R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: [1] 2 3 ... 8   Go Down

Author Topic: cables - facts and fiction  (Read 52214 times)

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
cables - facts and fiction
« on: February 18, 2005, 11:59:18 AM »

First topic: cable capacitance for analog audio signals.

Is cable capacitance important for transmitting analog audio signal? If so, when is it important?

The ability to drive capacitive load (such as cable capacitance) is a requirement for the audio driver. The potential problem of a poor driver design is undesirable oscillations. One can design an ultra low impedance source (near 0 Ohms) capable of driving a capacitive load. A more common solution is to insert a small value resistor in series (between the driver and the cable). A typical value for that resistor in audio is 75 Ohms.

Preventing oscillations is an issue for the gear designer, not for the end user. The user should not be required to "help" prevent oscillations by using low cable capacitance.

A series R and a parallel C circuit (such as driver and cable capacitance) is a low pass filter, but at 75 Ohms source resistance (or less) the attenuation even at 20KHz is negligible. The load resistance at most cases is high and does not noticeably impact the flatness response. In fact, the lower the load resistance, the less high frequency attenuation (better flatness response).

Below is a plot showing attenuation at 20KHz with 4 loads -
1KOhm (red), 2KOhm (blue), 5KOhm (black) and 10KOhms (purple)
The driver is 75Ohms and the plot shows the effect of capacitance from 0 to 10000pF. Note that the loss at 20KHz is less then .04dB at the maximum capacitance. A high capacitance cable, say 100pF per foot would require 100 feet of cable to build up 10000pF.  

index.php/fa/679/0/

Applications where the source impedance is much higher can suffer from much higher attenuation. A 10000pF capacitance driven by 600Ohms into 10KOhm load will cause a loss of about 1.7dB at 20KHz. Clearly high impedance drive can be an issue, and such conditions should be examined by users of high impedance microphones driving long cables.

Of course, lower source resistance drive under the same conditions (attenuation at 20KHz into 10KHom load with 10000pF cable capacitance) can not hurt. A 10 Ohm source will yield a 0.0007dB loss at 20KHz.

What is the attenuation when driving speakers? Speaker drivers are very low impedance, and the speakers themselves are low impedance load. Let us first assume a .1 Ohm driver (and we include the cable resistance) - the attenuation at 2,4,8 or 16Ohm resistive speaker load is about 0.1 micro dB(!)

Lets raise the driver and cable resistance to 1 Ohm (very high value) and the attenuation will still be negligible - better then .006 mili dB for 16 Ohm load.

Yes, it is true that speakers do not behave like a pure resistance, but the numbers I stated can be off by a factor of 1000, and the attenuation due to speaker cable capacitance is still a non issue (0.006dB instead of 0.006 mili dB).

While cable capacitance is an issue at higher frequencies (higher then audio signals), it only matters when the source impedance is very high and the cables are very long. It certainly is a non issue for speaker wires.

Regards
Dan Lavry
www.lavryengineering.com
Logged

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: cables - facts and fiction
« Reply #1 on: February 18, 2005, 04:43:58 PM »

Second topic: cable resistance for analog audio signals.

Cable resistance and the cable driver output (source) impedance work as a voltage divider against the load impedance at the destination receiving the signal.

Most often, the destination is a fixed value and resistive for audio frequencies, such as 10 KOhm or 600 Ohms resistive load.
The cable itself behave as a constant impedance (fixed resistance, capacitance and inductance) so we end up with a constant voltage divider. Even 1000 feet of 18 gauge wire is only 6.4 Ohms (12.8 for a "round trip") which does not cause much loss against 600 Ohms or higher load.

However, speakers are not a constant load. Speakers impedance does vary over the audio frequency range. The speaker designer may attempt to make the end result behave as a constant resistive load, but the end result is often far from constant, and certainly not resistive over the frequency range.

The question is: what is the effect of the varying load impedance (over the frequency range) against the constant cable resistance? This question requires a good model of a speaker. Many such models are available, and vary from speaker to speaker. For the sake of generality, let me simplify the problem to model a speaker as a resistive load. Let me assume a speaker that varies over the frequency range between 3 Ohms and 30 Ohms. The question is: What would such load variations do to flatness response?

