• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Balanced interconnection and ground

Davide

Addicted to Fun and Learning
Forum Donor
Joined
Jul 6, 2020
Messages
655
Likes
314
Location
Europe
Hi everyone.

I'm modifying the case of my amplifiers to make it more compliant with safety and signal integrity requirements.

However, a doubt has arisen that I can't dispel (I'm not a great connoisseur of electronics).

Bruno Putzeys wrote a paper for Hypex about the famous ground.
The summary for the interconnections is practically the following:

Screenshot_20240120_084156_Drive.jpg


Now. It's not clear to me where those two points I circled in yellow should be connected. I'm guessing at the case, somewhere away from where the screen was connected.
But looking at the pinout of the Hypex NCxxxMP modules, does it mean that it is the GND pin that has to go there?

Screenshot_20240120_084408_Drive.jpg


Because if so, then I don't understand why some amplifier on the market hasn't such a connection (seeing photos here and there).
Worse, some seem not to even connect pin 1 to the chassis. Look at Audiophonics.

audiophonics-mpa-s250nc-xlr-power-amplifier-class-d-stereo-ncore-nc252mp-2x250w-4-ohm.jpg


The three xlr conductors go to the board, therefore assuming that pin 1 (shield) is connected to the GND pin of the amplifier. Which is the common mistake Bruno talks about in the document. Calling GND the pin 1 leads to this error. In reality the differential signal does not need to conduct the ground, the signal is the potential difference between hot and cold, there are no other references.



Can anyone help me clarify?
 
Last edited:
It's generally referred to as The Pin 1 Problem, and many manufacturers get it wrong. Pin 1 should be seen as an extension of THE CHASSIS, not Audio ground. For reasons of cheapness, many manufacturers do what Audiophonics have done, and hope to get away with it.

S
 
It's generally referred to as The Pin 1 Problem, and many manufacturers get it wrong. Pin 1 should be seen as an extension of THE CHASSIS, not Audio ground. For reasons of cheapness, many manufacturers do what Audiophonics have done, and hope to get away with it.

S
Ok, but where I have to connect the GND pin of the amp?
 
Best results I have measured in a chain (and not standalone) is when the forth pin is tied directly to chassis (and same with the cables,internally) .
Some don't have that pin but these are usually the bare ones (not anodized,painted,etc) where you can put it on one of the bolts.

For example,this one has it:

4th pin.PNG
 
Ok, but where I have to connect the GND pin of the amp?
I would connect it back to a star point where the incoming power ground connects to the chassis.

Though it depends on the design of the board. Whatever happens, you should only have ONE connection back to ground. If for example the mains ground were connected at the power supply end of the board - and that were internally connected to the amp section ground, you would not then want a separate ground from the amp ground pin back to chassis.


What does the board installation documentation say about grounding?
 
In the case of the NCxxxMP, to the chassis.
In any specific location?
And then, both channels in the same place?
Also, inside the chassis, does the xlr signal no longer need a screen?
 
In any specific location?
And then, both channels in the same place?
Also, inside the chassis, does the xlr signal no longer need a screen?
As close to the XLR as possible. I use one of the mounting bolts to take a short wire to Pin 1.
Each XLR can be taken to its own mounting bolt, but yes, if they're close together, then they can be connected to the same place.
Correct. Once inside the chassis, it doesn't need a screen as long as the two wires are closely twisted, but inside a Class D amplifier, especially one with a switch-mode power supply, there's a lot of RF about, so having a screened lead is a Good Thing, which should be attached to Audio Ground on the board, not to chassis.

S.
 
There is no such thing as an "audio ground" on a NCxxxMP.
So it is correct to use the shield of internal xlr cable to connect "GND" pin (J4.3 of previous image) to chassis, at any point (not necessarily near the earthing point of chassis)?
Because @sergeauckland says a different thing.
 
There is no such thing as an "audio ground" on a NCxxxMP.
By that I meant to connect to the grounding on the board, which I assumed was also where the audio ground reference is, rather than attach the cable screen to the chassis.

