Phantom ref shared by PSU signal ref

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

squarewave said:
This is what I would consider to be "ideal" for an audio device in a metal chassis with an external supply.
So where you will connect power commons/grounds/0 volts from local smoothing/decoupling/filtering caps in the audio device?
And 0 volts from LED signalling, relays etc?
Shouldn't two metal boxes be connected directly?
Are you aware of the fact that the power cable can be 5 or more meters long, and your returning 0 volts wire from diff. amp will be exposed to all EMI/RFI from adjacent wires and environment?
 
moamps said:
So where you will connect power commons/grounds/0 volts from local smoothing/decoupling/filtering caps in the audio device?
Analogue 0V
And 0 volts from LED signalling, relays etc?
Preferably a completely independent floating supply but at Neve we used the audio 24V supply without problems
Shouldn't two metal boxes be connected directly?
Yes
Are you aware of the fact that the power cable can be 5 or more meters long, and your returning 0 volts wire from diff. amp will be exposed to all EMI/RFI from adjacent wires and environment?

Do you mean the mains cable or the DC power cable?

... if the latter, as Abbey has pointed out, the impedances are so low the EMI should not be a problem but  for certainly I would use a screened DC cable

Cheers

Ian
 
moamps said:
What a mess!
So where you will connect power commons/grounds/0 volts from local smoothing/decoupling/filtering caps in the audio device?
And 0 volts from LED signalling, relays etc?
Shouldn't two metal boxes be connected directly?
Are you aware of the fact that the power cable can be 5 or more meters long, and your returning 0 volts wire from diff. amp will be exposed to all EMI/RFI from adjacent wires and environment?
Within the audio device, the various grounds would be separate. That's the whole point!

LED / relay driver ICs can cause very sharp high current transients that no amount of filtering will remove. So, even though it's not shown in my diagram, I strongly recommend using a 3rd ground wire. Although ADCs and DACs should use analog ground.

The two metal boxes are connected together. If the power connectors have solder lugs, I think it would be fine to run a wire directly from it to the chassis bolt and then also have a second wire like in my diagram. But if we're trying to analyze what is "ideal", I would say the power connectors should probably be PCB mount so that a) the ground plane can be utilized as the "filter cap ground" or main branch point for all grounds and b) each pin can have a chip inductor and small cap to keep RF out.

Noise induced in the power cable will be common mode noise and so it should be rejected by the CMRR of the inputs. Having said that, 5 meters is a pretty long power cable. I'm not going to commit to predicting what exactly would happen in that case.
 
abbey road d enfer said:
Do you mean that all the various grounds are joined in the remote PSU only?

Yes, I believe that’s what he means.  This is how Altec and Western Electric specified grounding between modular units.  Their specs also built out to include rack frames and grounding rods. 
 
[whinge]
I get confused when different people refer to different things as grounds. Can we agree some common unambiguous terminology?

Safety earth is to stop you electrocuting yourself under fault conditions. I suggest we call this "Earth"

The chassis and some cable screens act as equipotentials and help prevent electrostatic interference reaching signal wires. I suggest we call these "Chassis"

The internal zero volts reference for the active electronics is connected to the audio power supply 0V. I suggest we call this "Analog 0V"

The zero volts reference for the phantom power supply (which may or may not be internally connected to Analog 0V inside the power supply) provides the return path for the 48V phantome supply to external microphones. I suggest we call this "Phantom 0V"
[/whinge]

Cheers

Ian
 
ruffrecords said:
[whinge]
I get confused when different people refer to different things as grounds. Can we agree some common unambiguous terminology?
The term "ground" is one of those misnomers that is generally accepted. There are lots of tech notes and such by Analog Devices and Rane and so on that all refer to "Grounding".

Also, the "Chassis" (ground) and "Phantom 0V" are the same and the chassis and "Analog 0V" must be connected together at the chassis bolt so I'm not sure what you mean by "may not be internally connected". They most certainly must be connected together.
 
squarewave said:
The term "ground" is one of those misnomers that is generally accepted. There are lots of tech notes and such by Analog Devices and Rane and so on that all refer to "Grounding".

Also, the "Chassis" (ground) and "Phantom 0V" are the same and the chassis and "Analog 0V" must be connected together at the chassis bolt so I'm not sure what you mean by "may not be internally connected". They most certainly must be connected together.

I don't mind what people call the topic as a whole, but within it, as engineers, we should have unambiguous terms for each of them.

What I was referring to is the common power supply design that creates  +15V, -15V and +48V from a single transformer winding and therefore unavoidably connects phantom 0V to Analog 0V at that point. They will have to be connected at some point but it does not have to be implicit.

Cheers

Ian
 
ruffrecords said:
[whinge]
I get confused when different people refer to different things as grounds. Can we agree some common unambiguous terminology?
Mature technology and at least some terminology is defined.

"Ground" is overused and often misapplied. Ground is not a voltage, it is typically a circuit node or path connection.
Safety earth is to stop you electrocuting yourself under fault conditions. I suggest we call this "Earth"
It is already called EGC (Equipment Ground Conductor), at least in the US, by UL and others.
The chassis and some cable screens act as equipotentials and help prevent electrostatic interference reaching signal wires. I suggest we call these "Chassis"
"Chassis ground"  is widely used (and typically bonded to EGC so internal fault can trip fuses or breakers or GFCI/RCD ).
The internal zero volts reference for the active electronics is connected to the audio power supply 0V. I suggest we call this "Analog 0V"
I think I call that "Audio 0V", or "audio low", or "local audio ground" depending on the context of how used (breaking my own rule about not calling anything ground, but "local ground nodes" make sense in the context of explaining use of differential circuits to forward and back reference clean audio signals between different "local ground nodes". They can all be connected together but still exhibit different voltage potentials at different locations.
The zero volts reference for the phantom power supply (which may or may not be internally connected to Analog 0V inside the power supply) provides the return path for the 48V phantome supply to external microphones. I suggest we call this "Phantom 0V"
[/whinge]

Cheers

Ian
How about pin 1? .

