Cable and EMC discussion (Split from "What if the Audio Critic was completely right")

[wakibaki]

 
Oh, please stop offering to explain things, it's hard to see it as genuine, the impression is more one of intent to be patronising and insulting.
 
Cheers, w

 

[jnjn]

Quote:

 
OK, firstly, with regard to cable geometries and impedance, you prove my point precisely.
 
You twisted the lamp cable, the impedance changed. 6 turns per foot is trivial to achieve by accident, twisted pair is specified with turns per inch.
 
You exceeded the bend radius of the co-ax, the impedance changed.
 
No surprises there then.
 
jnjn, do you stand by this statement? 'even putting the box in renders comparisons meaningless'
 
Cheers, w
 
 

copied and red for response..
 
OK, firstly, with regard to cable geometries and impedance, you prove my point precisely.
Honestly, we can't figure out what your point is.  First you said zip wire has no characteristic impedance, which was soundly trashed by someone who simply performed one measurement.
 
Then you said it can't be twisted.  Again, totally trashed by somebody who actually twisted zip and measured the characteristic impedance with various pitches.
 
Now this hedging:
You twisted the lamp cable, the impedance changed. 6 turns per foot is trivial to achieve by accident, twisted pair is specified with turns per inch.
Which is trashed by simple inspection.  He twisted the cable to various amounts and reported the values.  Only at 180 degrees per inch did it start to reduce, and it's unclear if it was a response to insulation modulus, copper cross section ovality, or helicity based magfield cancellations. (I'll explain that if you do not understand).
 
And twist pitch is determined by the size of the insulated wire.  I twist 32awg to specified pitches, I twist the 4/0 cables to a different pitch, I twist the 535kcmil pair to another.  Twisted by definition, is just that....twisted.
 
You exceeded the bend radius of the co-ax, the impedance changed.
If you had paid attention to the actual post, you would have realized that he mentioned an air tube.  Not a coax.  And he lamented the inability of many weird cables to take a bend and maintain impedance.  I deal with bend radius concerns on a daily basis, as I'm installing somewhere in the order of 10 kilometers of LMR 240 in tray, as well as LMR 400, LMR 900, and 3/8ths diameter heliax (that stuff's a PITA).  There, it is extremely important to maintain bend radius for the simple reason that a tight bend causes the inductance to go up, the prop velocity to go down, and the wires are bundled in groups of 4 with inter-wire delay uncertainty to be kept below 60 picoseconds even for the 100 foot runs.  Most cable tray waterfalls have a 4 inch radius, and even that scares me with the LMR 400.
 
No surprises there then.
Your changing your stance was certainly no surprise to me.
 
jnjn, do you stand by this statement? 'even putting the box in renders comparisons meaningless'

I have absolutely no idea what it is you are asking.  Perhaps you should read the full sentences within their contexts. I was quite clear in my statements, culling small portions of a sentence as a form of entrapment is rather disingenuous.  Your audience here is considerably beyond that, you should as well.
 
As to your other post....I am not in the business of publishing.
 
Cheers, jn


 

 

[wakibaki]

 
 

There are systems out there which are not susceptible to EMC problems

So presumably a system for inserting a switch without invalidating comparison testing is not impossible to engineer.
 
Since you have such a great understanding of the issues, why not construct such a system, perform some tests, and demonstrate to the whole world of blind testing where they've been going wrong all these years.
 
It'd make your reputation.
 
Or break it.
 
Cheers, w 

 

[jnjn]

 
Quote:

 
Oh, please stop offering to explain things, it's hard to see it as genuine, the impression is more one of intent to be patronising and insulting.
 
Cheers, w

How else are you going to learn?? 
 
You've no understanding of EMC.  You seem challenged with transmission lines.
 
I've offered to provide you links to material so you can learn, yet you flat out refuse to learn the material.  Your lack of knowledge on the topics clearly present no hindrance to you typing away.
 
