Speaker cables don't influence harmonic distortion!

[vzaichenko]

15m cable, dummy load, measuring at far end (the load)

OK... got some material 😀
First - raw spectrums for different measurement configurations.
In the end - summary table.

Starting with dummy load as a reference.

 

[vzaichenko]

3m cable, speaker, measuring at the amp terminals

From now on - it will be a speaker, rated as 6 ohm.
Starting with the relatively short 3m cable.

 

[vzaichenko]

3m cable, speaker, measuring at the far end (speaker)

Now - same 3m cable, but measuring right at the speaker.
All measurements are done at 30, 50, 85, 100, 200, 500 and 1000 Hz.

 

[vzaichenko]

15m cable, speaker, measuring at the amp terminals

OK, now - the long cable, measuring close to the amplifier.

 

[vzaichenko]

15m cable, speaker, measuring at the far end (speaker)

Finally - the same long cable, measuring right at the speaker.

 

[vzaichenko]

SUMMARY

Here are the table and the diagram.

All measurements are done at 4V RMS at the output of the amplifier.
I could not go higher as it was rather loud already with the speaker - I couldn't hear normally for about an hour 😛 At 200Hz the lights at the ceiling were vibrating, adding some "ringing" to the sound from the speaker.

Pretty interesting. Distortion peak at 200Hz is clearly visible - that's where the speaker's impedance is at the minimum.

With the long cable, low-frequency signals experience noticeable distortion increase.
However, with a 3m cable, performance at the speaker end is pretty good.

Cheers,
Valery

Edit: added a pdf with much better resolution 😎

 

[BesPav]

Finally - the same long cable, measuring right at the speaker.

Now you know the truth. Feedback need to be taken straight from load terminals or non-feedback amp must be placed straight at load. Output LR-filter must be compensated at all audioband. Wire resistance must be compensated at least to 1 kHz. Passive filters must be avoided. So best system is an active-DSP system with per-channel amplifying.

Distortion peak at 200Hz is clearly visible - that's where the speaker's impedance is at the minimum.

Yes. Load there are mostly current-consuming and damping factor (for a feedback'ed amp) mostly determined by cable resistanse. So voltage drop on the wire resistance is at maximum and you see distortion at load side.

 

[DF96]

For one thing, you can’t test *all* cables, and certainly high impedance thin gauge cables can’t be the same as low impedance thick conductor cables. If an amp is sensitive to load impedance it will have an effect. If a speaker is sensitive to an amp’s output impedance, it will have an effect and sound different. I know that the DF of an amp really changes with cable and without a cable attached to the dummy load.

Cable resistance does not generate distortion; metallic conductors are highly linear. However, high cable resistance (i.e. a poor cable, either too long or too thin) can expose other problems such as damping or speaker nonlinearity.

Pretty interesting. Distortion peak at 200Hz is clearly visible - that's where the speaker's impedance is at the minimum.

Minimum speaker impedance means maximum effect of speaker nonlinearity making distortion current visible. The distortion current is still there with no cable but you can't see it so easily. You are actually measuring an aspect of speaker nonlinearity, not cable nonlinearity. This may disappoint some people.

 

[BesPav]

You are actually measuring an aspect of speaker nonlinearity, not cable nonlinearity.

<——of non highly damped speaker——>

 

[vzaichenko]

Cable resistance does not generate distortion; metallic conductors are highly linear. However, high cable resistance (i.e. a poor cable, either too long or too thin) can expose other problems such as damping or speaker nonlinearity.


Minimum speaker impedance means maximum effect of speaker nonlinearity making distortion current visible. The distortion current is still there with no cable but you can't see it so easily. You are actually measuring an aspect of speaker nonlinearity, not cable nonlinearity. This may disappoint some people.

Exactly - the cable is no problem. The "bad thing" is the speaker - a rather non-linear system. Especially at the low frequencies. The longer the cable - the higher output impedance is seen by the speaker - the less non-linear speaker is controlled by the amplifier's output.

The "closer" the speaker to the amplifier - the lower the distortion is. We can make it closer either with a shorter cable, or with a"fatter" cable. These properties are not equal, but still - it makes sense to use the shorter and the fatter cables for better control of the speakers = lower distortion. Assuming the amplifier provides the output impedance low enough and the output current high enough when required.

