GND Scheme questions

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

spoontex

Well-known member
Joined
Jun 14, 2019
Messages
193
Location
Spain
Hello,


I've some questions about a good GND scheme. I read a few articles and I didn't see any consent about it. So, I draw two schemes. Wich one is better?


Scheme 1.
each XLR cable has its shielded cable. And chassis to pin 1. IEC earth anywhere bolt on the chassis, near the IEC connector.

Scheme 2.
Only first input have the shield cable to the pin 1. Pin 1 distribute shield to the others connectors, rest of the XLR cable only have shield at one end. Chassi and IEC earth soldered to pin 1 of input connector.

I read that the earth connection of the IEC socket need to be near to input power, bolt on chassis. Some others says that the earth connection need to be soldered at pin 1 of the XLR input connector.

Any suggests?
 

Attachments

  • gnd scheme2.jpg
    gnd scheme2.jpg
    137.9 KB · Views: 2
All the grounds of the circuit are connected at one
single point to the chassis. The PE gets its own connection to the chassis, according to the rules in your country.
 
Neither. The shielding of all XLR inputs have to go the shortest way connected to the chassis. It is not good to lead shields coming from the outside through the inside of the chassis. This can lead to RF contamination.
Which mean each an every XLR pin one needs to go to chassis right at the connector.

Cheers

Ian
 
I've some questions about a good GND scheme

You will be less confused when you understand the reason for various connections, and the requirements that derive from that. The first step is to not use the term "GND" because it is ambiguous and will just lead to confused thinking.

First step is proper protective earth (PE) connection for the power entry. That connection has one and only one function: protect you from electrocution in the event of power line connection to the chassis. As such there must be a short and direct connection between power entry protective earth and your chassis. Solder is not permissible, because it can melt in a high current fault event and allow the connection to separate. Connections should be crimped or held with screws. Most jurisdictions require a toothed nut or other locking style connector which can "bite" into the metal and hold a firm connection without vibrating loose.
To reiterate, this connection is protecting you from electrocution, so not negotiable. There are very specific requirements for when a chassis can have no earth connection at all, involving insulation strength of the high voltage components and how many layers of insulation you have, but again that involves design which can result in potential electrocution hazard, so you need to know for sure what you are doing before attempting a non-earthed power design.

The next step is shield connection of the cables. The shields of the cables are to provide protection from electrostatic fields, so to provide that function to as high a frequency as practical the shield connection of the connectors, i.e. pin 1 of XLR connectors, the sleeve of phone plug connectors, the shell of BNC, etc. should have as short and direct a connection to the chassis as possible.

A secondary result of connecting the shields directly to the chassis is that chassis-to-chassis leakage currents now flow on the shields. Any connection has resistance, which results in any current flow through that connection generating a voltage. That voltage will primarily be related to power line noise, so you need to make sure that there is no path for that noise through the circuit reference node. Inappropriate connection that allows the shield currents to flow on the circuit reference is referred to as the "pin 1 problem" from a June 1995 Journal of the Audio Engineering Society paper which popularized that term.

So the terms I have highlighted, "protective earth," "shield connection," and "reference node" all have very specific functions, but people will commonly refer to any or all of those as "ground." Without keeping the relevant functions separated in your mind as you plan a design, you can end up with only tangentially related items connected together in ways that limit the audio performance.

Bill Whitlock (MisterCMRR in these forums) has provided access to a very good tutorial on this subject. It is well worth the time to read through:
Whitlock grounding and shielding seminar

[Edit for posterity: I found some clarification on the "solder is not permitted" claim I made earlier in this post, which is not strictly correct without caveats. What you should not do is rely on solder alone to connect a wire to a chassis, there should be a bolt and wire terminal, rivet, etc. that physically secures the earth wire to the chassis, and if the wire connection to the power inlet or terminal is soldered, the wire should be hooked or wrapped through the terminal lead before soldering so that if the solder does begin to melt the wire cannot fall away.]
 