I can teach you, others can teach you, you could learn it from links or books, yet you refuse to learn.
 
Cheers, jn
 

 
 

 

[jnjn]

Quote:

 
 

So presumably a system for inserting a switch without invalidating comparison testing is not impossible to engineer.
 
Since you have such a great understanding of the issues, why not construct such a system, perform some tests, and demonstrate to the whole world of blind testing where they've been going wrong all these years.
 
It'd make your reputation.
 
Or break it.
 
Cheers, w 

People are already doing that.
 
The fact that you do not know of them, well, I am trying to get you to learn.
 
Cheers, jn
 
 

 

[wakibaki]

 
Quote:

4.  ABX is flawed whenever a switchbox is used.  Changes in the system caused by even putting the box in renders comparisons meaningless.  This is from EMC concerns.
 

 
Is this what you wrote?
 
Do you stand by it?
 
A simple yes or no will suffice.
 
Please don't feel that I am trying to trap you, I just want to understand.
 
Cheers, w

 

[wakibaki]

Quote:

Originally Posted by jnjn /img/forum/go_quote.gif
 
I am trying to get you to learn.
 

 
I am trying to get you to learn, too, or perhaps you feel that you have nothing further to learn. Perhaps we could learn together?
 
Cheers, w

 

[wakibaki]

 

Quote:

Originally Posted by jnjn /img/forum/go_quote.gif
 
You've no understanding of EMC.  You seem challenged with transmission lines.

 
I'd remind you that you're the one who said this:-
 
Quote:

When a signal in one channel is passed from the source to the amp, it does so by driving a voltage on the center core wire of that channel's coax.

 
Whereas anybody who knows thing 1 about transmission lines knows that a signal on a tx line propagates as a wave front with electric charge being established on both conductors. It's the rate of propagation that defines the line's velocity factor. Look it up, or perhaps even think a little. Here's a reference... http://www.allaboutcircuits.com/vol_2/chpt_14/3.html
 
So that's just 1 thing you could learn from me.
 
Cheers, w

 

[anetode]

Maybe I'm just dense, but if there's no audible ground loop hum and you're using cables and components of reasonable quality, then why worry further about EMC?
 
If something is wrong with the signal transmission to such an extent that the audio is noticeably degraded, most reasonable people will figure out a solution without having to buy into mad overpriced tweakery. I have a simple little dot tube amp which picked up interference and was susceptible to hum. It's like an ancient artifact, really. So I moved it a few feet away from all the other equipment and picked up an 8 dollar isolator and that was enough to address the supposedly staggering EMC concerns.
 
You have to remember that in a "Top Ten" magazine article there's not much space to go into all the manifold interpretations of physics. In the vast majority of cases "R, L, C" is all the customer has to grasp.
 
We're not talking about a peer-reviewed study posted in a scholarly journal. I'll take Aczel's cynicism over the self-deluded ramblings of 6moons anyday.

 

[mikeaj]

The point being made about driving voltage down the center line (yes there's a wave front going down the return lines as well), was that the return current may be split into multiple different paths with certain cabling configurations.  If in fact the return current doesn't match the current being sent on one coaxial line, then you've got a potentially non-trivial problem.
 
What I don't see is how a potential problem will always impact proper operation in a non-trivial way.  Certainly there have been ABX tests with switchboxes where the configuration suffered no ill effects as a result of the cabling, switchbox internal structure, or any detail of how it was connected?

 

[jnjn]

I've been asked why I respond to trolls.  My belief is that a discussion even with trolls can be valuable to all.  For example, this discussion:
 
Quote:

 
 
I'd remind you that you're the one who said this:-
 
Quote:
 
Whereas anybody who knows thing 1 about transmission lines knows that a signal on a tx line propagates as a wave front with electric charge being established on both conductors. It's the rate of propagation that defines the line's velocity factor. Look it up, or perhaps even think a little. Here's a reference... http://www.allaboutcircuits.com/vol_2/chpt_14/3.html
 
So that's just 1 thing you could learn from me.
 
