Triaxial interconnects

Hi,

I've just bought some silver plated triaxial cables and planning to make some interconnects with it.

BELDEN 7784AS CABLE Red

The triaxial cable is made out of a solid core, a guard braid and a an outer braided shield. The solid core and guard is made out of silver plated copper.

Now the question I have is this,

1) If I want to make them into an interconnect, what is the best way to do it.

Possibility
a) The guard is connected to the preamp "live" wire. However only the core carries the signal through to the amplifier. In which case the The guard and centre core is at the same potential and takes away the effect of the dieletric insulator. Outer shield is used as a ground.

b) The guard is used to carry the ground return, the outer shield is connected to the ground of the pre amp only, the return current is not passed through the outer shield.

c) The guard is used to carry the ground return, the outer shield is connected to the ground of the power amp only, the return current is not passed through the outer shield.

Thanking you all in advance for your feedback.

Oon
 
Last edited:
The interconnect must use two conductors to carry the Flow current from the source and carry the Return current back to the source.
If you omit one of these conductors your interconnect CANNOT work.

Don't use the word ground, it just leads to more confusion.

EVERY connection between electronic modules is a two wire system. Flow and Return or Hot and Cold.
Design all your connections to meet that requirement.
Then look at your proposal to see if any reference voltages need to be made between modules. The sch should identify any missing voltage reference connections.
Finally look at how you can safely assemble your mains powered equipment and design in (before you start building) those last few SAFETY connections.
 
Last edited:
Option A. But drive the inner shield through a resistor around 10K. That way you don't load the output driver too much. You can go as low as 2.2K. But shouldn't need to.
Option A connecting the screens/shields to ground breaks the return current route and breaks the connection.
It will not work, unless the amplifier finds an alternative route to return current to the source via some circuitous route that may include a remote star ground and/or chassis connection.

It really does look like both you and Simon have read this and not understood how circuits work
However only the core carries the signal through to the amplifier.
This is wrong.
I'll repeat again
EVERY connection between electronic modules is a two wire system. Flow and Return or Hot and Cold.
 
Last edited:
oon_the_kid said:
a) The guard is connected to the preamp "live" wire. However only the core carries the signal through to the amplifier. In which case the The guard and centre core is at the same potential and takes away the effect of the dieletric insulator. Outer shield is used as a ground.
You still have dielectric between the inner shield and outer shield. All you have done is increase cable capacitance, which is the opposite of what you want to do if you are worried about dielectric. Using a resistor, as simon7000 suggests, does not change this but merely adds some phase shift at the HF end.

Given that cable (a waste of money in my opinion) I would use (b) or (c).

AndrewT said:
Option A connecting the screens/shields to ground breaks the return current route and breaks the connection.
I think he means use the outer shield as the signal return conductor.
 
If a shield/screen is only connected at one end. Then it should be the send end.
Bill Whitlock did a paper on this subject.

But if the send end is not powered (like some mics) then it should be connected at both ends.
If the shield/screen is part of a 3wire connection and is ONLY an interference shield, then it can be connected directly at the send end.
Optionally it can be capacitively connected at the receive end, or directly connected at the receive end.

If the shield/screen is part of a 2wire connection, then it is the Return route for the signal current and it MUST be connected to the Audio return pads/pins at the audio PCBs at BOTH ends.
 
Hi Andrew T,

Not to worry I understand exactly the need for the return current.

The idea of going triaxial is it will have a core and 2 shields.

My question is which is the best way to use the shield. Triaxial is not all that new and actually used in interconnects. See this for instance.

Van Den Hul - 3T The Mountain Hybrid (halogen free) | Product (cable)

It is also an established technique for sending video signals over long distances. (Which is what those cables were originally for)

In the previous lab where I used to work it is used for low current measurement (pA, fA)

I am trying to figure out which is the best configuration for audio interconnects.


Thanks.

Oon
 
You still have dielectric between the inner shield and outer shield. All you have done is increase cable capacitance, which is the opposite of what you want to do if you are worried about dielectric. Using a resistor, as simon7000 suggests, does not change this but merely adds some phase shift at the HF end.

Given that cable (a waste of money in my opinion) I would use (b) or (c).


I think he means use the outer shield as the signal return conductor.

Actually, the way it works is the center core is connected to both the transmitting end and receiving end. The guard only to the transmitting end. The guard and core are both at the same voltage therefore the signal that is travelling through the center conductor does not see the insulator surrounding it. Normally this is done with a buffer. In this case, the pre amplifier will see the capacitance of the cable but the signal carrying core does not see any capacitance.

This is how it is done for semiconductor measurement with an additional buffer. I just wonder if this is the way it might be done by audio interconnect companies as well.

Power Systems Design (PSD) empowers global innovation for the power electronic design engineering community.

Oon
 
Option A. But drive the inner shield through a resistor around 10K. That way you don't load the output driver too much. You can go as low as 2.2K. But shouldn't need to.

Hi Simon,

You seem to have experience in this matter, would like to hear more from you.;)

Adding the resistor to reduce the effect of the capacitance makes sense because the capacitance will be much higher between the guard and outer shield. Normal co-ax is about 50-100pF/m, and the surface area is much smaller, I would not be surprised if the capacitance is as high as 1nF.

Oon
 
oon_the_kid said:
Actually, the way it works is the center core is connected to both the transmitting end and receiving end. The guard only to the transmitting end. The guard and core are both at the same voltage therefore the signal that is travelling through the center conductor does not see the insulator surrounding it. Normally this is done with a buffer. In this case, the pre amplifier will see the capacitance of the cable but the signal carrying core does not see any capacitance.
The signal on the guard will see whatever dielectric there is between it and the outer shield. This is likely to have much greater capacitance than between the inner core and the guard. Therefore you have made worse the very thing you believe you are making better. The source will see this capacitance, as it is driving it.

Adding a resistor as Simon suggests removes this load, but then renders the guard less useful due to phase shift at higher frequencies. To the extent that you 'hide' the guard-shield capacitance you nullify the effect of the guard.
 
The signal on the guard will see whatever dielectric there is between it and the outer shield. This is likely to have much greater capacitance than between the inner core and the guard. Therefore you have made worse the very thing you believe you are making better. The source will see this capacitance, as it is driving it.

Adding a resistor as Simon suggests removes this load, but then renders the guard less useful due to phase shift at higher frequencies. To the extent that you 'hide' the guard-shield capacitance you nullify the effect of the guard.

Care to calculate the phase shift from 10,000 ohms and 30 pF/ft at 20,000 hertz? And then use that to estimate the increased load on the center conductor?

Assuming 100 ohms source impedance and 10,000 ohms load, about .2%. A bit less than if the shield was directly around the center conductor.

Even at the 70 pF I normally see on concentric shielded cable it stay less than 1%. And of course if you are custom building cables they are often shorter than 6 feet.

However even using a driven shield from a second buffer makes a trivial difference in the signal transmission.
 
If the 10k resistor has no effect, then it has no purpose. If it is small compared with cable reactance then it doesn't materially affect the load on the source so no point in having it there. If it is large compared with the cable reactance then it will introduce phase shift so the guard acts less like a guard. So you choose: ineffectual guard or ineffectual resistor?

I agree that having the guard makes almost no difference anyway. As I said, I would not buy triaxial cable for an audio interconnect. The OP is trying to solve a non-problem.