Last edited:
I need to wiring this way, and it's a bit confusing for me. It's correct the scheme that I draw?

The Audio In / Out that I need.

The jacks are isolated.

The output doesn't worry me, because is straight to XLR outs.
 

Attachments

  • sc_audio_balanced.pdf
    16.6 KB · Views: 0
  • gnd scheme4.jpg
    gnd scheme4.jpg
    94.4 KB · Views: 2
It's correct the scheme that I draw?

No. Cable shields are still cable shields on outputs and on TRS jacks. Actually you have TS jacks, but I would recommend TRS jacks, and just have tip connect to the same place as XLR pin 2, and ring to the same place as XLR pin 3, and sleeve connects to chassis, XLR pin 1 connects to chassis. Cable shield connections do not connect directly to your PCB.
 
Last edited:
No. Cable shields are still cable shields on outputs and on TRS jacks. Actually you have TS jacks, but I would recommend TRS jacks, and just have tip connect to the same place as XLR pin 2, and ring to the same place as XLR pin 3, and sleeve connects to chassis, XLR pin 1 connects to chassis. Cable shield connections do not connect directly to your PCB.
Yes, but If you see the original schematic it says: Balanced XLR and unbalanced 1/4".
 
I hate ground loops. I would tie all the shield grounds to 1 chassis lug. If you must ground the chassis thru the power plug I've done it through 2 6A1 diodes, in parallel anodes to cathodes in parallel with a 10K resistor. Anything above .6 volts in either direction will cause the diodes to conduct in either direction long enough to trip a breaker in case of a short. Below .6 volts the diodes are off and the 10k resister will bleed off minor ground currents essentially limiting the current that can flow between grounds. This was done by Acoustat in their direct coupled tube amplifier/electrostatic speaker panels.

Of course your studio should be properly wired with grounds intact.
 

Attachments

  • Screen Shot 2022-11-30 at 6.06.14 PM.jpg
    Screen Shot 2022-11-30 at 6.06.14 PM.jpg
    38.4 KB · Views: 3
the diodes to conduct in either direction long enough to trip a breaker in case of a short

In general silicon is a faster fuse than thermal or magnetic breakers. I do not believe you could get a diode isolated design UL listed. It is much better to design your input stages properly to reject common mode noise than to possibly risk your life with non-standard PE connections.
 
Since you probably need a ground from circuit board to chassis, bring that from the ground in the power supply if it's on the board.

You will probably have some AC leakage from the power transformer. If it's chassis mounted then it's all ready connected to chassis. If it has a interwinding shield that should go to the IEC ground lug. You may need to experiment to see what cause and/or cures a problem.
 
Little (1A) silicon will go first and will short. The 6A1 peak current is 400 amps for 1/2 sine wave. Big silicon 25 amp diodes will survive. They have very high peak current. If your fuse/breaker doesn't pop you need to have one that will. I tested my circuit as a direct short on a 20A breaker. Worked as directed. The diodes even survived.

But it's only a suggestion for annoying ground loop problems. No arm twisting.
 
Little (1A) silicon will go first and will short. The 6A1 peak current is 400 amps for 1/2 sine wave. Big silicon 25 amp diodes will survive. They have very high peak current. If your fuse/breaker doesn't pop you need to have one that will. I tested my circuit as a direct short on a 20A breaker. Worked as directed. The diodes even survived.

But it's only a suggestion for annoying ground loop problems. No arm twisting.
here's a post from 2013 on the subject To lift pin 1 or not I did a bunch of work on this back in the 80s/90s. At Peavey I momentarily considered getting UL to test and approve this but using proper pin 1 design discipline makes it unnecessary (and UL would have cost a few tens of $k).

