Well not exactly.
In a single coaxial cable and insulated connectors, the ground is the shield of the cable and it is connected to the central GND at some point. The cable then still behaves like a capacitor with the internal conductor being the one plate and the shield being the second plate of the capacitor.
In the case where two coaxial cables are used to transfer the positive and GND of the signal in their internal conductors and the shields are connected to ground, there will still be shunt capacitance on the coaxial transfering the positive but not in the coaxial transfering the GND (if someone could use the term "transfering"..)
In the refered case, two coaxials are used for each channel. The shields are left unconnected on both, and the positive and negative signals pass only through the internal conductors, pretty much like your parallel pair of your speaker cables.
The unconnected shields form a kind of shielding which reflects the radio waves like a microwave dish reflects (and focuses) the radio waves into the desired direction.
Basic capacitor theory applies here, nothing special. If you leave one plate of a capacitor unconnected, you do not have a capacitor!
Imagine it like having a single wire which resides (and does not touch) inside a water pipe, the water pipe not being connected to the GND. No RF passes into the wire.
Another way to imagine it is to imagine it would be to tie an insulated wire with a piece of aluminum sheet, the aluminum sheet not connected to ground.
The idea is sound, and therefor (sorry...) has been tried before. You can buy cables that have two twisted wires that carry the 'hot' and signal gnd, overall screened with a screen that is either connected on one side only or on both sides (in the first case, often an arrow on the cable indicates where the screen is connected to ground). In fact this is even better than two separate cables as is proposed: the twisting of the two wires keeps them close together so that any interference tends to be the same in both, so the effective interference (the difference between the two) is minimized.
Of course, it's only one more step to balanced cables....
jd
Of course, it's only one more step to balanced cables....
jd
I am not able to measure so low capacitances like the shunt capacitance of a low length coaxial, however basic capacitor theory shows that a capacitor with one plate left unconnected is not a capacitor.
To my taste, shunt capacitance introduced by cabling has little effect though and it is of considering only if the whole audio system has been designed in a perfectionist sense or if cable lengths are quite long.
I see what you mean,
The balanced technique is used in UTP computer networks too for noise cancelation, but in not op-amp based audio I think there is no way to cancel this noise even if the cables are twisted (not entirely sure).
On the other side twisted pairs carying +/- sig still introduce some capacitance, each conductor is the plate and the insulation being the dielectric. The capacitance is much lower than coaxial cables though.
I am talking about a cheap way to achieve highly desirable characteristics. There is no lower capacitance than two wires not touching each other.
The problem that I see with this is that if the shield of the coax is not connected to ground at either end it essentially becomes an antenna that is capacitive coupled to the center conductor. With the coax shields greater cross sectional area it will pick up more RF than a wire the same size as the center conductor and transfer the RF energy to the center conductor. This I think would make you more susceptible to RF than an unshielded 22awg wire.
Oh, and two wires not touching each other is basically the description of a capacitor.
(two conductors separated by a dialectric)
Oh, and two wires not touching each other is basically the description of a capacitor.
(two conductors separated by a dialectric)
Last edited:
> and two wires not touching each other is basically the description of a capacitor.
Yes of course, the capacitance is too low in air dielectric so it is not of much importance here.
I have never thought it that way, It is like having a transmitter microwave dish radiate backwards? It does not make sense to me if comparing it with a dish, but your acquisation is completely right for longer wavelength RF. The shield becomes an antenna which picks up RF and capacitivelly couples it to the internal conductor.
Although for microwave frequencies it may actually repel RF, as any metalic object acts like a mirror for microwaves either connected to ground or not. Also, this coupling capacitor behaves like a high pass filter which cutts-off completely the lower frequencies if the capacitance is too low.
Combined these two effects (cutt-off for low frequency RF, and mirror for microwaves) should give the desired result of moddest shielding, although one should use low capacitance coaxial cable to ensure higher cut-off.
Does it make sense?
I think for reasonable interconnects the cap is a non-issue. For 3 feet, at say 250pF/feet you're looking at 750pF. If the Zout of your driving side is 100Ohms, a common value, your freq response has a -3dB point at f=1/(2*pi*R*C)= about 200kHz. I for one wouldn't lose sleep over that.
jd
I would not lose any sleep over that.
