[Speedskater]

Well "DBT" is Double Blind Test.  There are many different "DBT" protocols.  The basic meaning is that neither the Listener nor the Procter know what model/unit is being tested.  So tests are done using automated switch and in others it's done by a technician who is hidden (sight & sound) from the Listener and the Procter.

[Speedskater]

The most sensitive "DBT" is the "AB/X" test.  In this test, the Listener has to decide whether "X" is "A" or "B".  In a well run "AB/X" test, they use very short sound samples and almost instantaneous switching.  But the two big problems are: it's hard to deal with all the uncontrolled variables and it's way too sensitive!  Listeners can correctly identify "X", but the difference is so small that it's meaningless preference wise.

[kinku]

Again a difficult task in real life. Thanks speedskater.

[jneutron]

jn,

Your insight on the insulation thickness is an interesting insight. Somehow I am lost on the significance of 2.7.

That is the relative dielectric constant.  When making a coax for example, the equation LC = 1034 DC will be exact, with DC being the relative dielectric constant of the insulation material between the core wire and the shield.  I've used 2.7 for pvc, kapton, ptfe.

For a single pair of wires, the equation LC = 1034 (EDC) is used, where EDC is the effective dielectric constant.  It will be higher because it also accounts for the magnetic field that is outside the wires.  As the number of twisted pairs goes up, the EDC will asymptotically approach DC.

jn

[jneutron]

John,it is hard.There is science everywhere you have to find it.. So using PTFE does not reduce the capacitance too much I guess ,except if it is lower in thickness than PVC.

As the insulation gets thinner, the inductance will go down and the capacitance will rise.

jn

[cheap-Jack]

As the insulation gets thinner, the inductance will go down and the capacitance will rise.jn

How come? 

Let's understand what is inductance first.

Please read the statement taken from a long established cable manufacturer's 'cook book':-

"When current flows in a wire, it creates a magnetic field about the wire, which generates voltages along the same wire as the current changes. These opposing voltages act to limit the rate at which the current can change. This effect is termed as inductance & is measured in units called henrys.

The self inductance of a round straight copper wire is on the order of 0.4 micro-henry/ft & is relatively UNaffected by the diameter or length of the wire.

The self inductance of twisted pairs of wires is on the order of 0.08 micro-henry/ft; while the mutual inductance of a coaxial construction is 0.14 Lg10 (D/d) micro-henrys/ft."

The insulation material jacketing a round wire will reduce the propagation velocity(v.c.) of a signal wave to travel along the wire.

Taking vacuum as 1, still air is 0.95 (i.e.bare wire, NO insulation), foamed PE 0.8, Teflon will be 0.7, PE is 0.69, PVC is 0.5.
We can see PVC, such a common insulation material used in most electronic hookup wires, is low low down the v.c. list.

The thickness of the insulation is to safeguard the maximum voltage of the current flowing thru the wire against leakage.

So please tell us how the "insulation" thickness of a wire goes down with its inductance?

c-J

[jneutron]

How come? 

That is very simple.  First, I'll address all the points that are in error.

Let's understand what is inductance first.

Please read the statement taken from a long established cable manufacturer's 'cook book':-

"When current flows in a wire, it creates a magnetic field about the wire, which generates voltages along the same wire as the current changes. These opposing voltages act to limit the rate at which the current can change. This effect is termed as inductance & is measured in units called henrys.

Close enough description for a layman.  However, much of what you've pasted is inaccurate.  Knowing who this source is to correct their errors would be good.

More rigorous definition of inductance is: the relationship between the current flowing through a system, and the magnetic energy that is stored in that system.

The equation is E = 1/2 L I(squared).

The self inductance of a round straight copper wire is on the order of 0.4 micro-henry/ft & is relatively UNaffected by the diameter or length of the wire.

The terms used can get one into trouble here.  I'll explain.  The inductance of a wire is in two parts, that within the conductor, and that external.

The internal inductance of a cylindrical wire (the total magnetic energy stored within the conductive material) is 15 nanohenries per foot.  This is independent of the diameter of the wire for uniform current distribution, and will change only if the current is time varying.

The external inductance of a wire is absolutely dependent on where the return current is.  Current can only flow in a loop, and how it returns will govern the system inductance.  There is NO SUCH THING as the external inductance of a conductor if the return path is not specified.

The self inductance of twisted pairs of wires is on the order of 0.08 micro-henry/ft; while the mutual inductance of a coaxial construction is 0.14 Lg10 (D/d) micro-henrys/ft."

