard." It simply says that there can be variations which there are. Here are two references on stranded wire gauges. The first one is from Calmont wire and the other from George Washington University:



DC Resistance is the last two columns:



As you see, there is wide variation here and no one value has been anointed as the 'standard.'


First, this is not a "study." It is a set of ad-hoc tests. Second, the crossection cannot be measured accurately with the tools that I have. My micrometer compresses the fibers and different wires compress differently than each other depending on the make up of the strands. That data by the way was in the table as best as I could measure. Note that the weight included the jacket/insulation. I originally intended to strip the wires and measure just the conductor but it was too tedious and for now, I thought the data as is would be helpful to some.

And yes, being aluminum does not fully explain the high resistance of some wires. The purpose of my testing was not to explain the "why" but the "what."

I am open to someone else creating a test fixture that is better and report their results. I have yet to see an analysis of such values from multiple manufacturers. So it would be a useful endeavor. From my point of view, I like to include more wires in the table so that we get a good feel of their relative characteristics which I think is fairly solid.
Thanks.

Your quoted tables from Calmont wire and the other from George Washington University only support my position - all the resistance values for AWG12 wire per foot are between 1.46 and 1.75 mOhm. You used the value of 1.15 mOhm, in disagreement with the value from the Belden datasheet of around 1.559 mOhm, creating an error that propagates through the whole table and IMHO distorts the global view about other cables adherence to the AWG12 quoted value.

Please note that I am not questioning your choice of Belden or debating semantics - just the measured reference value, that is so far from the theoretical value of the one strand copper wire. Although measuring low resistances absolute values is not easy, measuring many values with relative accuracy is an easy job - you just connect all of them in serial, flow a stable current through them and measure the difference of potential between the extremities of the wires, avoiding the connecting points. Long ago I have done it measuring very low value resistances with a precision of one per thousand ( .1 %) in a project of a current source for a tungsten calibration lamp. If I had all the wires I would be very happy to participate in your project, but considering distance and customs it would not be an easy task.
 

CGabriel

Industry Expert
Nov 1, 2013
587
0
16
WA, USA
www.shunyata.com
#19
Very good work Amir. It is good to see a good logical thought process. Microstrip has some very good points and suggestions. First, let me be clear - I am not criticizing but just contributing to what you are trying to do.

As many of you know, we at Shunyata Research do many tests including wire tests in the development of products. I will confine my comments to the specific tests that Amir has conducted. We use a Hioki 35-11/50 LCR tester for these tests. In wire testing it is important to test long lengths of wire due to the very small resistance of the wire and the technical difficulties for measurement equipment to deal with short lengths of wire. I used a 50 foot length of 12 gauge stranded wire for a reference reading.

The DC resistance of the 50ft 12AWG wire works out to 1.66mOhm/ft

With the same tester I measured the same wire at three different lengths: 6ft, 5ft and 3 ft with the following readings

6ft = 1.48mOhm/ft
5ft = 1.38mOhm/ft
3ft = 1.67mOhm/ft

There is a variance of about 20% across these readings which simply indicates that the test unit has difficulty giving an accurate result with short lengths of wire. So my suggestion would be to go back and do the test again with significantly longer lengths of wire.

Next, I know that there are some people that claim that the only significant factor in speaker cables is the DC resistance. Here is something to think about. The inductance of the 50 foot length of 12 gauge wire is .274uH per foot. The measured impedance (which is a combined DC, L and C) of the wire is 2.36mOhm/ft while the DC resistance is 1.665mOhm/ft. This means that the impedance of the wire is about 40% greater than the DC resistance. And this is without any effects from geometry. This is the reading from a single wire - end to end.

Regarding geometry: Let's look at the effects of simple L and C in relation to a signal wire and a return which is what's required to send a signal to a speaker. We will measure a + wire and a - wire that are separated in space. Then we will measure the same set with the wires directly parallel to one another. And then the last will be the wires with a standard simple twist geometry. These readings are for a 6 foot set of leads and they are actual readings.

Wire separated: Z=.0342 Ohms
Wire parallel: Z=.0213 Ohms
Wire twisted: Z=.0213 Ohms

The measured DC resistance of the 6 foot cable is .0190 Ohms. So, the impedance of the wire at 1Khz is about 80% greater than the DC resistance when the wires as run separately. There is 12% greater impedance when the signal and return geometry is closely coupled electromagnetically as in the parallel and twisted measurements.

Of course reactance varies according to frequency so the difference in DC readings and total impedance will rise as the frequency rises.
 
Oct 1, 2010
910
0
16
Cleveland Ohio
#20
Those above charts have the same values as in my post #14. The information in those charts came from the 'National Bureau of Standards Copper Wire Tables - Handbook 100". Some time between 1985 and the present the wire industry stopped using different values for each stranding and now all strandings have a value of 1.71 Ohm. A couple of years ago, I was enlightened to this in another thread in another forum.