Ground question – finding the correct star ground point.

[SSLtech]

Two things here;

Can we PLEASE treat electrical safety ground (the power conection to the metal case) and audio/signal/0volt ground as two seperate things.

Even for unbalanced/asymmetric signals this can be done, gear is safe, it meets the single biggest UL requirement, and it shields quite effectively by doing so.

Audio ground NEED NOT BE CONNECTED. If if never is, the loops which wavebourn describes simply CANNOT happen.

I'll say that again.

They cannot happen unless they are connected. NO audio ground connection is required for electrical safety; only power ground.

Now, of course, linking them somewhere almost certainly does no harm.

But linking them in two (or usually more) pieces of connected gear is a recipe for hum. I'm fully familiar with the 'interlingual' situation described above. Yes, that's how I see it happening too.

But here we are discussing cutting safety connections in order to allow the maintained connection of the chassis to an AUDIO ground, when UL says it MUST be connected to electrical safety ground. -MADNESS!

Signals don't have to be unbalanced/asymmetrical to suffer from hum, so why make the distinction? -It implies that less care may be taken with balanced signals, where in fact while any induced hum may be lower, the SAME care sould be taken.

Certainly, transformers do help solve the problem even more.

But a thumbleful of prevention is worth a gallon of cure.

Keep Signal Grounds And Power/Electrical safety grounds separate. If you REALLY feel the need, provide a link or switch to bond the two.

That's it.

Now one other thing;

I'm slightly familiar with the life and major works of Anton Chekhov, but I must confess that the "clever likes to learn, stupid likes to teach" quote is new to me. -Where's it from, I'm curious...

Keith

 

[JohnRoberts]

[quote author="Larrchild"]But if you route the rf present on the incoming cables "around the outside" of the unit, whilst retaining good star ground hygiene inside, you are properly addressing both.

Keep the rf currents out of the unit's inside, and use a good audio ground scheme.[/quote]

Yup.. RF and AF are definitely different animals. We kind of beat the topic to death over on gearslutz several months ago with a thread about floating or grounding both ends of cable shields. I won't attempt to paraphrase the whole discussion but IMO for balanced interfaces it's a non-issue and for unbal, there is an approach AKA "Hybrid grounding" where the RF shields can be coupled via ceramic disc caps while floated at AF, so you have a best of both worlds scenario.

Larry, is absolutely right. Keep RF outside the box (at least the external RF). Another point worth mentioning, if you properly differential all of your signals you can pretty much ignore ground loops since the signal integrity is not affected. In the example I mentioned of the power amp using the chassis as a multiple point ground connection, the purity of this ground is not reliable. When you rack a power amp into a 19" EIA rack, who knows what kind of ground currents will flow between different chassis? Properly used it doesn't matter since the signals are treated as relative to their local grounds.

For SE consumer interfaces where a common ground connection is also an audio reference, ground purity is important, so YMMV.

JR

 

[Guest]

[quote author="SSLtech"]


Certainly, transformers do help solve the problem even more.
[/quote]

It is exactly my point. Or 12V power from something like BrickRack -- one cheap transformer per device, or multiple input/output transformers (expencive solution).

Keep Signal Grounds And Power/Electrical safety grounds separate. If you REALLY feel the need, provide a link or switch to bond the two.

That's it.

Great!
Now, the question is, where to connect that switch? Randomly, or in some certain well selected point?

I'm slightly familiar with the life and major works of Anton Chekhov, but I must confess that the "clever likes to learn, stupid likes to teach" quote is new to me. -Where's it from, I'm curious...

I don't remember. But I am certainly acting like a fool putting my ideas in public domain instead of patenting them to sell a production.

 

[Larrchild]

The good folks at UL and CE don't care if the contents of the box are patented. They just care if the green wire is grounded!

And I'm a big fan of the green wire on things that can kill you.
So it's simply how gear has to be designed to be sold.

So design with that as a prerequisite. Sorry.

 

[Guest]

[quote author="Larrchild"]\

And I'm a big fan of the green wire on things that can kill you.
[/quote]

Me too. That's why I am against a single randomly selected point because it can't be single by definition if to consider that green wire we have to connect without exceptions. And it does not matter, is it connected directly, or through a switch, anyway it can't be the single point anymore.