UL safety ground bonding, tests for voltage rise with several tens of amps fault current. While its true that diodes will generally fail as a short circuit .....BUT small 1A diodes will vaporize (open circuit) shortly after melting. My solution back then was to use several amp, diode bridges. The extra mass from the epoxy bridge package stops them from vaporizing long enough to trip the breakers (melted silicon is a low resistance conductor).

The good news is that none of this is necessary with good input stage design and proper ground path routing.

JR
 
Great. A 25 A diode bridge will work fine. This diode has 100 A repetitive current rating: VS-HFA25PB60-N3. How fast does a typical breaker take to open a dead short? I saw 6 milliseconds as a general rule.

If the current is lower but above the breaker rating it may take minutes to open but that has nothing to do with the diodes. Without them it will take just as long.

Despite everyone's best efforts to avoid ground loops in their studio and designs they do occur because we use different AC runs to the breaker box with different leakage currents, like from a Fender ground switch, old vintage gear or poorly designed stuff that creates a current. Or it's simple: the nearest outlet will do.

Lighting strikes will cook a lot of gear in a flash. The idea is safety in case of a short shorts happen, rare but they do.

Usually people solve the hum problem with a 3 to 2 prong plug adapter with no ground. I'm ok with my design. No problems so far. The 10k resistor makes it work.
 
Last edited:
You can't always trust the grounds in house or studio wiring or on the power pole. Open neutral on the power pole burned out some of my GF's appliances. No longer 120-0-120, every thing was now in series and not balanced. The power company paid for the damages but my cheap GF didn't pay me for my work. No longer my GF.
 
Great. A 25 A diode bridge will work fine. This diode has 100 A repetitive current rating: VS-HFA25PB60-N3. How fast does a typical breaker take to open a dead short? I saw 6 milliseconds as a general rule.
its been decades but as I recall it was low-mid single digit amp bridges... the benefit was the larger thermal mass that prevented the molten silicon from evaporating and conducting long enough for the fuse/breaker to trip.
If the current is lower but above the breaker rating it may take minutes to open but that has nothing to do with the diodes. Without them it will take just as long.
a shorted diode is very low resistance so the current will be limited by the service.
Despite everyone's best efforts to avoid ground loops in their studio and designs they do occur because we use different AC runs to the breaker box with different leakage currents, like from a Fender ground switch, old vintage gear or poorly designed stuff that creates a current. Or it's simple: the nearest outlet will do.

Lighting strikes will cook a lot of gear in a flash. The idea is safety in case of a short shorts happen, rare but they do.

Usually people solve the hum problem with a 3 to 2 prong plug adapter with no ground. I'm ok with my design. No problems so far. The 10k resistor makes it work.
I still do not condone using ground lifts with three wire line cords. UL uses different safety standards for two wire line cords (double insulated primary wiring) vs. three wire SKUs that rely upon a safety ground bonded chassis.

JR
 
This diode has 100 A repetitive current rating

Whitlock's presentation referenced fault currents from 150A to 1000A, and trip times from under 10 ms to 2.5 s, noting that "the quoted range of fault currents and trip times is based on a UL study (circa 2003-2005) of over 1,000 residential 15A and 20A outlets."

I would rather teach people how to make designs which are not sensitive to leakage currents than to bodge in schemes that might not create a dangerous fault if the device happens to hold up outside its specified use long enough.
 
No. Cable shields are still cable shields on outputs and on TRS jacks. Actually you have TS jacks, but I would recommend TRS jacks, and just have tip connect to the same place as XLR pin 2, and ring to the same place as XLR pin 3, and sleeve connects to chassis, XLR pin 1 connects to chassis. Cable shield connections do not connect directly to your PCB.
I think that I get.

So, every connector, XLR, jack... needs his own connection to chassis, right? I can't tie all the shields together and bolt on chassis, no?

And the circuit GND needs to tie chassis no? It's correct if a use a metal stand off that contact the GND circuit to chassis? Or where is the best place to tie the GND circuit?
 
Back
Top