Given that we are talking about speaker wire, let us take the following wire gauges (AWG):
00 with diameter of .36 inch
2 with diameter of .25 inch
4 with diameter of .2 inch
6 with diameter of .16 inch
8 with diameter of .13 inch
10 with diameter of .1 inch
12 with diameter of .08 inch

For the 7 wires, the deviation between 3 and 30 Ohm load is plotted below (in dB voltage loss) for distances between 10 and 50 feet of a speaker wire pair.

index.php/fa/681/0/

A 50 feet of 12AWG cable may cause about 0.4db deviation in voltage (.2dB power deviation). The same 12AWG wire at 10 feet length will cause about 0.08dB voltage attenuation (.04dB power). A 10 AWG wire at 20 feet is good for .1dB voltage deviation.

Clearly I am not stating where the deviation occurs, and the argument is built on an assumption that the load varies by a factor of 10 to 1 and is resistive. I will take some time to model the behavior of real speakers, but this ball park exercise shows that the deviations in flatness response due to speaker wire are very small compared to the fluctuations in flatness response of the speakers themselves, where a few db deviations over the frequency range is rather common.

From resistive analysis point of view, for most speakers, a reasonable length (say 20 feet) of 12AWG wire would do very well.

I would not rule out that one can pick differentiate between say 50 feet of 12AWG and 50 feet of say 6AWG. A 0.4dB voltage loss at the frequency where the speaker resistance drops to 3 Ohm may be noticeable in an ABX test. At 20 feet with say 10AWG, the deviation is only .1dB.

The lion share of the problem of deviation from flat response is the speaker itself. The cable resistance is nearly a non issue.

I will post the results of a more accurate model in the near future.

Regards
Dan Lavry
www.lavryengineering.com  
Logged

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: cables - facts and fiction
« Reply #2 on: February 18, 2005, 05:07:01 PM »

Third topic: Speaker wire and Inductance

Speakers are complicated devices. A good approximation to view a speaker (a driver), often used by speaker designers is an equivalent circuit developed by Thiele and Small, and the equivalent circuit parameters are often available from the driver manufacturers.

Note that the parameters consist of both electrical and electromechanical resistances, inductances and capacitance so that the designer ends up with a electrical network describing a speaker driver. Each driver (be it a sub woofer, woofer or tweeter) is far from optimal or desirable device, having an electrical peak (resonance), a mechanical resonance and certainly a need to be corrected (compensated) with a network.

The compensation is mostly done via conjugate filters - a circuit made of "real" resistors, capacitors and sometimes inductors to "counteract" the combination of electric and electromechanical behavior of the driver. One such common network is a Zobel Circuit, but there are others... So a speaker cabinet can be looked at as 2-3 drivers with some crossover networks and compensation networks, and one of the goals is to have a constant restive impedance across the audio range (such as 8 Ohms)...

I looked at some of the networks wondering about how much series resistance and inductance it would take to alter the response or ruin the compensation of the drivers or the crossover networks. At first glance it does not seem possible that the resistance of a 10-30 feet of say 10 gauge wire could do much to the flatness response. Can we have just enough inductance to interact with the network in a way that will cause a little bit of high frequency loss? I am including a plot for a very simplistic model, assuming a perfect speaker (8 Ohm resistive load behaviour at 20KHz and the effect of a cable inductance, assuming 1uH per foot. The picture shows amplitude response to 20KHz for 10feet, 20 feet and 30 feet distance.

Note that we are talking about PASSIVE speakers. All I am saying is does not hold for active (powered speaker), where the whole issue of cable type goes away completely.

Below: the red line (10 feet cable) shows around .1dB loss at 20KHz. The Black (30 feet cable) line shows almost a dB at 20Kh and a 1/4dB at 10KHz. It all assumes 1uH per foot and an ideal 8 Ohms speaker. This plot is just a rough first approximation.

index.php/fa/682/0/

Obviously cables have resistance, capacitance and inductance. Does any of it matter for audio frequencies? The capacitance matters least; the resistance does not matter much. The inductance seems to have a slightly higher impact, not much but enough to make me wonder why people buy the huge heavy wires with less resistance but higher inductance.  

I guess the answer is "marketing".

Regards
Dan Lavry
www.lavryengineering.com
Regards
Logged

JGreenslade

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 824
Re: cables - facts and fiction
« Reply #3 on: February 19, 2005, 10:22:50 AM »

Hi Dan,

Considering your forum does not allow debate based on subjective reasoning (how much power does subjectivity give marketing?), I wonder if you would consider it right if I were to email a link for this topic to some of the "exotic" cable manufacturers? I wonder if they could find the time to respond - they could well prove to be terribly busy developing new products...