 

[JohnRoberts]

The "green" wire is a safety ground so by definition diffuse, subject to loops, and not a single point.

You pick wherever you desire to define your single audio ground reference.

JR

 

[cannikin]

IF you combine Audio Ground with Safety Ground you get wonderful Hum generator for a unit --- Check out my Federal Limiter Clones (this is with input attenuator @ zero) NICE!!!!!

Still trying to unravel the Ground Problems...... I'm sure poor design in regards placement of parts doesn't help ...... live and learn.

 

[Guest]

[quote author="RogerFoote"]I don't know, I am still a little leary of keeping the safety ground isolated form audio gnd (especially in tube gear).

What if there is a potential between them and you touch the audio gnd (through the shell of a connector) while touching the metal part of the chassis that is connected to safety ground? Faults can happen.[/quote]

That's why it has to be connected!

Now assume, you connect a mixer pult, power amp, graphical EQ, a notebook with A/D converter; all of them already have that green wire connectde, plus some of them have EMI filters containing capacitors between 120V and zero, middle point of them is connected to the safety ground, so all RF nasty noises go from 120V wire to the safety ground through the capacitor; all their chassis' are already connected together once through a power outlet.

Now, suppose your left leg randomly decided to connect to ground different points in each gear, in one closer to input, in another closer to output, in the third closer to whatever... As the result, one ground near the input is connected through that green wires to power supply ground in another peace of gear, etc... And current peaks from your notebook's switching power supply partially go to that wire, i.e. to input of some gear... Whistles, boiling eggs, etc...

Now, the recipe from Dr. Wavebourn: the best place to interconnect all grounds is as close as possible to the minimum volt/minimum dynamic resistance point in each peace of gear, and exactly this point connect to the chassis, better through the ground lift switch. This minimum volt/minimum dynamic resistance point is located in the power supply, on the leg of the final filtering capacitor, where feedback to all voltage regulators is taken from.

However, it is not necessary to follow this prescription; everyone has the right to reinvent own wheel...

 

[JohnRoberts]

The audio interfaces do not need to be bonded (shorted) to power ground but good practice suggests a reasonably low impedance between audio and power ground. Probably less than 100 ohms (with a cap in parallel for RF). This compliance between grounds can prevent loops in SE hi-fi type interfaces between multiple 3 wire safety grounded chassis

Back in '80s I also used back to back power diodes which would be open circuit for modest ground potentials but provide safety bonding above +/- 1/2V. I recall talking with UL about this and since the (power) diodes would routinely fail as a short circuit when overloaded they were even safe for bonding if sized to handle fault current without vaporizing. (as I recall UL was agreeable, but this was never formally UL approved, $$ to run full tests, etc).

JR

 

[Guest]

[quote author="JohnRoberts"] Probably less than 100 ohms (with a cap in parallel for RF). This compliance between grounds can prevent loops in SE hi-fi type interfaces between multiple 3 wire safety grounded chassis

[/quote]

Can you translate this into English, please? Loops through capacitors and resistors are not loops anymore? How it helps if hot 120V shorts accidentally with a signal ground? And what is "SE hi-fi type"?

Bob Cordell on DIY forum suggested 5 Ohm resistor with diodes in parallel. Indeed, 5 ohm is more than resistance of cables, but diodes can't save from 120V, they will burn like matches...

I vote for ground lift switches on all peaces of gear except a console, because a console has more ins/outs by definition. Powering from 12V will help a lot. In such case a power supply only will be connected to the safety ground, the rest will be interconnected through a signal ground only.

 

[Guest]

[quote author="RogerFoote"]Good case for external open frame PSU... Although if the PSU faults, theoretically you could get 120Vac in contact with DC terminals and carry that into the non safety grounded gear.[/quote]

No DC terminals! I use AC terminals and rectifiers/regulators in each peace of gear. Signal ground is well connected to the chassis in the mixer console without any possibility to lift the ground (no ground switch!), however a notebook wants this privilege, but I run it on batteries.

I use backward connected 12V transformers in my mic pres for B+ voltage.
In BrickRack each individual transformer has own fuse.

