Should a balanced interconnect be shielded as well (shielded vs twisted pair)

[EB1000]

Hi

I've build many 125ASX2 based DIY, using Ghent's pre-made interconnects (BTL mode, feed from a balanced output).

Ghent cables are shielded using two sperate shielded cables for the positive and negative signals of the balanced connection. Ghent told me this is the best approach.

However, I had a discussion with Guntars Smits, which is another DIY class D supplier and builder, and he's using non-shielded single braided cable, in which the positive, negative, and the two ground are all twisted together.

He claims that this approach is much better for balanced cables, as you don't need shielding due to high noise immunity of balanced, and by eliminating the shielding, you reduce capacitance and thus increase bandwidth.

So which of is correct? Ghent or Guntars?

I've tried both, and could not tell the difference in sound using my own ears. I wonder what measurements will reveal

This is the video Guntars posted. You can see that all 4 wires (two grounds, pos and neg) are bundled and twisted together, while, Ghent solution is using two shielded cables

https://www.ghentaudio.com/cable/125asx2pro.html

What do you think?

Thanks

 

[Speedskater]

The has been some success in the pro audio field using unshielded Cat5 cable.
But CAt5 is vary, vary different from non-shielded single braided cable, in which the positive, negative, and the two ground are all twisted together.
On the other hand, balanced XLR interconnects are a very robust system. For a 1 or 2 meter interconnect almost any cable will work.

 

[mansr]

Ghent cables are shielded using two sperate shielded cables for the positive and negative signals of the balanced connection. Ghent told me this is the best approach.

That's the _worst_ way of making a balanced connection. The proper way is twisted positive/negative signal wires surrounded by a shield.

 

[Blumlein 88]

That's the _worst_ way of making a balanced connection. The proper way is twisted positive/negative signal wires surrounded by a shield.

Have you listened to it done both ways? AH! I thought not.

In one method you improve the CMRR with tight twisting. Star quad helps even more. In the other you reduce the CMRR somewhat by shielding the twisted wires from each other so the fields can't cancel out. Maybe too much CMRR is like too much negative feedback. You have to listen and read the advertising instructions to know for sure.

/sarc

If you want a better explanation as to why mansr is correct:
https://www.svconline.com/news/cmrr-balanced-interfaces-part-2-369515

 

[BDWoody]

Have you listened to it done both ways? AH! I thought not.
......

/sarc

Holy crap...

I thought my head was going to explode for a few seconds.

 

[DonH56]

Balanced lines' rejection bandwidth depends upon the twisting and spacing of the twisted pairs. The shield provides additional protection from EMI/RFI and additionally provides a separate (chassis) ground path for reducing externally impinging noise upon the signal conductors. Ideally the shield creates a (quasi) Faraday cage protecting the signal conductors within.

Or something like that, long, long time since my EM classes...

 

[andyc56]

Balanced XLR is about as good as it gets, provided proper grounding technique is observed in all the components in the system. The theory and practice of this were established by Neil Muncy in his classic 1995 article, which was later put into the AES48 standard. This article by Bruno Putzeys explains it further.

 

[DonH56]

Balanced XLR is about as good as it gets, provided proper grounding technique is observed in all the components in the system. The theory and practice of this were established by Neil Muncy in his classic 1995 article, which was later put into the AES48 standard. This article by Bruno Putzeys explains it further.

Yah... Part of the problem is a lack of understanding what "balanced" means and how return currents flow (if they even understand current operates in loops). Another is the huge body of XLR circuits called "balanced" that range from fully-differential to quasi-differential to impedance-matched (usually resistive) with common-mode rejection ranging from "decent" (20-40 dB) to almost nil (6 dB or less).

And I know you know all this, ranting... - Don

 

[andyc56]

Yah... Part of the problem is a lack of understanding what "balanced" means and how return currents flow (if they even understand current operates in loops). Another is the huge body of XLR circuits called "balanced" that range from fully-differential to quasi-differential to impedance-matched (usually resistive) with common-mode rejection ranging from "decent" (20-40 dB) to almost nil (6 dB or less).

One of the more interesting aspects of the Muncy article is how little the CMRR matters to hum rejection.

 

[EB1000]

The guy from the video (Guntars) told me that his connections are very short, thus no shielding is required, and longer shielded cabled will add capacitance, which will reduce the bandwidth. Ghent told me that for BTL, he can make me a shielded cable in which both pos/neg conductors are inside the same shield.

Also, have anyone noticed that for balanced headphones cables. the most expensive ones are braided and not shielded? So I'm still not sure what is the best balanced cable configuration. Twist or twist and shield? Also, do you ground the shield on sending/receiving end, or both?

Each article gives a different suggestion...

Thanks

 

[mansr]

The guy from the video (Guntars) told me that his connections are very short, thus no shielding is required, and longer shielded cabled will add capacitance, which will reduce the bandwidth.

Nonsense. Sure, a shield increases the capacitance a little, but that doesn't matter. Cat 8 Ethernet cable uses shielded twisted pair and has a bandwidth of several GHz.

Also, have anyone noticed that for balanced headphones cables. the most expensive ones are braided and not shielded?

I suspect aesthetics is a major factor in headphone cable design. The amount of variation seen only goes to show that it doesn't really matter much.

Twist or twist and shield?

Shielding keeps a bit more noise out. Whether it's necessary depends on the application.

Also, do you ground the shield on sending/receiving end, or both?

Both, unless you have a specific reason to do otherwise.

 

[Speedskater]

The guy from the video (Guntars) told me that his connections are very short, thus no shielding is required, and longer shielded cabled will add capacitance, which will reduce the bandwidth.

Pros often use 100 meter (300 foot) XLR cables (and sometimes much longer) at on location sites with no capacitance problems.

 

[DonH56]

One of the more interesting aspects of the Muncy article is how little the CMRR matters to hum rejection.

View attachment 111641

IME most "hum" problems stemmed from ground loops, or poor grounding (return path) execution in general that was outside the CMRR path, which matches that completely.

The other thing about CMRR is it tends to reduce very, very quickly with frequency. The cable itself (shielded twisted-pairs) improves HF (EMI/RFI) common-mode rejection, but the circuits themselves (drivers/receivers) CMRR usually falls off well below 100 kHz.

 

[DonH56]

Mogami quad-star cable is around 35 pF/ft to the shield and 24 AWG (0.04 ohms/ft). Given a preamp with 600-ohm output impedance (on the high side) and 100 k-ohm amplifier input impedance, here is a quick simulation showing 500 feet of cable yields about 450 kHz of bandwidth considering just one wire to the shield. This is not a full simulation with distributed RC and all that jazz, but seems like it should not be an issue...

HTH - Don

 

[pma]

Real life preamp CMR vs. frequency (measured), simplified schematics attached.
OP, yes, use shielded twisted pair.