In other words, connect XLR socket pin 1 to chassis only. Connect XLR Pins 2 &3 to the audio cable inners, and leave the cable screen unconnected at the XLR end, and only connected at the amplifier board end.

S.
 
If we stop using the lower case 'ground' a lot of these misunderstanding go away.
* * * * * * * * * *
Better words:
chassis (metal)
shield/screen
DC supply common
analog/audio circuit common
digital circuit common
Safety Ground/Protective Earth (EGC)
Planet Earth
* * * * * * * * * *
yes all of these labels have continuity to each other, but each has its own tasks.
 
Last edited:
By that I meant to connect to the grounding on the board, which I assumed was also where the audio ground reference is, rather than attach the cable screen to the chassis.

In other words, connect XLR socket pin 1 to chassis only. Connect XLR Pins 2 &3 to the audio cable inners, and leave the cable screen unconnected at the XLR end, and only connected at the amplifier board end.

S.
If you leave the screen unconnected on one end, it's loosing it's function.
 
If you leave the screen unconnected on one end, it's loosing it's function.
It's an electrostatic screen only, and for a balanced circuit, doesn't really have much function anyway. The amplifier's common-mode rejection and the twisting of the two signal wires should in all but the most extreme cases be sufficient. I've seen whole radio stations wired with UTP cable with no hum or other interference problems at all.

In this case, if using a pre-made cable with a Molex or similar connector at one end, and if XLR Pin1 is correctly attached directly to the chassis, only the two signal wires need to be connected to XLR pins 2 & 3.

S.
 
It's an electrostatic screen only, and for a balanced circuit, doesn't really have much function anyway. The amplifier's common-mode rejection and the twisting of the two signal wires should in all but the most extreme cases be sufficient. I've seen whole radio stations wired with UTP cable with no hum or other interference problems at all.

In this case, if using a pre-made cable with a Molex or similar connector at one end, and if XLR Pin1 is correctly attached directly to the chassis, only the two signal wires need to be connected to XLR pins 2 & 3.

S.
It's an electrostatic screen only, and for a balanced circuit, doesn't really have much function anyway. The amplifier's common-mode rejection and the twisting of the two signal wires should in all but the most extreme cases be sufficient. I've seen whole radio stations wired with UTP cable with no hum or other interference problems at all.

In this case, if using a pre-made cable with a Molex or similar connector at one end, and if XLR Pin1 is correctly attached directly to the chassis, only the two signal wires need to be connected to XLR pins 2 & 3.

S.
While I agree on the theory of everything you write,
- the NCxxxMP board ground should be connected to chassis to avoid nasty noises
- the signal connector is mapped very close from the NTC on mains, twisting has unfortunately not a lot of effect here.
So better connect the shield on both sides.
 
The three xlr conductors go to the board, therefore assuming that pin 1 (shield) is connected to the GND pin of the amplifier. Which is the common mistake Bruno talks about in the document. Calling GND the pin 1 leads to this error. In reality the differential signal does not need to conduct the ground, the signal is the potential difference between hot and cold, there are no other references.



Can anyone help me clarify?
It's possible what you see here is pin 1 being connected to the chassis directly by the socket - Neutrik and others sell XLR sockets with a sharp point that punches into the chassis to make the shortest possible connection between Pin 1 and the enclosure.

And then, as Boxem says, it would make sense for the GND pin on the NCxxxMP input to be connected to the shield (which is also connected to Pin 1, and hence the chassis), since it needs to be connected to a ground somewhere, and the cable is already there to do it.

The NCxxxMP modules don't use a safety earth connection (they only have live & neutral AC power inputs). So there is a separate question of whether you connect the chassis to safety earth or not. For DIY it's probably not a good idea to leave it disconnected unless you are very experienced or using a non-conductive enclosure (but then you have the Pin 1 problem again but in different form)
 
Back
Top Bottom