 

[JohnRoberts]

[quote author="Wavebourn"][quote author="JohnRoberts"] Probably less than 100 ohms (with a cap in parallel for RF). This compliance between grounds can prevent loops in SE hi-fi type interfaces between multiple 3 wire safety grounded chassis

[/quote]

Can you translate this into English, please? Loops through capacitors and resistors are not loops anymore? How it helps if hot 120V shorts accidentally with a signal ground? And what is "SE hi-fi type"?

Bob Cordell on DIY forum suggested 5 Ohm resistor with diodes in parallel. Indeed, 5 ohm is more than resistance of cables, but diodes can't save from 120V, they will burn like matches...

I vote for ground lift switches on all peaces of gear except a console, because a console has more ins/outs by definition. Powering from 12V will help a lot. In such case a power supply only will be connected to the safety ground, the rest will be interconnected through a signal ground only.[/quote]

I was speaking english, but I'll try slower and louder.

Instead of a dead short even a few ohms will prevent the heavy current flow in a shorted low impedance turn. 5 ohms would work, I can’t really comment on Cordell’s choice without knowing more about the specific application.

I addressed the size of the diodes in passing by suggesting power diodes. I actually used small bridges back in '80s (two pair in parallel) and they were adequate to hang until breakers tripped. Recall that diodes fail as a short circuit so dissipation is even less than while they were good diode junctions.

SE= single ended, i.e. unbalanced, 2-wire, etc.

hi-fi = nonprofessional, higher impedance, unbalanced, whatever.

The 60 Hz loop through a .1uF ceramic disc is not going to be much current, sorry if that wasn't obvious. Standard hybrid grounding technique to avoid LF loops.

JR

 

[Guest]

[quote author="JohnRoberts"]

I was speaking english, but I'll try slower and louder.

[/quote]

Thank you! I can hear you now! :grin:

[quote author="JohnRoberts"] 5 ohms would work, I can’t really comment on Cordell’s choice without knowing more about the specific application.
[/quote]

The specific application was connecting signal ground to chassis, i.e. to safety ground.

SE= single ended, i.e. unbalanced, 2-wire, etc.

Thanks, I thought of a standard usage of the term, i.e. opposed to PP (Push-Pull). Modern OpAmps have symmetrical Push-Pull outputs, but unbalanced ones (not SE!).

hi-fi = nonprofessional, higher impedance, unbalanced, whatever.

Lot of professional gear has unbalanced ins/outs, for example, inserts. So we have to deal with them. Also, some professiona gear has transformerless OpAmp balancing, so ceramic capacitors will be enough to make a noise...

The 60 Hz loop through a .1uF ceramic disc is not going to be much current, sorry if that wasn't obvious.

I meant RFI, sorry if it was'nt obvious.
It was a good practice in vintage receivers, though, because people connecting antenna did not bother to ground the chassis...

For audio couple of diodes will be better than a capacitor. But they must be powerful enough, I saw a lot of them popped like corn in a microwave.

 

[Fredrik]

[quote author="RogerFoote"]
Interesting discussion![/quote]

Yes, but it hasn´t helped me to make this any clearer quite the opposite actually. Sorry :?

I´m concerned about safety more than sonics here. What is a safe and a good rule of thumb that could be used?

Is this true?

1. Ground wire from mains power ALWAYS have to be connected to the chassis?

2. Signal ground i.e ground pin in audio sockets and the ground plane on the pcb must be connected to a common ground reference somewhere. If this is the same point as the power ground is what is discussed here?

In my DIY attempts I have a wire running from:

1) Power ground from from mains connector,
2) Ground pin on one of the audio connectors
3) Ground plane on the PCB

to the the same point at a screw terminal in the chassis.

Is this a correct and safe way of doing this?



:guinness: :sam:

 

[Guest]

1. Green ground wire must be connected to the chassis, as required by EL.
2. The point of minimal dynamic resistance between output of a power supply ground and powering voltages it supplies may be connected to the chassis. This point on the power supply is your common reference point for the star ground.
3. Connection of the signal ground (goto 2.) may go through a ground lift switch that will allow you to choose which point of your multiple pieces of a gear will be grounded.
4. You may use a 5-100 Ohm resistor with pair of powerful diodes (you may use ~ taps of a powerful bridge rectifier) to connect your common reference point to a chassis.