That is off by almost a factor of 2.  The minimal inductance of a wire pair results when the conductors are almost touching each other, and that number is approximately 150 nH per foot if I recall correctly, it's been a few decades..  The correct number can be derived by using the Terman equation, as it is comprised of the internal inductance, the spacing based inductance, and a correction term for loops which do not have a high aspect ratio (there is enhancement when the wire loop is wide in comparison to it's length.

As to the statement about coax.... There is no mutual inductance between a coax core and it's shield.  This is a simple result of geometry.  In fact, there is no internal magnetic field caused by current within a shield.  The inductance of a coax is the sum of the internal inductance of the core (15 nH per foot) and the integrated magnetic field from the core wire surface to the inner surface of the shield conductor.  Note that it is a truncated integral, it does not extend to infinity as a bare wire would.  This is a consequence of the braid's external field being equal to and opposite the core's field in the space outside the braid.  Also note that this cancellation only works if the currents are equal and opposite, and the core is concentric to the braid, placing the current centroids on the same line in space.

The insulation material jacketing a round wire will reduce the propagation velocity(v.c.) of a signal wave to travel along the wire.

That is in essence almost correct.  The actual prop velocity will be 1/sqr(LC).  Another form which is applicable to free space is 1/sqr(epsilon mu).  For coax, it is 1/sqr(DC), DC being the relative dielectric constant..

By varying the dielectric constant of a zip pair's insulation (and keeping thickness constant), the velocity will indeed be inversly proportional to the dielectric constant of the insulation.  However, since the external inductance of a wire pair is also increased by the spacing of the conductors, the velocity will in fact be slower than what the simple equation would predict.

The thickness of the insulation is to safeguard the maximum voltage of the current flowing thru the wire against leakage.
So please tell us how the "insulation" thickness of a wire goes down with its inductance?

c-J

Now for the simple answer.( Note: you reversed cause and effect for some reason.  The insulation thickness changes the inductance, not the other way around.)
Teflon typically has a higher dielectric withstanding capability than other plastics.  As a result, for the same voltage rating, it will be thinner.  Kapton is even better than teflon, it can withstand 6 kV per .001 inch.

When the insulation is thinner, the wires in a twisted pair will be closer together.  Since the external inductance of a wire PAIR is entirely dependent on the spacing between the two current carrying conductors, when they are closer together, the inductance will go down.

Where did you get that erroneous information, who is that cable manufacturer, and who wrote the "book"?  I assume that if they are a large cable manufacturer, they would like to know they have bad content.  Geeze, I hope it's not Steve from belden, I haven't been in contact with him in years..

jn

ps.  I'll admit that some of the concepts I present are outside the scope of most of the readers here, but just ask questions and I'll be happy to explain.

[Speedskater]

See what happens when you ask a person who's full time job in a science lab, is getting more & more electricity through smaller & smaller wires, faster & faster.

"So please tell us how the "insulation" thickness of a wire goes down with its inductance?"

Short answer:  That's backwards,  it's as the insulation gets thinner the inductance goes down.

But really it's not the insulation getting thinner, it's the two conductors getting closer together.

[jneutron]

See what happens when you ask a person who's full time job in a science lab, is getting more & more electricity through smaller & smaller wires, faster & faster.

"So please tell us how the "insulation" thickness of a wire goes down with its inductance?"

Short answer:  That's backwards,  it's as the insulation gets thinner the inductance goes down.

But really it's not the insulation getting thinner, it's the two conductors getting closer together.

Cmon...I did say that.  It's all in there...somewhaaare....

This ain't rocket science...I know, cause I have talked to rocket scientists..  They're weird..not like us injuneers...we's is normal people..

jn

[Speedskater]

I know all about not being a 'rocket scientist', I saw that bit on "Big Bang Theory".

[jneutron]

I know all about not being a 'rocket scientist', I saw that bit on "Big Bang Theory".

I really don't watch that show.

I'm not sure if it's because I really despise how technically oriented people are depicted in prime time, or if it's because it is too close to the mark.
jn

[andy_c]

For coax, [the velocity] is 1/sqr(DC), DC being the relative dielectric constant..

Really?

[jneutron]

Really?

Um, besides yes, what is the question?

You do understand it's normalized to lightspeed, yes?  If the coax dielectric is vacuum, prop velocity is 1.  If the dielectric constant of the insulator is 4, then the prop velocity will be .5 lightspeed.