BTW, before you start a thread on "high-end" mains cables, may I suggest posting a primer for anyone willing to participate in the debate, in the form of a temporary sticky link explaining how psu reservoir capacitors and rectifier diodes function...

Regards,
Justin
Logged
Audio is a vocational affliction

"there is no "homeopathic" effect in bits and bytes." - HansP

Joe Crawford

  • Full Member
  • ***
  • Offline Offline
  • Posts: 107
Re: cables - facts and fiction
« Reply #4 on: February 19, 2005, 05:07:29 PM »

Dan,

If you haven't already considered it, this looks like a good subject for another one of your white papers. Add in the skin effect discussion from the other thread and it just about covers "studio cabling for dummies".  Thanks again and keep the information flowing!

Joe Crawford  
Logged

Sahib

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 429
Re: cables - facts and fiction
« Reply #5 on: February 20, 2005, 01:47:28 PM »

thermionic wrote on Sat, 19 February 2005 15:22

Hi Dan,

BTW, before you start a thread on "high-end" mains cables, may I suggest posting a primer for anyone willing to participate in the debate, in the form of a temporary sticky link explaining how psu reservoir capacitors and rectifier diodes function...

Regards,
Justin


I think this would be a very short debate but I am in.

Cemal
Logged

Kendrix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 988
Re: cables - facts and fiction
« Reply #6 on: February 21, 2005, 10:57:43 AM »

Dan,
Aside from the minor impacts to signal amplitude across the audible frequency range that you've calculated...
What about any frequency dependant phase effects of the cable impedance?  
Is it possible that such phase-effects may account for perceived sonic differences attributed to cables?  
I've never experienced these differences myself.  
However, people who are far better at this than I have reported hearing cable-related differences.

Logged
Ken Favata

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: cables - facts and fiction
« Reply #7 on: February 21, 2005, 12:49:27 PM »

Kendrix wrote on Mon, 21 February 2005 15:57

Dan,
Aside from the minor impacts to signal amplitude across the audible frequency range that you've calculated...
What about any frequency dependant phase effects of the cable impedance?  
Is it possible that such phase-effects may account for perceived sonic differences attributed to cables?  
I've never experienced these differences myself.  
However, people who are far better at this than I have reported hearing cable-related differences.




I got rather curious and spent some serious hours gathering a handful of speaker models - circuits that represent a speaker behavior by use of networks made of inductors, capacitors and resistors. The network are pretty complex. A 2 way speaker model may require 7 resistors, 5 inductors and 4 capacitors for the 2 drivers, and then you get to add some the crossover network – more inductors and caps…

Well, I got “lazy” and decided to fire up the analog spice software (circuit analysis program). I modeled 2 cases side by side:
1.   Direct connection
2.   50 feel connection via cable

Spice offers a 4 port transmission line model for the cable, and it is a pretty fine model. I got to plug in values for the resistance per meter, capacitance per meter, inductance per meter and even a value for the leakage (which is negligible in this case).

After all that work, my original conclusions are about the same. For me, spice is not a tool for precision anaysis, but it works great for some "ball park" estimates (say to better then +/- 25% estimate). If you want to be very tweaky about it, pay attention to cable inductance first, resistance second, capacitance last, and that does not correspond to the huge cables I see in stores – offering lower resistance at the expanse of higher inductance…

Regarding phase: Linear passive networks in the frequency regions we are talking about (no skin effect) are made of resistors, caps and inductors. When I see a flat response to say 100KHz, I am pretty sure that there are no phase problems up to 20KHz… (for example, a 1KHz pole or zero will impact the phase within an octave or so, not much further).  
But yes, of course I looked at phase and I do it as a matter of course. It was particularly interesting for me when modeling the speakers just around the crossover network frequency. That is where the main phase problems are, not in the cable. But we given that we are talking about cables, again, the cable inductance makes much more differance then its resistance, and the impact will be near 20KHz. I agree that a .25dB deviation in amplitude may be coupled with as much as 30 degrees deviation from linear phase at 20KHz.

Note: It is difficult to model all the cases, because the drivers (power amplifiers) come into play. It is almost common practice for power amp designers to insert a series network (resistor in parallel with an inductor). At low frequencies the inductor shorts the resistor, and at high frequencies the inductor “disappears” (high impedance) leaving a series R to combat possible high frequency oscillations. As a rule, that network seem to have much more effect then the cable itself.

