Modelling Audio Cables

[shadders]

Originally Posted by jcx

gnd return R,L modeling doesn't change conclusions, the "one sided model" is the universal way the lumped approximation is shown in the literature

I think that as a series circuit you will have a hard time finding practical difference from modeling both wire's R,L


for a real cable model accuracy improvement, in my linked post I mention that skin/proximity effect could be modeled with added mutual L, steering high frequency current to some parallel R branches, "concentrating" the current, increasing apparent resistance as the other R paths are starved

Hi,
I modelled quickly - so could be mistaken the lumped element and distributed element of a cable - from Hifi News - a UK publication.

I assumed 3metres - stated parameters in Hifi News are 0.36uH/m, 4.6mOhms/metre, 295pF/m.

I ran the lumped element into 8ohms resistance, and distributed element also into 8ohms.

B2Spice allow you to implement the graph as a formula.

For the lumped element i used the vdB probe - which is always taken to 0volts.

For the distributed element i used dB[magnitude(voltage across 8ohms)] which has slightly different results.

Lumped element rolls off at 18dB/octave

Distributed element rolls off at 12dB/octave.

The lumped element as a clean roll off, where as the distributed element has two resonant peaks before rolling off - both simulated to 100MHz.

So - i may have my simulation wrong as it was completed is haste, but i do believe that lumped element is different to distributed element.

Regards,

Shadders.

[deanxxx]

Originally Posted by shadders

I assumed 3metres - stated parameters in Hifi News are 0.36uH/m, 4.6mOhms/metre, 295pF/m.

EWB say -1dB @ 23kHz

I think it's fair to say we get a similar shaped curves for the two models. Out of interest, what did you get for -1dB point?

[deanxxx]

Originally Posted by shadders

i do believe that lumped element is different to distributed element.

Apart from the curves being different only beyond audibility (gentle roll-off v "spikey" looking response), I get the -1dB (which I think I read as being the all important cut-off point) to be the same for lumped and distributed simulations.

I'm no expert by any means so I'm not all that sure I'm doing anything useful or that relates to the real world. I hope to actually compare the EWB sims with real-world measurements at some point.

[shadders]

Originally Posted by deanxxx

EWB say -1dB @ 23kHz

I think it's fair to say we get a similar shaped curves for the two models. Out of interest, what did you get for -1dB point?

Hi,

At the -1dB point, i obtain 600kHz for the distributed response and 700kHz approx for the lumped response.

I think the 1dB at 23kHz may mean you may be using too many elements - or the cable length is quite long ?

Regards,

Shadders.

[DUG]

Originally Posted by shadders

Hi,

I assumed 3metres - stated parameters in Hifi News are 0.36uH/m, 4.6mOhms/metre, 295pF/m.

0.36uH/m seems a bit low

295pF/m seems a bit high

Do you know what gauge and spacing were they using?

[shadders]

Originally Posted by deanxxx

Apart from the curves being different only beyond audibility (gentle roll-off v "spikey" looking response), I get the -1dB (which I think I read as being the all important cut-off point) to be the same for lumped and distributed simulations.

I'm no expert by any means so I'm not all that sure I'm doing anything useful or that relates to the real world. I hope to actually compare the EWB sims with real-world measurements at some point.

Hi,

I may be wrong - not sure regarding the simulation of lumped and how i measured the signal of interest - across the 8ohm load.

As you have stated the differences are at the higher frequencies beyond the hearing capability of people.

A frequency response will only indicate the steady state levels at the speaker terminals - but a transient response may be more revealing.

I don't have golden ears - so i would not know if there are differences between cables - but too many people report that there is - this is in publications, not just forums - so scientific analysis must be pursued to resolve where the percieved differences occur.

Regards,

Shadders.

[shadders]

Originally Posted by DUG

0.36uH/m seems a bit low

295pF/m seems a bit high

Do you know what gauge and spacing were they using?

Hi,

This was a Highly Commended cable in Hifi News - states as above the specifications - other have lower 105pF, down to 30pF and up to 925pF.

So i used one that was recommended and had high capacitance and slightly below average inductance.

Regards,

Shadders.

[jcx]

multiple RLC sections are still called lumped element approximations - Spice also has a "continuous" transmission line model

some like to use just a single RLC section - may often be good enough for audio - but multi section RLC lumped TL can approach the continuous model in accuracy at increasingly higher frequency as the number of sections/cable length increase

and, as usual, audio frequency "transients" don't stress the accuracy of even single section RLC cable models for domestic <10m lengths compared to the continuous TL models - as long as both are ignoring skin/proximity effect

[Speedskater]

About ten years ago, Cyril Bateman did some cable modeling and testing. It quickly got complicated.

http://www.waynekirkwood.com/Images/...nteraction.pdf

http://www.waynekirkwood.com/Images/...ing_Cables.pdf

[davidsrsb]

Continuous models and lumped are identical as long as each lump is less the 10% of the electrical wavelength. It is unusual for either model to take into account changes of resistance due to skin depth or capacitance with dielectric constant over frequency.