But, cascade three of these cable filters over three different interconnects and you have ~-9dB at 200kHz and a low Q of ~0.4 and that extends the attenuation down into the accepted audio bandwidth.
If one of the Rs were 500r then the output would be 3dB down @ ~ 40kHz due to this single pole filter and a further 2dB to 3dB down due the the other two interconnects, i.e. ~-5dB at 40kHz and again of a low Q extending well down into the audio bandwidth.
But, cascade three of these cable filters over three different interconnects and you have ~-9dB at 200kHz and a low Q of ~0.4 and that extends the attenuation down into the accepted audio bandwidth.
If one of the Rs were 500r then the output would be 3dB down @ ~ 40kHz due to this single pole filter and a further 2dB to 3dB down due the the other two interconnects, i.e. ~-5dB at 40kHz and again of a low Q extending well down into the audio bandwidth.
If I get it right you are talking about LPF here?
I was talking about HPF formed by the capacitor between the UNCONNECTED shield and the central conductor, which cuts-off the lower RF frequencies. For microwaves I would not worry.
The audio is direct coupled srom the source to the sink.
when I looked at the proposed interconnect, I got the very distinct impression that the signal circuit was deliberately broken. That cannot be equated to direct coupled.
True. But 500 ohms Zout is already pathological in my book
Although some tube preamps are guilty here with very (relatively) high Zout. Interestingly, there is some indication that the earlier hi-freq roll off does give subjectively better mid range because the ear-brain will more focus on that range and pick up more details. That in turn *could* be a reason why some like tube preamps. But I have no hard proof for it.
jd
Maybe this is not clear enough.
The amplifier is direct coupled to the speakers and it is this design: http://www.audio-constructions.com/monstre/Hiraga-Monstre-Monster-Class-A-amplifier-schematic.png
The central conductor of one coaxial is the positive going from the amplifier to the speakers.
The central conductor of the other coaxial is the negative going from the amplifier to the speakers.
Both shields of the coaxials are left unconnected.
So we have no shunt capacitance from positive to groung (let's ignore the air dielectric etc)
But as a previous member states, the unconnected ground acts as an antenna. It collects RF and couples it to the central conductor through the dielectric of the coaxial. This is ANOTHER capacitor (not shunt)
If we use the standard coaxial way we have capacitor A
If we use the proposed coaxial way we do not have A but we have only capacitor B
But at microwave frequencies capacitor B is not formed because the shield reflects the RF even if it is not grounded, like a dish reflects microwaves.
At lower RF frequencies capacitor B couples RF energy to the cable, but It also behaves like a HPF, if the dielectric of the cable has low capacitance and the length of the cable is kept low. What is the cut off, I do not know..
Attachments
Hi,
are you saying his diagram that shows + signal and -signal are actually the signal flow and signal return of the signal circuit?
What about the loop area of the flow and return pair?
Surely, he is not proposing that you separate the signal flow from the signal return.
Look at microphone cable, twisted flow and return inside the common screen.
Look at balanced interconnects, the hot and cold (flow and return) are twisted together, sometimes without a screen.
Look at standard Coax, the signal flow and signal return are together concentrically.
Look at HF and a ground plane. The return route mimics the flow route to minimise impedance even though the return route is longer than the direct route straight across the conductive plane.
Why would all these standard methods go out the window, if they could be improved by separating the flow and return conductors resulting in big loop area?
are you saying his diagram that shows + signal and -signal are actually the signal flow and signal return of the signal circuit?
What about the loop area of the flow and return pair?
Surely, he is not proposing that you separate the signal flow from the signal return.
Look at microphone cable, twisted flow and return inside the common screen.
Look at balanced interconnects, the hot and cold (flow and return) are twisted together, sometimes without a screen.
Look at standard Coax, the signal flow and signal return are together concentrically.
Look at HF and a ground plane. The return route mimics the flow route to minimise impedance even though the return route is longer than the direct route straight across the conductive plane.
Why would all these standard methods go out the window, if they could be improved by separating the flow and return conductors resulting in big loop area?
Last edited:
This is not a techical answer, we are trying to improve things or at least to explore the possibilities.
Back to the technical, the cable lengths are the same, inductance is kept the same, large shunt capacitance does not exist (like two wires appart from each other, not twisted). So the point is the RF interference and how much the unconnected shield will reflect or couple RF energy to the internal conductor or not.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.