Signal and power grounding

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi,
Today I just read the nice grounding article http://www.diyaudio.com/forums/diya...udio-component-grounding-interconnection.html

And then I have planned to construct my new integrated amp with balanced inputs. I have a question regarding the interconnection of volume and input wires.

The article states to grounding them to the chassis as below image.

fig4.6-1.png


But I want to have the star signal ground(signal reference) on pcb and ground them to this one, signal ground will connected to the star power ground.
Because of it is easy to do this way in case of my 4 channels discrete resistors/relay-based volume. lots of ground branches from resistors have to be grounded.
As I ever seen inside chassis of some hi-end integrated amp, they have very less outside pcb wiring, but it sound awesome and low noise.

Please give me your opinion. Is this way can also be the alternative grounding?
 
Last edited:
Hi Guys

Star grounding should be avoided unless you wish to maximise noise and intermodulation effects.

Rather, use the Galactic Ground, which us a buss with small stars for local circuitry and local distributed filtering.

As far as connecting balanced inputs/outputs: Rane has a series of "Rane notes" that discuss this and the "pin-1 problem", written by recognised experts in audio.

You need to get the balanced signal into the chassis before you change its level, so place the balanced to SE circuit ahead of the level control. The grounding shown in your drawing will allow chassis noise to enter the audio path. Pin-1 should generally go to the chassis right where the jack is, to keep the RF path as short as possible. The hot and cold lines go to the diff-amp, which provides rejection of ground differences and common-mode noise. This feeds the level control, which can now be one section per audio channel and can also be reduced in value to help keep noise generated within your chassis low.

Similarly, at the output, the pin-1 ties to chassis either directly or via a small cap. Hot and cold provide the relevant audio signals, happily "not referenced to ground".

The chassis represent noise and shielding. First off, it is tied to the mains safety ground to assure that fault currents are sent directly through that connection via a grounded wall socket. Second, the chassis provides a small amount of shielding against outside noise. Any noise current induced into the chassis is just that - a noise current - and you really do not want this to contaminate the audio currents.

The audio ground should only connect to the chassis at one point. This does NOT mean that all the audio circuit grounds are tied to one chassis point, rather as above: each gain stage has associated leak resistors, gain-set Rs/Cs, etc, plus a local filter cap. All of these tie to one star which is tied to a buss that connects all the other stars together into a "galaxy" - hence, a Galactic Ground. This organisation controls all the circulating currents for DC bias and signals, keeping them separated and providing the lowest possible noise and IM distortion.

Have fun
Kevin O'Connor
 
Hi Guys

Star grounding should be avoided unless you wish to maximise noise and intermodulation effects.

Rather, use the Galactic Ground, which us a buss with small stars for local circuitry and local distributed filtering.

Hi Kevin, Thanks for the reply. So you advise to do the local star ground for each circuitry and connect all of them to the noisier ground (PS cap common), let say like do partitioning them like this scheme, Am I right?

An externally hosted image should be here but it was not working when we last tested it.


Hi Guys
Similarly, at the output, the pin-1 ties to chassis either directly or via a small cap. Hot and cold provide the relevant audio signals, happily "not referenced to ground".

An externally hosted image should be here but it was not working when we last tested it.


So please advise where point 1 have to be connected to?
 
Last edited:
Hi Guys

Your interpretation is exactly a star ground and completely NOT a Galactic Ground. Please re-read the description.

The fact you can recognise that the PSU common to the filter caps is noisy is a PLUS in your favour. That detail should tell you right away that your following supposition is false. From those first filter caps power feeds to regulators in most opamp-based circuits, which are followed by output filers. As currents circulate from one node to the next they are cleaner. The PSU attachment to the rest of the circuit should come from the final output filers.

Note that _electrically_ the common for the first filters, the refs for the regulators, and the commons of the output filters are all "ground" and all nominally the same "point", but... These three circuit nodes must be considered as separate points. The regulator refs can be tied to the output filter common, but the rectifier feed _into_ the first filter MUST NOT BE the same wire as the output feed _from_ those filters. There must be a physical separation or else the rectifier noise will modulate the output DC. There is not much we can do about the leads on the caps themselves other than make them short.

Similar care should be taken with the circulating currents in the rest of the circuit. There should be enough detail provided here for extrapolation to the rest of the circuit.

See "The Ultimate vol.3" for complete details.

The issues of where to tie pin-1 of XLRs (S on TRS) has been the bain of audio system designers for decades. Almost every solution is a compromise if balanced and single-ended connections are to be accommodated.

The issues of feeding split-rail power to several circuits introduces a compromise in the ideal flow of currents and the wiring needed for them. Every multi-channel circuit, from stereos to large mixing consoles, deal with this issue and have to accept major compromises, particularly for more than 2-channels. Fortunately, since most such products incorporate hard regulation for the low-level circuitry, many sins can be committed with respect to ideal wiring without major detriment to hum and buzz levels. IM and THD will be a bit worse, but not to where average designers and products will be compromised beyond their (likely) other flaws.

Have fun
Kevin O'Connor
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.