I would be glad to redo the whole experiment on other speaker models, if anyone has such models to share.

Regards
Dan Lavry
www.lavryengineering.com        
Logged

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: cables - facts and fiction
« Reply #8 on: February 22, 2005, 04:29:50 PM »

Kendrix wrote on Mon, 21 February 2005 15:57

Dan,
Aside from the minor impacts to signal amplitude across the audible frequency range that you've calculated...
What about any frequency dependant phase effects of the cable impedance?  
Is it possible that such phase-effects may account for perceived sonic differences attributed to cables?  
I've never experienced these differences myself.  
However, people who are far better at this than I have reported hearing cable-related differences.




For the record, I am not saying that one can not hear cable differences. The data I posted in the previous posts show that it is very possible to achieve some differences, especially at high frequencies, with deviations in both gain and phase.

At issue is not the ability to recognize that there is a difference. The issue is to go for better results (such as better flatness response).

Clearly, a a 50 foot run of 20AWG is a real bad choice - at 10 Ohms per 1000 feet, you end up with 1 Ohm of resistance, and if your 100 watt amp is trying to power an 4 Ohms load, you have 4.5 Ampere on the wire - a loss of 20 watts! Running that much current on a 20AWG is silly. But at 10AWG, the resistance is about .1 Ohms for 50 feet distance. So at 4 ohms you lose 2.5 watt. At a more realistic distance of 10 feet distance it becomes 0.25 watt and in fact 0.1 Ohms is so small that there is no reason to increase diameter.

Meanwhile, the inductance that actually causes more harm to gain and phase is being "neglected". Instead, some cable makers sell you on BS such as skin effect and other nonsense like directional speaker cables with arrow on the sleeve! It is a crock!

Regarding skin effect, yes there is such a thing, and those of us that worked in high speed electronics know that at low frequencies (such as audio) it does not matter, because the effect is so small. The most critical ears and measurements do not demand better then say .01dB flatness response, and skin effect on a 20AWG at audio (20KHz)over 50 feet is less then 0.0001dB into 4 Ohms speaker. It is not even funny.

Regarding the wire directionality - it is sick! Copper DOES NOT have directional preference, and anyone claiming otherwise is lying. And even if I were wrong (which I am not), the signal is AC - bidirectional, and in fact, the same amount of current going in one direction fro amp to speaker will come back from the speaker to the amp...

There is a lot of deception in that speaker wire business. Unfortunately, some companies made a lot of money with such deceptive practices, and much of the crock moved to other type of cable, not just speaker wire.

Regards
Dan Lavry
www.lavryengineering.com
Logged

dcollins

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2815
Re: cables - facts and fiction
« Reply #9 on: February 23, 2005, 01:04:14 PM »

danlavry wrote on Tue, 22 February 2005 13:29


There is a lot of deception in that speaker wire business. Unfortunately, some companies made a lot of money with such deceptive practices, and much of the crock moved to other type of cable, not just speaker wire.



Yeah.  People forget that there are far more more demanding  areas of signal transmission than audio (medical, aerospace, etc. ) and if _any_ of the mechanisms claimed for audio cables existed they would be the guys who would have discovered the effects.  Not some guy writing for a Hi-Fi magazine!

Do you know how much industry would pay for a "micro-diode?"  (this is a typical audiofool claim to explain otherwise nonexistent cable effects)

Too bad there is no such thing....

And J. C. Maxwell never talked about Phi, either.  Why is that?

http://www.cardas.com/content.php?area=insights&content_ id=6&pagestring=Golden+Section+Stranding

DC

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: cables - facts and fiction
« Reply #10 on: February 23, 2005, 05:45:19 PM »

dcollins wrote on Wed, 23 February 2005 18:04

danlavry wrote on Tue, 22 February 2005 13:29


There is a lot of deception in that speaker wire business. Unfortunately, some companies made a lot of money with such deceptive practices, and much of the crock moved to other type of cable, not just speaker wire.



Yeah.  People forget that there are far more more demanding  areas of signal transmission than audio (medical, aerospace, etc. ) and if _any_ of the mechanisms claimed for audio cables existed they would be the guys who would have discovered the effects.  Not some guy writing for a Hi-Fi magazine!

Do you know how much industry would pay for a "micro-diode?"  (this is a typical audiofool claim to explain otherwise nonexistent cable effects)

Too bad there is no such thing....

