Measurements of speaker cables in frequency and time domain

[preload]

More measurements

I have just measured the effect of 2 speaker cables on frequency response, with my speaker dummy load. The dummy load simulates behaviour of a 2-way closed box speaker. It should be understood that the effect of a speaker cable on frequency response depends on complex impedance of the speaker (load) as was explained, with formulas, in post #1.

This is the dummy load impedance
View attachment 126994
We can see that the impedance is capacitive at high frequencies and this will have interesting consequences.

Two speakers cables were tested:
1) a flat cable, length 2m
2) a zip-cord, cross section 2x1.5mm2, length 5m

The flat cable looks like this:
View attachment 126995
It has bigger distance between the wires (than a zip-cord) and this increases inductance and reduces capacitance of the cable.

Again, a voltage transfer ratio of voltages behind the cable / before the cable as a function of frequency was measured, which shows the frequency response of the cable in isolation, with the load used.

This is the result:
View attachment 126996
Please see the rise of frequency response above 5kHz due to capacitive character of the load at high frequencies. Also see the 0.2dB dip in zipcord FR at 150Hz due to high resistance of the 2x1.5mm2 zipcord with 5m length.

Conclusion
It is impossible to say that "cables make no difference". It is not true. The cables depending on length, construction and speaker used may make an audible difference, even if they are as short as 2m. 5m of 2x1.5mm2 zipcord makes 0.2dB deviation at higher bass, into quite standard speaker load. This starts to be audible. The speaker cable should be as short as possible and monoblocks placed near the speaker are the best option in case of passive speakers.

Edit: the zipcord used was 2x1.5mm2, length 5m.

This all looks correct to me and supports what I’ve also pointed out on several occasions on the variation in FR caused by speaker cable inductance (and DCR) as a function of the speaker’s impedance vs freq curve.
The only thing I would question here is the explanation for why there is a relative rise >5khz in then cable with higher capacitance. Rather I would still attribute this to the differences in cable inductance and DCR. The nF of capacitance in the cables you tested, predicted based on the cable geometry, shouldn’t be great enough to start affecting HF in the audible range. I’m happy to be corrected, but this is my understanding.

 

[Ashley Salmond]

Hi, this to a layman looks like me a comprehensive test. What would make a I treating test is try same overall volume awg cable of twisted and weave. Then it might be possible to access their differences and to see if there is any advantage to using a type of cable. I am no expert, but my thought is that depending on the impedance of a speaker and use g the same amplifier there may be some increase in frequency responce or possibly decrease . Either way, it would certainly inform the audio consumer of his options as to a systems performance.

 

[preload]

Hi, this to a layman looks like me a comprehensive test. What would make a I treating test is try same overall volume awg cable of twisted and weave. Then it might be possible to access their differences and to see if there is any advantage to using a type of cable. I am no expert, but my thought is that depending on the impedance of a speaker and use g the same amplifier there may be some increase in frequency responce or possibly decrease . Either way, it would certainly inform the audio consumer of his options as to a systems performance.

If done correctly, the twisted weave cables tend to have lower inductance and higher capacitance compared to single pairs. As long as the equivalent gauge is the same (i.e. DCR is similar), the net result should be LESS deviation in FR. There's an older aes paper that compares cable LCR characterisrics and I think the twisted weave design was included.

 

[pma]

I also made a simulation of FR of the 5m zipcord loaded with a dummy load as of the post #14. The simulation gave almost same plot as the measurement:

which proves that the measurements are correct.

I would like to mention that the speaker cable, though often underestimated and laughed, may have much bigger impact to the sound than a change of average DAC or preamplifier. This observation might be interesting in the light of attention that is paid to preamp and DAC tests and to tiny differences in the SINAD. The speaker cable in fact is a kind of EQ with respect to the speaker used and its influence might be eliminated by using software EQ.

 

[Matias]

@Amir could you also test speaker cables into dummy loads that simulate actual speakers?

 

[RayDunzl]

I would like to mention that the speaker cable, though often underestimated and laughed, may have much bigger impact to the sound

Excellent thought.

It's boring to listen to dummy loads.

How about a test of these electrical microvariations as shown in the response of a speaker in a room taken at the listening position?

 

[oupee]

Amir does a great job, but you could practically say that his measurements are incomplete. Square wave measurement is missing in almost all types of products. Great @pma

 

[solderdude]

I have only low electronic skills, please can you explain why a resistor with 5 times higher value parallel to a loudspeaker can do any termination (4/5 of the electric power is going thru the speakercoil) ?

Most speakers have substantially increasing impedance for inaudible high frequencies. Given the usage of square-waves (harmonics are very high up) so PMA used a fixed resistor of a high enough value. PMA states that this works better than a Zobel.
When the impedance of a speaker near the resonance point comes close to 40 ohm the 47 ohm will have an influence.
PMA's dummy load reaches 36 ohm.

I would not use a high power squarewave test using speakers though. The huge amount of harmonics will fry the tweeter. This also means that one cannot test amps with actual loads at higher power levels. The dummy load is essential here.

 

[amirm]

@Amir could you also test speaker cables into dummy loads that simulate actual speakers?

I can but I ran the very test on a real speaker in my last review just as PMA is doing here: https://www.audiosciencereview.com/forum/index.php?threads/morrow-sp3-review-speaker-cable.22834/

I used a Klipsch speaker as the load. Here are the results:

And I showed the difference between the two cables:

I am using proper scale so you get a sense of how small these numbers are (one thousands of a dB). Note that my Audio Precision analyzer is far more accurate than OP's scope measurements. Digital scopes are made for speed so their dynamic range is quite poor (usually 8 to 10 bit digitizers).