Cheers, w

It a tad deeper than that.  Please don't rely on internet links to learn this stuff, typically it's "dumbed down" to levels where non engineers can start to get a grasp of the concepts.  I recommend the hyperphysics website, as they have a well organized structure allowing one to peruse the various ee stuff in a logical fashion.  Caution is warranted however, as most of the material does require pre-existing knowledge to allow a full understanding.
 
Now..
 
A coaxial cable used for transmission of audio signals shields via two mechanisms.
 
1.  Electric field shielding.  The outer braid interrupts the permittivity of space, (in essence a physical barrier to penetration of e-fields), such that the inner core wire does not capacitively couple to external sources of e-field.  In the world of EMC, this condition is more relevant when the victim circuit as it were, has an impedance above 377 ohms.  This number is of course, free space impedance, and is a good general threshold between magnetic effects on victim circuits, and electric field effects.  To be an effective shield against e-fields, generally requires the shield be continuous around the signal wire.  However, even braids below 95% coverage generally works very well at audio frequencies, but as the braid "holes" get to the physical size of the inner core dielectric, the e field shielding effect rapidly diminishes.
 
2.  Magnetic field shielding.  At audio frequencies, copper and aluminum are terrible shields.  Magnetic fields at lower frequencies easily penetrate normal outer conductor materials.  When an audio signal is being transferred (as it were) from a source to a sink via coax, most people think of the return currents to the source as travelling along the outer braid of the coax.  In a standalone situation where two devices do not have any common connections other than a single coaxial cable, this is true.  Unfortunately, if there are two coaxs being used to convey two channels, this no longer holds true.  At DC, the return current will flow back to the source based on the conductivity of the paths.  With two IC's, there are two shields connecting the units, therefore there will be two paths of equal resistance for the return current to take.  As the frequency rises, the loop inductance formed by the other's shield and the first IC will begin to increase the reactance of the alternate path, so more and more current will return via the IC which has the signal current flowing.  At very high frequencies, all of the return current will be via the shield of the coax which is "sending" the current. 
 
Since audio frequency magnetic fields will go through coaxial shields, cancellation of external magnetic field errors has to be via a different mechanism.  For a twisted pair, the two wires have what is called a "common centroid", a line in space where on average, all the signal current to the sink and all the signal current from the sink have a common geometric center.  As a result of this common centroid, EMF induced on both wires will be identical.  A balanced drive system relies on this common centroid construction to reduce signal errors between the two conductors, and by subtraction, removes the common mode error caused by the external magnetic field.  A coax also has a common centroid, this being the exact center of the wire.  The name "coax" stems from coaxial, or common center.  Again, externally coupled magnetic fields will induce the exact same EMF on the braid as well as the core wire. edit:this tenet breaks down if the coax goes through a very high magnetic field gradient such as the octupole field of a superconducting antimatter confinement trap, or the corner of an E core transformer running close to saturation.
 
 
When two coax IC's are used, problems arise.  Recall that at the lower frequencies, the current from one channel gets to the amp via the core of one coax, but it returns to the source via both coax shields.  This by design, creates a loop in space between the sent signal of the source, and half the return signal back.  This destroys the common centroid feature of the signal path that we were relying on to magnetically "shield" us from external magnetic fields.  Any time varying magnetic field that goes within this loop will generate a differential signal between the signal wire, and the common shield.
 
This problem becomes worse when we add in a pair of power cords with grounding conductors, as the ground path provides an additional way for the current to return to the source, so opens the loop even further.  And even WORSE that that, the power cords are  carrying currents with peaks from 10 to 20, 30 amperes with harmonics out the wazoo..power cords have magnetic fields,edit: and those fields are intimately coupled physically to the ground loop formed by the power cord ground.  If hot and neutral couple exactly equal and of opposite value to the ground loop, then power cord currents will not cause a problem.  A power cord which is asymmetrical in this regard can cause problems.
 
