Simple power supply grounding questions

[PermO]

I'm having trouble to wrap my mind around this...

As there's only one planet Earth where you can stick a grounding rod in the soil... they will always meet at some point.

(I've been working on a offshore project, and there's two huge copper busbars, one labeled green yellow, mains ground and one labeled, black yellow, instrumentation ground... so first thing I did was take my multimeter to the busbars to find there was 0 Ohms between them ..... aaaargh .... just shoot me !)

But appearantly it is important where they meet...

To check if I see this right;

AC ground from the IEC socket connects to chassis (all metal parts) and to the PSU after rectification (I usually connect this point to my last capacitor negative pole on the PSU output)...
Now there's also an audio circuit, it has input and output transformers, so this audio circuit is actually 'floating' between the two transformers and there's no physical connection to the outside.
So this part connects to the signal ground comming from the XLR input ?

Is there a point in the unit where these two meet ? ...they will shurely meet somewhere outside the unit as there is only one planet Earth...

edit;
Wouldn't it be best practice to leave the floating part floating ?
Does there really need to be a reference to ground ?

I also noticed on most really really old gear, there's no ground connection at all ?

(my interest is with this situation as I am planning to build this) ..So with this transformer balanced circuit it sort of makes sense to me.

But if we have a transformerless single ended design mains ground and signal ground are allready connected at the input, so it will no longer matter what you connect where ?

Ian, thanks for the document, but I'm still not shure if I really get this.

I've never encountered any hum problems in my builds so I guess I'm doing it right...

Aaaargh... just shoot me !

 

[Matt Syson]

Hi
The most important aspect is safety for which I think Ian's comment about the mains earth 'arriving' has a connection with a (usually) dedicated solid connection to the metalwork and then often to all conductive panels throughout the enclosure. 'Best' places for audio 'performance may well be different and then takes you swiftly into the realm of EMC compliance. As the world has physical dimensions there is also the fact that all conductors are aerials, at a resonant frequency determined by dimensions with the upshot that the 'simplistic' 'star ground' concept is great for electrical safety (at power line frequencies) one end of your 'ground wire might actually have received a signal peak voltage at the other end. Microwaves (your mobile phone) have wavelengths only a few centimeters/inches long.
In the 'audio' world placing of grounding can be critical, for example pin 1 of XLRs, at least for phantom powered microphone circuits as you are linking the common of the 48 Volt supply to the chassis. Juggling in 3 dimensions is great fun.
Although this is a digression, the placing of grounding is one pf the significant causes of problems with computers and audio convertors as the mains earth of a computer is not physically taken to the same point as a USB port and although the computer 'box' may contain the high frequency 'noise', it can be superimposed on the external wiring in some circumstances (I think).
The common thought that 'balancing' is a universal solution is again only partly correct, or I suppose IS correct but falls down in the real world as 'Common Mode' Signal Rejection only works for a specific frequency range. Usually pretty well for powerline and it's lower harmonics at least but when confronted with 'noise' from switcher supplies is woefully inadequate, so filtering is necessary.
Even standing on your mic cable can upset the 'balance' in extreme situations as well as wiring where the inphase conductor follows a slightly different physical route to the out of phase conductor.

 

[PermO]

I get the safety part...
Though in the old days they seemed to care a lot less about it...

And I have mistaken pin 1 for signal ground, which it is not, it's screening, and screens can be tied to chassis without causing trouble.

Wouldn't it be best practice (for audio specs) to leave the floating part floating ?
As it sits in a grounded metal box, with a grounded powersupply, why would you mess up a balanced floating circuit by tying one leg to ground ?

(So far my builds have always been single ended, or with just a plate to line output transformer)

 

[abbey road d enfer]

I get the safety part...
Though in the old days they seemed to care a lot less about it...

*Not only safety, also EMI/RFI, which has become a most preoccupying subject in the meantime.

Wouldn't it be best practice (for audio specs) to leave the floating part floating ?

As it sits in a grounded metal box, with a grounded powersupply,

There's all sorts of radiation and capacitive coupling that would interfere with sensitive circuitry. Referencing the circuit to the casing reduces considerably these.
In addition, in case of defect, referencing the audio circuit to the chassis/earth makes sure no lethal voltage can be presented to the outside world.

why would you mess up a balanced floating circuit by tying one leg to ground ?

With balanced connections, none of the legs is grounded.

 

[PermO]

Yes, with balanced connections none of the legs is grounded, I understand...

Maybe I'm overcomplicating things ...

I'm looking at the Federal AM864 schematic and indeed I see a ground connection at R13 R14 and R9 but the circuit is still balanced as both legs refer to ground over a resistor, I think I see it now.
So this point will connect to the bolt at the IEC input / chassis.
All XLR pin one can be tied together and connected to this same grounding bolt (a groundlift switch could be applied here)
Then there's R8, current control wich will tie to the PSU ground, or would it be better to tie this to the same bolt ?