And, it is only for propagation of signals which have a voltage and current relationship consistent with the characteristic impedance of the cable.

jn

[andy_c]

So you give somebody a big lecture about propagation velocity and provide three examples, two of which are absolute and one of which is relative to light.

Not too rigorous if you ask me.

[jneutron]

So you give somebody a big lecture about propagation velocity and provide three examples, two of which are absolute and one of which is relative to light.

Not too rigorous if you ask me.

If you can't follow the discussion, just ask. If all you are capable of doing is nitpicking trivial details, go for it.  I'm sure I spelled some words incorrectly somewhere as well, and grammar probs while you're at it.

Otherwise, you have a problem I cannot assist you with.

jn

ps..you call two sentences a big lecture?  Sheesh..
pps. next day edit:  CJ is to be commended on bringing content into the discussion.  While there are errors in that content, that is not CJ's fault.  He pointed to conflicting information from a presumed reliable source, and it is good fodder for discussion nonetheless.  There are many sources out there which have conflicting information, and can be difficult to figure out which is good.

As to discussion of velocity in terms of C or in terms of absolute, each equation has it's strengths and pitfalls.  Using 1/sqr(LC) carries with it the problem of units and orders of magnitude.  Capacitance is in pf, which is 10e-12, inductance is uH, or  10e-6, velocity is then in the foot/ns level (10e-9).  Using 1/sqr(mu epsilon) has the same type of issues...4 pi 10e-7 henries per meter, 8.854 10e-12 farads per meter, 3 10e+8 meters/sec.. exponents out the wazoo.

Using 1/sqr(Dielectric coefficient(relative)) uses numbers like "4" or "9", with results like "1/2" or "1/3rd".  And it is consistent with the units CJ used (he said taking vacuum speed as 1).

Converting the first two to the units being discussed, while possible, does not simplify the context.  It buries the understanding deep in equations.  I am glad CJ used the units he did.  And I am suprised anybody would choose that aspect as a bludgeon.

[rich2ch]

JN,

I am kind of curious about how some of my old cables would stack up. Is there an easy way to measure the characteristic impedance of a cable?

Thanks,

Rich

[rich2ch]

JN,

If I were going to make a cable with a characteristic impedance of 6 using cat 5 cable would I need 8 runs of cable or 4? I think 8 is correct but I am not sure.

Rich

[jneutron]

JN,

I am kind of curious about how some of my old cables would stack up. Is there an easy way to measure the characteristic impedance of a cable?

Thanks,

Rich

If you measure L and C of the wire, you can use z = sqr(L/C).  Accuracy can suffer because the L is very low for typical cables, and it is not easy to measure at that level correctly.  Even if you can correct all the L measurement errors, you still have to contend with how the meter actually calculates inductance, many meters are different.

C is relatively easy to measure, nature of the beast.

What you may end up with is a reasonable value of the high frequency impepdance of the wire, but no indication of what the impedance will be at audio frequencies.  That will be anywhere from 2 to perhaps 6 times the hf impedance, and that is generally due to the distributed R of the cable, which L and C meters will tend to compensate for.

As a result of the error possibilities, I recommend using simple approximations. and the equation LC = 1034 EDC.

Measure C in pf.

Calculate L in nH assuming the EDC consistent with the wire geometry and typical plastics...anywhere from roughly 3 to 5 for most zips.

Then Z = sqr(L/C).  Watch the units though, that is where lots of errors occur.

Zips generally run 100 to 150 hf Z, and 250 to 400 lf Z.

JN,

If I were going to make a cable with a characteristic impedance of 6 using cat 5 cable would I need 8 runs of cable or 4? I think 8 is correct but I am not sure.

Rich

each twisted pair is 100 ohms.  Assuming you connect all the stripes together as one leg , and all the solids together as another leg, the impedance will be 100/n, with n being the number of twisted pairs.  One cat5e will have n=4, Z = 25.  8 runs of cat5e is 32 pairs, so z = 100/32 or about 3.  4 runs is 16 pair, 100/16 =6.25.

Just beware of oscillation if your amp is a high bandwidth one, you might need a zobel.

jn

[cheap-Jack]

Hi.

(1) The external inductance of a wire is absolutely dependent on where the return current is.  Current can only flow in a loop, and how it returns will govern the system inductance.  There is NO SUCH THING as the external inductance of a conductor if the return path is not specified.