Systems with only two prong cords have far less of an issue at 60 hz in one respect in that the loop formed by the ground now has two transformer interwinding capacitances in series with the loop.  Ground loop currents at low frequencies will create voltage across these capacitive barriers, and the currents will be far lower.  This falls apart at higher frequencies as the capacitive reactance goes down, but if it's sufficiently low capacitance, the break frequency for the ground loop can be above the frequency where all the signal current flows back on it's own shield.
 
As for signal propagation along a T-line, please crack open Jackson, or Becker, Rojansky, or Shadowitz.  They each have relatively good sections you can learn from.
 
Cheers, jn
 
please note...I will only respond to coherent questions from you from now on.  If I do not respond to a post of yours, you'll know why...
 

 
 

 

[jnjn]

 
Quote:

Maybe I'm just dense, but if there's no audible ground loop hum and you're using cables and components of reasonable quality, then why worry further about EMC?
 
If something is wrong with the signal transmission to such an extent that the audio is noticeably degraded, most reasonable people will figure out a solution without having to buy into mad overpriced tweakery. I have a simple little dot tube amp which picked up interference and was susceptible to hum. It's like an ancient artifact, really. So I moved it a few feet away from all the other equipment and picked up an 8 dollar isolator and that was enough to address the supposedly staggering EMC concerns.
 
You have to remember that in a "Top Ten" magazine article there's not much space to go into all the manifold interpretations of physics. In the vast majority of cases "R, L, C" is all the customer has to grasp.
 
We're not talking about a peer-reviewed study posted in a scholarly journal. I'll take Aczel's cynicism over the self-deluded ramblings of 6moons anyday.

You are not dense.  It is an excellent question. Fortunately (or unfortunately), it requires a bit of explanation..
 
For a system problem to arise, you need a victim, an agressor, and a coupling mechanism.
 
The victim is simple enough.
 
The agressor can be the amplifier itself as a result of it's current draw, or it's leakage of it's output signal along the ground.  Or, it can be other current draws in the branch circuit causing magnetic fields which get trapped, or it could be a device nearby which creates a large pulse of magnetic energy on the wires that feed it, like an AC compressor during startup or turn off.
 
The coupling mechanism is the set of wires we use to power and connect the equipment. 
 
The coupling constant (as it were) will have a frequency dependence.  Faraday's law of inductance tells us that the voltage created around a closed loop of time varying magnetic field is proportional to the field's rate of change.  If you hear 60 hz, obviously the coupling constant is sufficiently high at 60 hz that it is audible.  But if it is not audible at 60 hz, that doesn't mean there is no coupling, merely that it is low enough at 60 hz.
 
If the primary mode of intrusion is just a simple induction law based coupling, the response will be frequency dependent. If the primary mode of intrusion is caused by an induced ground loop current....inducing another voltage into the system, you now have two frequency dependencies.  In other words, the system is now coupled as frequency squared.  In the first case, 600 hz will be coupled at a value ten times that of the 60 hz, in the second it will be 100 times the 60 hz value.
 
That is where my amp bit me.  Even though the system did not generally hum when turned on, as it began to draw power via it's line cords, the primary coupling introduces errors consistent with the pulses the power supplies draw from the line (huge levels of odd harmonics), and the second level of frequency squared coupling bit my crossover capacitors big time.
 
In general, if the system doesn't hum, we stop looking for problems.  Why fix it if it ain't broken. 
 
No hum does not necessarily mean no coupling however.  edit:  nor, does the existance of hum mean all the couplings exist.
 
In the vast majority of cases "R, L, C" is all the customer has to grasp.