 

[abbey road d enfer]

I'm looking at the Federal AM864 schematic and indeed I see a ground connection at R13 R14 and R9 but the circuit is still balanced as both legs refer to ground over a resistor, I think I see it now.
So this point will connect to the bolt at the IEC input / chassis.

It is eventually connected to the earth/chassis. But I would bet the best point to connect it is not the earth lug. Remeber every conductor has resistance. What is drawn as a single node is actually a series of low value resistors intertwined and sharing currents.

All XLR pin one can be tied together and connected to this same grounding bolt (a groundlift switch could be applied here)
Then there's R8, current control wich will tie to the PSU ground, or would it be better to tie this to the same bolt ?

You must learn to evaluate what belongs to the audio ground and what belongs to the earth/chassis. The point you describe must be connected to a busbar (very bad practice to connect it to the chassis, although it's been sanctified by time), according to rule #101 of grounding: "ground follows signal". Replace all the ground connections in a schematic with resistors and analyse. Disonnect chassis and audioground and watch teh consequences; does it still work?
After two decades you'll have the feel for it.

 

[PermO]

Yes, I've always used a piece 2.5mm solid copper wire for all ground connections, I don't like the chassis to be 'part of the circuit' ...
I'll be doing a point to point build, so I can experiment, though I would like to get it right the first time.

In two decades I'll turn 70, grow a beard, become a hipster and will do a startup campaign to become a new gear manufacturer, my products will be overprized, plastic, make no sense and sound like sh*t

 

[ruffrecords]

Yes, with balanced connections none of the legs is grounded, I understand...

Maybe I'm overcomplicating things ...

I'm looking at the Federal AM864 schematic and indeed I see a ground connection at R13 R14 and R9 but the circuit is still balanced as both legs refer to ground over a resistor, I think I see it now.

I think you are misunderstanding balanced. A balanced signal exists only between the hot and the cold. Pin 1 is a screen and does not carry signal current.

So this point will connect to the bolt at the IEC input / chassis.
All XLR pin one can be tied together and connected to this same grounding bolt (a groundlift switch could be applied here)
Then there's R8, current control wich will tie to the PSU ground, or would it be better to tie this to the same bolt ?

The analogue 0V, the chassis and the mains safety earth pin of the IEC should all connect to a bolt close to the IEC. All XLR pins 1 connect directly to the chassis at the XLR and nowhere else.

Cheers

Ian

 

[PermO]

Initially I thought when applying a ground connection in a balanced circuit (between the transformers) it would mean that the negative phase would be tied to ground... I now see from the schematic you can reference both signal halfs to ground and remain having a balanced circuit.

I've only done single ende tube circuits sofar.. this topology is new for me, but I always try to understand what I am building, and shipping companies make shure I have a lot of time for that now... where are my parts !?

Thanks for the tip on the XLR's, I will apply that.

I found your grounding document a couple years ago and I think it is very well written, big thanks for that !

 

[ruffrecords]

Initially I thought when applying a ground connection in a balanced circuit (between the transformers) it would mean that the negative phase would be tied to ground... I now see from the schematic you can reference both signal halfs to ground and remain having a balanced circuit.

There are not two halves to the signal. As I have said many times before, a balanced signal exists only between the hot and cold (XLR pins 2 and 3). There is no other reference and the voltage between hot and ground or between cold and ground is not defined. There is no requirement for the signal between hot and ground to be half the normal value. This only occurs in (usually semiconductor) differentital outputs where they do refer each output to ground. There is no need to do this and it is not a requirement for a balanced system. In telephone systems they send hot and cold down miles of twin cable with no screen or ground at all and connect them to a transformer at the other end.

I've only done single ende tube circuits sofar.. this topology is new for me, but I always try to understand what I am building, and shipping companies make shure I have a lot of time for that now... where are my parts !?

Thanks for the tip on the XLR's, I will apply that.

I found your grounding document a couple years ago and I think it is very well written, big thanks for that !

Think of a single end audio amp with an output transformer feeding a speaker. You can take a pair of wires from the transformer and connect them direct to a speaker. Neither one needs to be connected to ground. This is balanced signal connection. It is as simple as that.

Cheers

Ian

 

[Matador]

In many cases it's best to substitute 'impedance balanced' whenever you see the word 'balanced', because it will immediately clue you in to the fact that it's an interconnect topology, not a description of the time-varying behavior of the signals with respect to each other. On the other axis, you have 'single ended' versus 'differential' signals, which just describes whether or not you have a single signal which is referenced from a fixed potential, versus two signals which are referenced to each other (to be pedantic, even single ended signals are differential, it just one of the signals is a fixed potential, often 0V). For a differential signal, there is effectively no 0V level, because it has no meaning in that context.

For example, It's not uncommon to send a single-ended signal down an 'impedance balanced' interconnection. You could also send a differential signal down a non-impedance balanced interconnect (like coax), however I'm not sure why you'd want to.

