As Scott correctly pointed out with an example, this is indeed incorrect.
A coaxial cable by design will have zero external magnetic field beyond the braid, this because the core field and shield field are exactly opposite outside the shield. As inductance is the relationship between the current in the system and the amount of energy stored in the magnetic field, having no external magfield will indeed appear as if there is no inductance. However, one cannot neglect the magnetic field energy that is contained between the core and shield, nor within the core wire itself (15 nH per foot at LF).
Between the core and the shield, there is both E field energy stored within the dielectric's capacitance, and M field energy stored within the dielectric region within the permeability of the dielectric (typically free space).
The relationship between the inductance and capacitance of a coax cable can be expressed as:
L(nh/ft) times C(pf per foot) =1034 times the dielectric coefficient.
In Scott's example, plugging in the L and C, the DC is 1.6. As most plastics have a DC in the 3-5 range, this cable is obviously a foamed dielectric, as foamed can have a DC as low as 1.05. The problem with getting too close to 1 is that the dielectric will lose physical strength, so the cable becomes too delicate.
To find the inductance of a normal coax, just measure the capacitance per foot, use a DC of 3, and calculate inductance.
Just a note...
When the load end of the coax is terminated by the characteristic impedance of that cable, the capacitive energy stored is exactly equal to the magnetic energy stored.
John
No. The state of zero external field is a consequence of the core current being equal to and opposite the shield current. If there is a parasitic path from the load to the source which allows current to return not through the cable, then that current will cause external magnetic field around the coax and around the sneak path.
For example, an amp and preamp, both grounded 3 prong, with a coax sending signal to the amp, will have a sneak path back to the preamp via the safety ground. If there are two coax (left and right), there are two sneak paths for each signal. At LF, the line cord ground will dominate. As frequency climbs, the line cord ground path (which has a higher loop inductance) becomes less of a player, and the other channel shield becomes half the return path. Only when the shield to shield loop inductance exceeds the coax impedance will the signal current return only via the same coax.
A current clamp put over one of the coax cables will confirm that at LF, the coax will have a magnetic field external.
This is what a ground loop is all about.
The termination impedance determines the balance between the capacitive energy storage in the cable and the inductive storage within the cable.
When the load impedance is below the cable impedance, there is more inductive storage in the cable than capacitive storage.
John
For example, an amp and preamp, both grounded 3 prong, with a coax sending signal to the amp, will have a sneak path back to the preamp via the safety ground. If there are two coax (left and right), there are two sneak paths for each signal. At LF, the line cord ground will dominate. As frequency climbs, the line cord ground path (which has a higher loop inductance) becomes less of a player, and the other channel shield becomes half the return path. Only when the shield to shield loop inductance exceeds the coax impedance will the signal current return only via the same coax.
A current clamp put over one of the coax cables will confirm that at LF, the coax will have a magnetic field external.
This is what a ground loop is all about.
The termination impedance determines the balance between the capacitive energy storage in the cable and the inductive storage within the cable.
When the load impedance is below the cable impedance, there is more inductive storage in the cable than capacitive storage.
John
I incorrectly confounded zero external field with zero inductance, what I should have said is "in theory coaxial cables have zero external magnetic field".
Thank you for your explanation of the splitting of screen return currents through sneak paths, which is helpful in the visualisation of the cause of earth loops.
I started using as short as possible, large cross-section multistrand wire for my common returns/grounding - where the power supply common is to be connected to the chassis ground. I paid strict attention to ensuring large contact areas between, for example, a large round lug and chassis bottom plate. I used crimping first, followed by soldering the crimped area as well, with 80W soldering iron. Would you say that this is better than crimping alone?
By the way, the improvement in stereo separation, instrument positioning and better, more natural harmonics was unbelievable, all that from this simple upgrade, that is almost always completely neglected.
Also, the input IEC mains RFI filter has a metal case surrounding it, which when mounted on the backplate ensures very large contact. Both are aluminium.
I pretty much started thinking along the same lines of properly earthing large AC VSD's to combat a large amount of noise and harmonics...
Thank you very much, extremely useful. I had to read your post few times until it sank in, but it does make sense 100%.
Just a question: should the text I highlighted in bold be saying this:
"...and the coax shields (both channels; each channel using one coax cable) become half the return path"
Thank you.
Nick
Last edited:
Coax made for RF applications and interconnects is typically not suitable for speaker connections because of either the mechanic properties (too stiff) or electrical properties (too small a centre conductor). However there have been manufacturers of coaxial cable suitable for speakers.
OCOS (Optimal Connection System) was a Swiss manufactured coaxial speaker cable favoured by Dynaudio, who at one stage even manufactured their speaker systems with a coaxial connector for OCOS.
Eurocable made 02N25C coaxial 2 x 2.5 sqmm speaker cable of which I still have a roll. It's the only 14 AWG speaker cable that will fit into the 2-pin DIN speaker connector used by Bang & Olufsen, for example.
Canare GS6, which is an instrument cable, is also suitable for speakers and is great for systems that use RCA or ¼" jacks for speaker connectors.
Star-quad (four conductor) speaker cables give much of the benefit of coaxial cable when configured correctly. In fact the star-quad cable construction was first devised for RF at the dawn of radio when it had not yet been possible to manufacture coaxial transmission cables.
There is a RF-Type of coax, which is indeed relatvely stiff, but the diameters of the conductors will work for the typical current for speakers....
RG213(copper)
or
RG214(silver)
And as the mechanical integrity is helpful for sound quality, as you said, johnmath, i would say the stiffer the better....
RG213(copper)
or
RG214(silver)
And as the mechanical integrity is helpful for sound quality, as you said, johnmath, i would say the stiffer the better....
Mechanical integrity matters; conductors that are not constrained move under the influence of electromotive force of currents flowing within them.
There is a lot of enthusiasm for RG11 as a speaker cable amongst some quarters in Australia so I made up a pair for a blind comparison. I could easily pick them and didn't like them much compared to star-quad, but I don't have any particular theory as to why. RG8 being 50Ω with a much larger centre conductor than a 75Ω cable with the same overall diameter would likely be a better choice for a speaker cable than RG11.
Any speaker cable appraisal is going to be affected by choice of amplifier and speaker, of course. I have occasionally found combinations where star-quad isn't the best choice.
Last edited:
I recommend exactly the opposite: the almost complete elimination of cable capacitance by using of single speaker wires - go to post #1110 under
https://www.diyaudio.com/community/...ake-any-difference.73632/page-56#post-7004937