Grounding 101

[CJ]

i see no plate choke, that circuit is therefore disqualified, 

 

[earthsled]

Does anyone have recommendations for grounding tube shields and transformer shields on a PCB? Should they be tied to the chassis, or should they go to 0V of the audio circuit?

Thanks!

 

[abbey road d enfer]

Does anyone have recommendations for grounding tube shields and transformer shields on a PCB? Should they be tied to the chassis, or should they go to 0V of the audio circuit?

Thanks!

Ideally, they should go to the chassis ground, but in fact in most cases they can go to any reasonably stiff reference node, meaning they could go to a cathode decoupled to ground, they could go to B+ (would be silly because' of risks), they could go to DC heaters voltage... connecting them to 0v is OK, except if there's heavy HF energy in the box; then, the metal parts, acting like antennas, would discharge the energy in the 0v node, which is not a nice thing to do.

 

[MagnetoSound]

There's stuff here that really should be put in the Meta pages, imho.

 

[jstark]

There's stuff here that really should be put in the Meta pages, imho.

Seriously.  This thread has done so much to tie-together and illuminate the fragments of what I've learned about grounding.  Between this, the Rane docs, and the excellent Bill Whitlock article about balanced interfaces, I finally feel like I have my head around proper grounding.

Thanks everyone.


peace.
david.

 

[dmp]

Going back to the dual ground v3 gif - I would be interested to understand better the flow of current and how that might create noise. This seems to me to be the necessary building block of information to create a quiet PSU.
The sketch makes it seem that a common ground for the rectifier, filter caps, and regulator will create quiet power rails. But isn't it true that all the current returns to the transformer center tap, which is at the opposite end of the ground flow from the audio ground return? This seems to me to be problematic, as the ground plane (say it is a copper wire) between audio ground and the transformer CT will be fluctuating slightly depending on the current going through it? If the ground reference for the regulators is fluctuating slightly (i.e. noise!) then the power rail will have some noise in it, yes?

 

[abbey road d enfer]

Going back to the dual ground v3 gif - I would be interested to understand better the flow of current and how that might create noise. This seems to me to be the necessary building block of information to create a quiet PSU.
The sketch makes it seem that a common ground for the rectifier, filter caps, and regulator will create quiet power rails. But isn't it true that all the current returns to the transformer center tap, which is at the opposite end of the ground flow from the audio ground return? This seems to me to be problematic, as the ground plane (say it is a copper wire) between audio ground and the transformer CT will be fluctuating slightly depending on the current going through it? If the ground reference for the regulators is fluctuating slightly (i.e. noise!) then the power rail will have some noise in it, yes?

The ground plane must avoid the CT and the smoothing caps gnd.
It should start at the reference point for the regulators.

 

[dmp]

The ground plane must avoid the CT and the smoothing caps gnd.

So the regulators and audio ground plane should only connect to the CT and smoothing caps ground in one place? I'm not sure exactly how to interpret 'avoid'.
I'm still visualizing all the current from the regulator and audio circuit returning to the CT and smoothing caps. Is this an incorrect way to think of it?

 

[abbey road d enfer]

The ground plane must avoid the CT and the smoothing caps gnd.

So the regulators and audio ground plane should only connect to the CT and smoothing caps ground in one place? I'm not sure exactly how to interpret 'avoid'.
I'm still visualizing all the current from the regulator and audio circuit returning to the CT and smoothing caps. Is this an incorrect way to think of it?

There is a voltage that develops along the conductor that goes from the smoothing caps to the CT; this voltage must not be allowed to pollute any other ground.
Avoid means the copper pour must not include the aformentioned portion.

 

[user 37518]

audio grounds should be star grounded but if two channel box, use different star grounds  for each channel.

I wonder what exctly what you mean when you say 2 star grounds for each channel, meaning to treat each channel as if it were alone inside the chassis? why would you want that, or how do you do that? what about the chassis (safety) ground, will it also be tied to both star grounds? and what about the PSU common? will it also be attached to both star grounds?

i built a 2 channel guitar amp one time and one channel had less hum than the other, even though each circuit was identical. the channel with the most hum was further away from the power trans, so why it had a tad more hum is still a mystery. :

Exactly the same thing happened to me once, the channel closer to the transformer had more hum, nothing i can do about it....

 

[earthsled]

Hi All,

I'd like to pick up this discussion once again...
Attached is figure from RANE's Grounding & Shielding of Audio Devices reference document. Can anyone enlighten me as to the advantages and disadvantages of the different connection techniques? Do some offer better results in audio equipment?

Thanks!

 

[abbey road d enfer]

I'll try to give some clues.
It's a vast subject, to which one could devote an entire life and several others more.

First option (no connection) means that some voltage may exist between chassis and audio ground. If this voltage is excessive, it may generate noise, by capacitive induction with the sensitive parts of the circuitry. This can happen if the audio ground has no galvanic connection to ground via another piece of gear, e.g. when the unit is fully balanced with input and output xfmrs. Then, if for some reason the audio ground is accessible, there is a risk for human creatures that may touch it.
Now how can a voltage develop there? The most common cause is capacitive leakage from the primary of the power xfmr; it is typically a few 100pF, up to several nanofards for large PT's. This is generally not enough to generate significant current at the fundamental mains frequency, but upper harmonics such as those created by power dimmers and other RFI can create relatively large current, because the leakage path's impedance decreases with frequency.
Second option (solid link): this will prevent any significant voltage between chassis and audio ground. Nothing wrong with it.
Third option (resistor): this will discharge to earth any low frequency mains noise (hum) but it may be too resistive to absorb higher frequencies due to RFI/RMI.
Fourth option (cap): will discharge EMI/RFI to earth , but not low frequency hum.
Fifth option (switch): offers choice between 1st and second option; can solve problems but ergonomically not acceptable for mobile/touring applications.
Sixth option (switch + cap): offers choice between 2nd and 4th option; same problem as precedent.
Seventh option (back-to-back diodes): limits the peak-value of the voltage at 0.7V.

Now if the unit has unbalanced connections, there is a safety aspect that is much more important than the noise performance.
The open circuit, the resistor, the cap and the switches are not acceptable in that respect; if the PT fries resulting in a short between mains and audio ground, one needs to make sure that the connection between audio ground and earth (chassis) is capable of limiting the voltage, resisting the surge and tripping the fuse/breaker. Only the solid link and the diodes (providing they are properly dimensioned) would conform to these multiple requirements.

The diode option is justified by the fact that the audio ground is floating (within limits) from the chassis (and earth). In a relatively simple electroacoustic chain, this may allow a unit to have its audio ground "adjusted" to the ground of another unit, but as soon as the system gets complex, it doesn't work anymore. Let's say you have such a unit, e.g. a mixer connected between a rackmount CD player and an amplifier mounted in a different rack: to which ground reference will it adjust? Such a concept works only if all the units except one are floating.
That's how many HiFi systems work; the sources CD, tuner, turntable... are floating, only the amp is grounded. This makes the use of dual-insulation PT's mandatory.

As usual, Ohm and Kirchoff are your friends there; consider any earth or ground as a resistor.
I guess there will be many comments; I have just scratched the surface.

Finally, there's one option that's not been included, although it is favoured by many manufacturers, it's a parallel RC circuit. It covers both LF and HF and is reasonably safe if the components are properly dimensioned. It's probably the most sensible choice.
Remember that all these arrangements except the solid link are utterly useless on fully balanced units.

 

[DerEber]

I really liked this reading on that topic:

http://www.valvewizard.co.uk/Grounding.html

THANKS VALVEWIZARD!!!!