I.e. all chassis' will be interconnected through "green wires", and additional connections between common reference point swill harm minimally.

However, the weakest points still are a computer with switching power supply & asymmetrical inputs/outputs and a mixer pult with asymmetrical inserts. Some modern gear with cheap opamp based balanced ins/outs may also represent weak points...

However, I don't idealize transformer balanced inputs since transformers have some interwinding capacitances...

 

[JohnRoberts]

[quote author="Fredrik"]

Yes, but it hasn´t helped me to make this any clearer quite the opposite actually. Sorry :?

I´m concerned about safety more than sonics here. What is a safe and a good rule of thumb that could be used?

Is this true?

1. Ground wire from mains power ALWAYS have to be connected to the chassis?
[/quote]

It is good practice with metal chassis solidly connected to wall ground. The goal is to provide a low impedance path so any fault condition will cause mains breaker to open and protect the user. If the transformer is dual insulated (like typical consumer gear), grounded chassis isn't required, but still good practice, especially if dealing with tube or high power amplifiers with elevated mains voltages.

2. Signal ground i.e ground pin in audio sockets and the ground plane on the pcb must be connected to a common ground reference somewhere. If this is the same point as the power ground is what is discussed here?

My understanding of law is that signal or audio grounds do not have to be bonded to mains ground "unless" the panel markings says that it is "ground" or "earth". It is probably good practice to provide a safety path to mains ground when interfacing with other gear since they may have a fault that dumps dangerous current into your audio ground. Never assume the other guy did it right. Even a UL approved design, plugged in with a cheater plug to lift the ground, could fail and dump fault current into your audio ground if connected.

In my DIY attempts I have a wire running from:

1) Power ground from from mains connector,
2) Ground pin on one of the audio connectors
3) Ground plane on the PCB

to the the same point at a screw terminal in the chassis.

Is this a correct and safe way of doing this?
:guinness: :sam:

Sounds safe assuming adequate gauge wire. :guinness: Whether it will sound quiet depends on the details of how audio interfaces are configured. :?:

JR

 

[pepe]

Ground wire directly connected to the chassis is to be considered as the basic safety rule number one. I don’t want to hear what else to do with this green wire :shock: .

To me it seems logical that when the device has many in/outs simply join the earth ground and circuit ground at only one place, also preferable through a ground lift to give the option. This involves star grounding all Pin1 on all xlr connections! Many seem to recommend the use of a passive device during this linking, at least a resistor in the range of 10-100 ohm. I assume this will isolate the ground noise to get into the audio path, but wouldn’t the use of a shottky diode be a better choice here? Then you have total isolation as long as no higher voltages comes creeping in the shield.

But life isn’t simple because…
Another recommendation (from many sources) is that pin 1 on all inputs xlr should go directly to the chassi, while pin 1 on the outputs should go to a toggle switch then to star ground. Should this alternative only be considered when building a micpre :? ?

 

[JohnRoberts]

Pin one can be hard connected to the mains ground if desired since the audio signal is extracted from pin 2 relative to pin 3. Pin 1 should be connected to a low impedance path to ground to sink away any stray noise currents the shield may pick up. You want to be careful that these noise currents don't contaminate any important audio paths as they are leaving the area.

Voltage on or developed across pin ones will have a secondary effect due to coupling through the shield into pins 2 and 3, but proper wire wrap for good CMR, balanced impedance for similar +/- pickup, and differential processing by receiver circuitry should minimize noise pickup.

JR

 

[cannikin]

[quote author="pepe"]
But life isn’t simple because…
Another recommendation (from many sources) is that pin 1 on all inputs xlr should go directly to the chassi, while pin 1 on the outputs should go to a toggle switch then to star ground. Should this alternative only be considered when building a micpre :? ?[/quote]

this has been recommended by a few people -- while I was trying to figure out my 60Hz problem --

"I believe that your problem is that you may have a ground loop. Audio components should be referenced to a ground buss, like a number 10 bare wire. Check to make sure that there is only one ground reference and that the reference buss is grounded as close to the input transformer as possible. (ONLY ONE GROUND INSIDE THE CHASSIS).

If you have say an output connector that is grounded to the chassis (pin 1 connected to the chassis instead of the reference point), you will have a nice antenna that picks up 60 cycle AC."