Actually, they do have more.  Once you have a bit of understanding, it is actually possible to reduce any issues you find.  In one of my cases, some simple guidelines have served me well.
I had a setup where my source was 125 feet away from my amplifier.  Plugging the source in where it was needed caused all kinds of problems, as the building had tens of kilowatts of solid state dimmers and air conditioning.  The use of two 125 foot balanced cables reduced the problem a tad, but not enough.  Turned out that the pin 1 currents were still going into the amp and source, and the IR drop into the star grounds as well as secondary induction, still gave me problems.  So my solution was:  plug the source into the same outlet as the amp using a 125 foot long cord.  Switch the two balanced line runs into one unbalanced pair, done by taking a mike cable and running left channel on the blue wire, right channel on the white.  And, I wrapped this stereo unbalanced cable around the line cord, being careful not to match the twist pitch of the extension cord.  This resulted in dead quiet operation, and the channel separation degradation was of no consequence to the application.
 
Summary:  I ran the line levels with the power cord to reduce the loop size, twisted them to actually zero out any external driven fields by twisted pair cancellation, eliminated the dual paths caused by two shields, eliminated pin1 current issues. If I had a home audio system which I had hum problems with, the first thing I'd do is wrap the IC's around the power cords to remove the loops..
Quote:

The point being made about driving voltage down the center line (yes there's a wave front going down the return lines as well), was that the return current may be split into multiple different paths with certain cabling configurations.  If in fact the return current doesn't match the current being sent on one coaxial line, then you've got a potentially non-trivial problem.
 
What I don't see is how a potential problem will always impact proper operation in a non-trivial way.  Certainly there have been ABX tests with switchboxes where the configuration suffered no ill effects as a result of the cabling, switchbox internal structure, or any detail of how it was connected?

My primary point is that the inclusion of any device into the system must be considered for it's impact on system performance.  If it is "alleged" that two sets of IC's sound different, and that difference is caused by loop trapping effects, then a switching device must be fashioned such that it doesn't alter the loop trapping.
 
The allegation that listeners cannot hear a difference in IC's because it has been proven via the use of a switchbox cannot be supported when the differences are a result of a loop trapping mechanism which is drastically altered during the construction of the test setup..
 
It doesn't mean that there are always differences between IC's or power cords, just that the test configuration used to prove no difference may in fact be the reason no difference was detected.
 
Given what I know of EMC, and the fact that very few components have been designed with EMC understandings,I cannot support logically an absolute statement that "there are no differences" based on tests which do not even consider the mechanisms responsible for possible differences.  That would be akin to the drunk looking for his keys at the lightpost because he can see there, vs in the dark where he dropped them.
 
Cheers, jn

 

[JRG1990]

Do you have any tests or measurements that shows the effects of EMC on audio frequences or any audio gear?.

Also what about power cords that have the ground connected to the shield and the shield cut back disconnected at the equipment end does that have any effect on EMC, Also what about ferrites that add resistance at high frequences what affects do they have?.

Also aren't all audiable ground loop problems solved with a ground loop isolator?.

 

[jnjn]

BTW, that link provided as a "resource" by wakibaki...not quite correct.  This is what I mean by trusting all kinds of net sources, they may look good but not be completely accurate...and the casual, uneducated (on em field theory) peruser will not be able to discern the errors.
 
The statement ""Velocity factor is purely a factor of the insulating material's relative permittivity (otherwise known as its dielectric constant), defined as the ratio of a material's electric field permittivity to that of a pure vacuum. The velocity factor of any cable type -- coaxial or otherwise -- may be calculated quite simply by the following formula:""
 
followed by the equation"" v/c =1/ sqr(k)""  where k is the relative permittivity, is not accurate for either case cited.
 
Consider two types of transmission lines, constrained and unconstrained.
 
Constrained lines are lines where the magnetic and electric fields are rigidly controlled or constrained.  A coaxial cable is one such constrained cable, and in the limit, striplines can be considered as constrained as long as the width of the dielectric is more than an order of magnitude bigger than it's thickness.
 