And J. C. Maxwell never talked about Phi, either.  Why is that?

  http://www.cardas.com/content.php?area=insights&content_ id=6&pagestring=Golden+Section+Stranding

DC




This is truly amazing! I never heard of that BS before, so I went to that sight and Saw a lot of interesting stuff. The first comment was:  

"The industry's best designers, constructors and reviewers: The Academy for The Advancement of High End Audio, awarded to George Cardas..."

I never heard of that "academy", and I have less then low opinion of many of the audio industry awards, including those that proport to be technical awards, some are given to products that are a complete and total crock!

But reading on I saw:

George introduced the concept of Golden Section Stranding to high-end audio, but Golden Ratio, 1.6180339887...: 1 is as old as nature itself. Golden Mean forms the mathematical proportions nature uses to shape leaves and sea shells, insects and people, hurricanes and galaxies...

Go measure the leafs, the sea shells, insects and people. I think you will not find that ratio anywhere Smile

I wonder what or if any entities in that cable are at a ratio of 1.6180339887. It is not cable length to diameter, so what could it be? Copper to insulator cross sectional area ratio accurate to 10 digits? What a Bunch of...  

But thanks for sharing. It is good to laugh but I feel sorry for the victims that buy it.

Phi? it is just like Pi but with an added h for horse manure Smile

Micro-diode? I never heard of that one but it sounds interesting Smile

Any other examples of cable BS?  

Regrads
Dan Lavry
www.lavryengineeting.com

Logged

mcfate

  • Newbie
  • *
  • Offline Offline
  • Posts: 6
Re: cables - facts and fiction
« Reply #11 on: February 23, 2005, 11:24:50 PM »

OK - please tell me what (the hell) is going on with this?

http://www.mitcables.com/technology/

Is this the pile that it seems like it must be, or do we all need to run out and buy $3000 10' XLR cables?  I actually was forced to "demo" these at one point - and uh, well,  yeah.
Logged

Sahib

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 429
Re: cables - facts and fiction
« Reply #12 on: February 24, 2005, 07:41:14 AM »

My two favourites are 'eddy current losses on speaker wires' and
'active signal cables'.

Of course the eddy current issue is patheticaly false but the active wire is interesting. It consists of applying a low DC onto the screen to 'bias' it. These people must really be spending days and days on trying to come up with all these cock and bull stories but as Dan said it is really said to see people falling for them.

Cemal

Logged

danlavry

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 997
Re: cables - facts and fiction
« Reply #13 on: February 24, 2005, 06:18:35 PM »

mcfate wrote on Thu, 24 February 2005 04:24

OK - please tell me what (the hell) is going on with this?

http://www.mitcables.com/technology/

Is this the pile that it seems like it must be, or do we all need to run out and buy $3000 10' XLR cables?  I actually was forced to "demo" these at one point - and uh, well,  yeah.


The URL does not open. I did a quick search on the cables and here is some of what I found:

"Reference Series speaker cables
8' MIT Oracle V3.1 speaker cable. $5,495/pair.
8' MIT Oracle V3.1 Bi-Wire speaker cable. $5,995/pair.
 
8' MIT Oracle V2.1 modular speaker cable. $9,495/pair.
(Click here for more info.)
8' MIT Oracle V2 Bi-Wire modular speaker cable. $9,995/pair.
 
8' MIT Oracle V1 modular speaker cable. $15,995/pair.
8' MIT Oracle V1 Bi-Wire modular speaker cable. $16,995/pair.
All Oracle Series cables are available in the following versions:

Ultrawide Bandwidth for use with Spectral and other high-speed amplifiers
V1.1 Ultrawide is known as Prism.
Wide Bandwidth for use with for typical solidstate and high current tube amplifiers
Wide Bandwidth EX for use with typical solidstate and high current tube amplifiers that need extra high frequency articulation.
Please feel free to call for a recommendation for your amplifier and system.
"

It is my considered opinion that such outrageously priced cables will not outperform an average ordinary 10AWG copper wire. The $16,995 price is about thousand times more expensive than in needs be. At 8 feet length, a 10AWG cable does not present enough capacitance, resistance or inductance to impact the signal.  

Regards
Dan Lavry
www.lavryengineering.com
Logged

dcollins

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2815
Re: cables - facts and fiction
« Reply #14 on: February 24, 2005, 11:04:25 PM »

Pages: [1] 2 3 ... 8   Go Up
 

Site Hosted By Ashdown Technologies, Inc.

Page created in 0.118 seconds with 19 queries.