(2) As to the statement about coax.... There is no mutual inductance between a coax core and it's shield.  This is a simple result of geometry. .. This is a consequence of the braid's external field being equal to and opposite the core's field in the space outside the braid.  Also note that this cancellation only works if the currents are equal and opposite, and the core is concentric to the braid, placing the current centroids on the same line in space.
That is in essence almost correct. 

jn

(1) So you stated a conductor also gets "external inductance" besides its self inductance or "internal inductance" as you quoted. Yet the external inductance is "absolutely dependant" on where the current returns.

The ideal situation is like a coax where the return current of equal value flows back thru its braid/foil shield & complete cancellation of mutual inductance is possible.

But in realworld situation for non-coax cables where the return cable can be placed far away from the incoming signal cable.
Then the usefulness of the cable external inductance is no long valid as no way the external inductance can interact each other in a distance. So why mention it vs its self inductance, it being so vulnerable to the cable positioning?

Even in the ideal situation of a tightly twisted pair of cable, like Cat 6 LAN cable. The return path of the return signal is frequency selective. Physical layout of the device can affect substantially the return pattern of the return signal currents. 
For lower frequencies, the return signal will go for return loop of minimum resistance, ie. the chassis. Only for certain HF & above, the return current will follow the loop of minimum inductance, which would be the shield of the cable.

(2) Such complete mutual inductance on a coax also 100% depends on the symmetric geometry of the centre conductor & its shield.

However in realworld coax cable building, it really impossible to extrude a coax with absolutely coaxial geometry. Plus any true
coaxial cable will be deformed on actual field laying, making complete mutual inductance cancellation impossible. Therefore
mutual inductance always exist in a coax.

c-J

[jneutron]

Hi.
(1) So you stated a conductor also gets "external inductance" besides its self inductance or "internal inductance" as you quoted. Yet the external inductance is "absolutely dependant" on where the current returns.

The ideal situation is like a coax where the return current of equal value flows back thru its braid/foil shield & complete cancellation of mutual inductance is possible.

As I stated, there is no mutual inductance between a core and it's shield.  It is necessary to be careful with terms here, they can be confusing. 

The core wire will create a magnetic field which falls off as 1/R outside the conductor.  The shield will also create a field which falls off as 1/R outside the braid.  Both falloff "R's" are measured from the current centroid, which is identical for a coax.  As a consequence, from the braid out, the fields will exactly cancel.  This leaves only the 1/R field of the core, and only between the core outer surface and the braid inner surface.

But in realworld situation for non-coax cables where the return cable can be placed far away from the incoming signal cable.
Then the usefulness of the cable external inductance is no long valid as no way the external inductance can interact each other in a distance. So why mention it vs its self inductance, it being so vulnerable to the cable positioning?

What is considered "distant" is not so much.  One foot, one meter, ten feet?  Regardless of spacings of this level, the loop will still create a magnetic field, and will still have an inductance based on the distance.  This is trivially proven, and the power companies have to use this for their HV transmission runs even though the MV lines may be 100 feet apart.

Even in the ideal situation of a tightly twisted pair of cable, like Cat 6 LAN cable. The return path of the return signal is frequency selective. Physical layout of the device can affect substantially the return pattern of the return signal currents.

Actually, no.  the size of the pairs, the twist pitch, and the jacket thickness conspire to make the cable relatively immune to physical layout effects.  If the cable were sensitive to what you say, we'd never be able to run it in conduit or cable trays.  Needless to say, we do.

For lower frequencies, the return signal will go for return loop of minimum resistance, ie. the chassis. Only for certain HF & above, the return current will follow the loop of minimum inductance, which would be the shield of the cable.

An accurate statement.  Somewhat out of context, but accurate nonetheless.

However in realworld coax cable building, it really impossible to extrude a coax with absolutely coaxial geometry. Plus any true
coaxial cable will be deformed on actual field laying, making complete mutual inductance cancellation impossible. Therefore
mutual inductance always exist in a coax.

All manufacturers of heliax or coax specify a minimum bend radius.  As you correctly state, deformation of the coaxial symmetry can indeed occur if the cable is overbent.  Users typically worry about variations of cable impedance resulting from such deformation, and can be measured via TDR for quality checks.  I worry about that AND the splashout of magnetic field caused by the failure of total and absolute magnetic cancellation.  Not for the cable users mind you, but for all the other users and systems sharing the cable trays and conduits.  When dealing with nanovolts, kiloamperes, and ghz frequencies, (not all in the same cable of course), I have to worry about interference.

For coax, it is not quite as bad as discussed at rf frequencies.  Proximity effects will tend to force the braid currents to redistribute around the braid to reduce the centroid errors.  As a result, rf coax will be less inclined to splash as a result of deformations.

jn