 

[29285]

This is a brilliant result. Comparing the plots of 15m cable with dummy load or with speaker load tells the hole story as stated in the previous posting.

 

[DF96]

<——of non highly damped speaker——>

Damping and nonlinearity are two quite separate issues. Don't confuse them.

The "closer" the speaker to the amplifier - the lower the distortion is. We can make it closer either with a shorter cable, or with a"fatter" cable. These properties are not equal, but still - it makes sense to use the shorter and the fatter cables for better control of the speakers = lower distortion. Assuming the amplifier provides the output impedance low enough and the output current high enough when required.

There are people who advocate going to the opposite extreme: current driving of speakers. Their logic is that it is current which moves the voice coil, so if the current is set by the amplifier (instead of the voltage) then the result will be more transducer linearity. If they are right then you can get some of the advantage of this method by using long thin speaker cables from a conventional voltage-source amplifier.

I am not convinced that they are right. Speaker nonlinearity has various causes but there is no a priori reason to assume that current drive is better. For one cause of nonlinearity (voice coil displacement in a non-uniform magnetic field) I think voltage drive is better: as the voice coil moves out of the gap it will generate less force but also less back emf so with a conventional system the current will increase to partly compensate; a current-driven system cannot compensate. Suspension nonlinearity could act in the same way: if anything tends to restrict the motion this will increase the current in a conventional system, but not a current-driven system.

That is one of the beauties of audio: whatever the physics says, there are always people who take the opposite view.

 

[BesPav]

Damping and nonlinearity are two quite separate issues. Don't confuse them.

Try to understand and differentiate the causes and consequences. This nonlinearity was caused by voltage drop on wire resistanse, which created by flowing current. Measurement setup indicates this as distortion.

If Valery could take measurements while speaker will be connected straight to amplifier output terminal (and amp’s feedback is taken from output terminals, not somewhere from the amp’s PCB) - there will be much less nonlinearity.

 

[RajkoM]

subscribe

 

[TNT]

...
That is one of the beauties of audio: whatever the physics says, there are always people who take the opposite view.

I'd call that a wart 🙂

//

 

[DF96]

Try to understand and differentiate the causes and consequences.

That is what I am trying to get you to do.

This nonlinearity was caused by voltage drop on wire resistanse, which created by flowing current. Measurement setup indicates this as distortion.

No. The nonlinearity is in the current. The cable resistance merely makes this visible; it does not cause it.

If Valery could take measurements while speaker will be connected straight to amplifier output terminal (and amp’s feedback is taken from output terminals, not somewhere from the amp’s PCB) - there will be much less nonlinearity.

No. The nonlinearity would still be there, but we would no longer be able to see it. The nonlinearity is caused by the speaker impedance being nonlinear. The cable is linear.

 

[BesPav]

No. The nonlinearity is in the current. The cable resistance merely makes this visible; it does not cause it.

No. The nonlinearity are caused by wire resistance.

No. The nonlinearity would still be there

Nonlinearity of speaker impedance doesn’t cause signal distortion. It can cause phase lag, it can cause huge current, but not distortion.

The nonlinearity is caused by the speaker impedance being nonlinear.

It will be damped/shunted by low amp’s output resistance connected in parallel. There will be some milliOhms and lower depending of amp’s feedback depth at a given frequency.

Distortion was caused by bad damping. Bad damping caused by improper feedback point and small but finite connection resistanse.

 

[dadod]

No. The nonlinearity are caused by wire resistance.

You are saying that wire resistance is non linear??😱

 

[xrk971]

Pure conducting ideal wires are different than real cables which have intrinsic parasitic capacitance and inductance. They also have metal to metal terminations of dissimilar metals and joints etc. I don’t think all of these system effects are linear.

 

[BesPav]

You are saying that wire resistance is non linear??😱

LMAO
Of course - no.
[emoji6]
(UPD
But really, your wire is a strongly capacitive for my amp based on MRF6S18060N)

It replaces relatively low amp’s output resistance and interacts with nonlinear load. Low load impedance at some frequency causes higher current to run through the load. That current causes voltage drop in all loop elements. Our measurement setup sees part of voltage drop, being caused by that current at wire resistance (or being divided at load voltage) as distortion.
Voila.