==
An old analogy I like is plumbing... we have clean and dirty water flows inside our plumbing and we don't want the sewage to corrupt our clean water...

This requires a bit of a mental stretch but try to visualize the currents flowing in conductors (or PCB traces), then think of conductors or traces as resistors with voltage drops (in some cases like high frequency, as inductors).

moamps said:
What about clean and dirt grounds ::)?
Back in the early 80s I got in trouble trying to use a parallel redundant "clean" and "dirty" ground system inside a large console. When installed in the main beam of an AM radio transmission tower those two ground networks looked like antennas.  :eek: I even recall the frequency of that f'n AM station as my two grounds had a couple volts of 960kHz swinging between them. Of course they were connected together at one common PS ground node, but that was not enough to bond them together at 960kHz several feet away.

I don't share this to scare anybody but to advise caution about over reliance on simple plans.

JR
 
abbey road d enfer said:
Not sarcasm, bewilderment. That means that all the grounds in the mixer are not equipotential. That maybe OK, maybe not.
Yes but if inputs have decent CMR then it should be OK. Maybe not with "a couple volts of 960kHz swinging between them". That's a hard scenario. Even really good transformers might have trouble with that. A 3rd "dirty" ground for LEDs and Relays is enough to keep those sharp currents out of the audio. But in practice there are going ADCs sampling audio for levels and such. Technically those interfaces would need to be differential as well but that scenario is much more forgiving since it's logic and not audio. If you want your LED meters to go down to -40dB, then it might be necessary to go differential (but using two ADC inputs instead of one and then do the "differential" bit in code might be enough).
 
JohnRoberts said:
Back in the early 80s I got in trouble trying to use a parallel redundant "clean" and "dirty" ground system inside a large console. When installed in the main beam of an AM radio transmission tower those two ground networks looked like antennas.  :eek: I even recall the frequency of that f'n AM station as my two grounds had a couple volts of 960kHz swinging between them. Of course they were connected together at one common PS ground node, but that was not enough to bond them together at 960kHz several feet away.

The biggest problem with AM signals is that they are demodulated on the first nonlinear element they hit (diodes, transistors, etc). FM is far more forgiving in that regard. The first analog mobile phones running at 400MHz overhere also produced major interference, especially in dynamic microphones, and these new smartphones produce absolutely no interference in my experience. The microwave bands are too high though.
 
squarewave said:
Also, the "Chassis" (ground) and "Phantom 0V" are the same
Why? Because of pin1 problem? You will find in many consoles power supplies (or power supplies for 500 type racks) that the Phantom 0V is connected to Analog 0V.
and the chassis and "Analog 0V" must be connected together at the chassis bolt
You are using word "must" too often, IMO. :)
 
moamps said:
Why? Because of pin1 problem? You will find in many consoles power supplies (or power supplies for 500 type racks) that the Phantom 0V is connected to Analog 0V. You are using word "must" too often, IMO. :)
I think that if you really examined one of those devices carefully you would find that pin 1 does in fact connect to the chassis. But as a practical manufacturing matter, it's not uncommon for all parts to be mounted directly to a PCB. Ideally, they should use XLRs that have the metal spike on the front that stabs the panel so that when the install the PCB pin 1 contacts the panel directly. But for various reasons, some manufacturers just didn't do that (the spike might be a relatively recent invention) and so the XLR mic input pin 1 is just soldered directly to the PCB ground plane which is of course also analog 0V. But if you look closely at the construction you would find a path to the chassis such as through a PCB mounting screw. In the schem it might not be labelled as chassis but it is connected to it. It's not ideal because lower frequency RF can get in, but manufacturers often compromise and deviate from what is ideal for all sorts of reasons.
 
moamps said:
The biggest problem with AM signals is that they are demodulated on the first nonlinear element they hit (diodes, transistors, etc). FM is far more forgiving in that regard. The first analog mobile phones running at 400MHz overhere also produced major interference, especially in dynamic microphones, and these new smartphones produce absolutely no interference in my experience. The microwave bands are too high though.
I have written about "rectification" before (in this forum). When active circuits can't slew fast enough to follow the AM radio signal the audio modulation gets crudely detected. This is a common problem in mic preamps because of mic cables (antennas), high gain, and sometimes less that adequate slew rates to cleanly reproduce modulated AM radio.

Trust me (or not) but my (dirty) power rail swinging a couple volts at 960kHz was a significant problem and not just to the input circuitry.

I shared the story to suggest that all power supply rails need to be low impedance at RF... It wasn't fixed with input filtering, but by stiffening up the two supplies with respect to each other with ceramic disc caps.

JR
 
80hinhiding said:
I'm changing other things as well.  In future I won't be doing point to point, even though I like that construction method.  For right now though, if you're like me, wanting to use less connectors, check out an option like PTFix.

This is an attempt to pay it forward, and hopefully I haven't mucked it up.

Adam
Thanks for the PTFix tip. I was not aware of them. They look really neat, especially the variety of fixing options.

Edit: Turns out they are available from Farnell.

Cheers

Ian
 

Latest posts

Back
Top