 

[RayDunzl]

Amir does a great job, but you could practically say that his measurements are incomplete. Square wave measurement is missing in almost all types of products.

I thought so too.

How could a signal with no edges (sine sweep) test for edges (impulse or square response).

So I compared calculated impulse and step response via sine sweep in REW vs actual response using an impulse and sqaure wave signal through the gear as recorded in the air with Audacity.

The results to my expectantly waiting eyeballs were close enough that I haven't tried the test since, nor worried about it not having been performed.

I'd rate it as an "Ah haaa!" moment that totally failed as an "I knew it!" moment.

If this makes you curious I can link to the results, yet again.

You're new here.

 

[preload]

I can but I ran the very test on a real speaker in my last review just as PMA is doing here: https://www.audiosciencereview.com/forum/index.php?threads/morrow-sp3-review-speaker-cable.22834/

I used a Klipsch speaker as the load. Here are the results:

And I showed the difference between the two cables:

I am using proper scale so you get a sense of how small these numbers are (one thousands of a dB). Note that my Audio Precision analyzer is far more accurate than OP's scope measurements. Digital scopes are made for speed so their dynamic range is quite poor (usually 8 to 10 bit digitizers).

Interesting. Which Klipsch test speaker did you use? If the load has a relatively flat impedance vs. freq curve, the cable properties matter a lot less.

 

[pma]

I have only low electronic skills, please can you explain why a resistor with 5 times higher value parallel to a loudspeaker can do any termination (4/5 of the electric power is going thru the speakercoil) ?

In the post #28 @solderdude has explained that the 47 ohm resistor also affects the impedance near woofer and port resonance, makes it more flat. But there is also another reason why this resistor is used, and it is the elimination of cable reflections in the frequency range above 10MHz and RLC resonances above 1MHz. Yes this is of course inaudible, but it may result in amplifier instability and even oscillations. This issue is very complex and difficult to discuss on the basic level. Examples of burnt amplifiers working into long speaker cables loaded with mostly inductive load (at HF) were described several times. I have also experienced an instability of one amplifier loaded with 10m coax speaker cable.
Essentially, the cable driven from very low impedance (amp output) and having no termination (the speaker load would be inductive at frequencies >1MHz) is the worst case regarding reflections and oscillations. It is not the same case as to load the amplifier with a capacitor of the same capacitance value. Lumped or transmission line parameters start to be important.
Now, in case that the cable is terminated by a resistor with resistance near to cable characteristics impedance, the reflections and possible oscillations issue are highly suppressed. This is the main reason why the R1 is used in the post #1. Of course it does not make much difference at audio frequencies.

Below you may see a simulation of the issue and also a real measurement

Simulated oscillations

Oscillations measured on a real world power amplifier with 10m speaker coax cable. Terminating resistor stops the oscillations.

 

[DerRoland]

that the 47 ohm resistor also affects the impedance near woofer and port resonance, makes it more flat.

Thank you for your good explanation. You have better skills, that is clear. I'm speculate self of the flatening feature with the resistor.

But I'm unsave lately, does this 47 ohm resistor have an influence to the passiv crossover and the speaker performance at 30 and 70 Hz?

 

[restorer-john]

The speaker cable in fact is a kind of EQ with respect to the speaker used and its influence might be eliminated by using software EQ.

Or by including the speaker cable in the feedback loop...

(think Kenwood Sigma Drive from the early 1980s)

 

[restorer-john]

I am using proper scale so you get a sense of how small these numbers are (one thousands of a dB). Note that my Audio Precision analyzer is far more accurate than OP's scope measurements. Digital scopes are made for speed so their dynamic range is quite poor (usually 8 to 10 bit digitizers).

True, but when it comes to square waves (as @pma is using also in this thread) , it's a completely different story.

 

[KSTR]

But there is also another reason why this resistor is used, and it is the elimination of cable reflections in the frequency range above 10MHz and RLC resonances above 1MHz. Yes this is of course inaudible, but it may result in amplifier instability and even oscillations. This issue is very complex and difficult to discuss on the basic level. Examples of burnt amplifiers working into long speaker cables loaded with mostly inductive load (at HF) were described several times. I have also experienced an instability of one amplifier loaded with 10m coax speaker cable.

Very good point. Unterminated "electrically long" cable, even when its "DC" capacitance is just a few nF, can have an equivalent capacitance of several orders of magnitude more, at specific frequencies (and at other it appears as highly inductive). When you load an amp (with no or inadequate series R//L) near its GBW with this, all bets are off. The amp might be fine with 9m and 11m of cable but go up in smoke at 10m...

 

[restorer-john]

The amp might be fine with 9m and 11m of cable but go up in smoke at 10m...

Love it.

We love to paint mental pictures of amplifiers combusting, tweeters smoking, voice coils shooting across the room and embedding themselves in the plaster board. Not to mention the odd capacitor going off like a gunshot and blowing a hole in the ceiling.

 

[DerRoland]

Sorry for my next "low level" questions:

Can this happen with common class D amplifiers (usally build-in filtering of high frequencies above 30-60 kHz), too? What GBW stands for?

 

[solderdude]

does this 47 ohm resistor have an influence to the passiv crossover and the speaker performance at 30 and 70 Hz?

Nope. It only changes the maximum resistance the amplifier sees at all frequencies.

 

[solderdude]

Sorry for my next "low level" questions:

Can this happen with common class D amplifiers (usally build-in filtering of high frequencies above 30-60 kHz), too? What GBW stands for?

With class D the intention is to filter all HF anyway so it does not matter there. class-D amplifiers are bandwidth limited anyway while one can build analog amps with >100kHz bandwidths.

GBW