 

[abbey road d enfer]

In two decades I'll turn 70, grow a beard, become a hipster and will do a startup campaign to become a new gear manufacturer, my products will be overprized, plastic, make no sense and sound like sh*t

Excellent resolution!

 

[ruffrecords]

In many cases it's best to substitute 'impedance balanced' whenever you see the word 'balanced', because it will immediately clue you in to the fact that it's an interconnect topology, not a description of the time-varying behavior of the signals with respect to each other. On the other axis, you have 'single ended' versus 'differential' signals, which just describes whether or not you have a single signal which is referenced from a fixed potential, versus two signals which are referenced to each other (to be pedantic, even single ended signals are differential, it just one of the signals is a fixed potential, often 0V). For a differential signal, there is effectively no 0V level, because it has no meaning in that context.

For example, It's not uncommon to send a single-ended signal down an 'impedance balanced' interconnection. You could also send a differential signal down a non-impedance balanced interconnect (like coax), however I'm not sure why you'd want to.

Excellent answer because the things that are balanced are the impedances not the signal levels

Cheers

ian

 

[PermO]

Think of a single end audio amp with an output transformer feeding a speaker. You can take a pair of wires from the transformer and connect them direct to a speaker. Neither one needs to be connected to ground. This is balanced signal connection. It is as simple as that.

This I understand, no problem.

I guess I'm not using the right terminology as I am not educated in circuit design, refering to the 864 circuit and it's ground connection points.

Now the speaker is a input transformer with a CT tied to ground ?
I'm trying to understand why these ground connections are there and where to connect them.

With sigle ended designs there's just one ground... my brain can cope with that one.

 

[abbey road d enfer]

Now the speaker is a input transformer with a CT tied to ground ?

A speaker is not a transformer...
The secondary of the output transformer is connected to the moving coil of the speaker.
Both the secondary and the moving coil are intrinsically floating (they don't need to be grounded or referenced to anything).
Grounding a leg of a floating connection is possible because there is no current, since there is no return path.
However, very often one of the ends of the secondary will be grounded. The main reason generally is that the voltage from the secondary is used to apply negative feedback to the input of the amp. Then it does not qualify as a floating connection anymore.

I'm trying to understand why these ground connections are there and where to connect them.

Very often they are not necessary. They maybe there either for assuming an auxiliary function (e.g. signalling) or for safety reason.
In case a transformer breaks its isolation, the output may carry high voltage.

 

[PermO]

A speaker is not a transformer...

Yes, but in both cases we are exiting a coil. Ian brought up the speaker as an example of a balanced connection.

The secondary of the output transformer is connected to the moving coil of the speaker.

In the case of a vacuum tube poweramp, yes, with a line amp it could be anything.

Both the secondary and the moving coil are intrinsically floating (they don't need to be grounded or referenced to anything).

Just like a transformer balanced line out / line in interconnection.

Grounding a leg of a floating connection is possible because there is no current, since there is no return path.
However, very often one of the ends of the secondary will be grounded. The main reason generally is that the voltage from the secondary is used to apply negative feedback to the input of the amp. Then it does not qualify as a floating connection anymore.

This is helpfull, thanks.

Very often they are not necessary. They maybe there either for assuming an auxiliary function (e.g. signalling) or for safety reason.
In case a transformer breaks its isolation, the output may carry high voltage.

I'll just stick to the schematic...

 

[neil.johnson]

AES48 defined the industry standard for shield connections on XLR pins, and the author (RIP) spent several years persuading pro-audio companies to stop doing the bad thing and start doing the good thing. So when someone points to a schematic I am always cautious as many companies did (and few still do) the bad thing.
If you're not a member of the AES you can find a draft edition here: https://www.aes.org/standards/comments/drafts/aes48-xxxx-190121-cfc.pdf

(attached for posterity)

 

[MisterCMRR]

I've written about these subjects for at least 25 years - and was part of the AES standards working group that created AES48. Both grounding and signal interfaces are widely misunderstood. I've attached here a version of my seminars created for Jensen (when I was the owner and chief engineer of the company). Hopefully, it will "connect the dots" so-to-speak for these subjects. Feel free to copy and distribute it ... all I ask is that it be properly attributed to me if portions of it are posted somewhere. Although I sold Jensen in 2014 (to Radial Engineering of Vancouver BC), the Jensen website still has hundreds of documents I created over my 25 years there - including the chapter "Audio Transformers" that I wrote for the "Handbook for Sound Engineers." I've always tried to teach in a conceptual way and avoid as much math as possible. Nothing will discourage a beginner more than pages of equations! This material seems especially relevant to this thread.

Enjoy,
Bill Whitlock
Whitlock Consulting
Ventura, CA, USA
[email protected]
IEEE Life Senior Member
AES Life Fellow

 

[PermO]

Thank you very much for this document, I saved a copy for my personal DIY database.

Love the presentation !

 

[warpie]

Very interesting notes. Thank you.