Unconstrained lines are narrow striplines, and parallel conductors.
 
For a constrained transmission line, the actual equation is:
 
v/c =1/sqr (epsilon relative times mu relative).  Epsilon relative is the exact same thing they mention, relative permittivity.  Mu relative was missing, that is the relative permeability.  The assumption of the relative permeability being 1 is simply that, an assumption. 
 
The more important error is the assumption that their equation holds for a parallel conductor transmission line...IT DOES NOT, and never can.
 
The reason is the result of the conductors not constraining the magnetic nor electric fields.  There are NEVER cases where parallel wire transmission lines have the propagation velocity set by the dielectric permittivity. 
 
The correct expression is:  v = 1/sqr(LC)
 
In the limit expressed above for striplines of sufficiently high aspect ratio, the velocity can be assumed to be a result of the dielectric permittivity only.
 
For any t-line, the following expressions hold.
 
LC = 1034 EDC..
 
L = inductance per foot, expressed as nHenries per foot.
C = capacitance per foot, expressed as pf/foot
 
EDC is the effective dielectric constant and is unitless.  For constrained cables, it will be mu relative times epsilon relative.  For unconstrained cables, it will be higher than unity only.
 
The prop velocity can be defined as v = c/sqr(EDC).
 
Typical zips will have an EDC ranging from 2 to greater than 10, depending on how much magnetic field they "spray about" because of wide conductor spacings.
 
When perusing websites where they provide the L and C for their cables, simply using the LC=1034 EDC expression can tell you if the data is accurate..If EDC is below 1, the data suggests superluminal propagation velocities, which only happens in Star Trek.  Clearly, not possible.  If the EDC is huge, like 100, that also means errors.  Nodost did this on their website for the valhalla cables..they have C as 18pf per foot (IIRC), and L as 9600 nH per foot.  They state prop velocity as 96%.  If you go through the math, prop velocity calcs out as 10% C, obviously incorrect as well.  They typo'd L, it is actually 96 nH per foot.  When I duplicated their cable geometry, it measured as about 100 nH per foot.
 
Cheers, jn
 
ps.. I wish some forum would come up with a decent equation editor, it drives me nuts...
 
pps.  wakibaki, I have indeed learned something from you.  You have no problem with presenting links to net content as "proof" without vetting the source for accuracy.

 

[jnjn]

Quote:

Do you have any tests or measurements that shows the effects of EMC on audio frequences or any audio gear?.
Also what about power cords that have the ground connected to the shield and the shield cut back disconnected at the equipment end does that have any effect on EMC, Also what about ferrites that add resistance at high frequences what affects do they have?.
Also aren't all audiable ground loop problems solved with a ground loop isolator?.

I have the results I've mentioned.  Unfortunately, when I fixed my amp and source back in 1981, there was no internet to speak of, nor scopes or spectrum analyzers that could capture results for later distribution.  All I had were chicken scratched on papers.  Now, I just fix stuff by rote.  It always happens that the problems occur at the last minute before showtime, and I end up "mcgyvering" a loop fix..
 
Tom Van Doren provides several very good demonstrations of all of this when he gives his two day seminar.  In addition, it's somehow been made available on the web, if you google his name, you'll find a link to one of his workbooks...I do not know if it is an authorized copy however. If a search is not productive, I can find the link somewhere.
 
If you break the loop, you reduce the effect significantly.  Not completely of course, but it is quite effective.  Unfortunately, it also raises the problem of safety bonding, as it could reduce the ability to protect the user from unintended failures within equipment.  For an EMC related problem, ferrites will not be effective at all because they cannot block the actual ground loop current, they are significantly better for rf issues, which is not what I discuss.
 
Isolators do work, but again, I worry about safety bonded chassis and electrocution hazard.  A two prong listed piece gets around that by design, test, and certification, so is not an issue.
 
Cheers, jn