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.

boji

Well-known member
Joined
Jan 6, 2010
Messages
2,375
Location
Maryland, USA
Hello yall,

I was reading Ian's helpful sheet on proper grounding of consoles, and I got to the part about phantom Ov issues that can arise. Advice was to have a separate supply for phantom, but in my case, the PSU shares signal com with phantom com over the same pin on the power connector.

Any advice how to best treat phantom with this setup? I can isolate P-com with a different PSU if needs be, or just run phantom 48v and P-com to 500 series backplanes that house mic pre's-- however I'm afraid since the backplanes are getting acom from the busbar well (and now also from phantom run), I'll have created a ground loop potential.
XWG3Ulh.jpg


ref: http://www.ianbell.ukfsn.org/EzTubeMixer/docs/EzTubeMixer/SimpleMixer/grounding101v2.pdf
 
Pin 1 of the XLR in should be connected to the chassis over the shortest possible wire to limit RF entry. This is why many female XLRs have a metal spike next to the screw hole to stab the metal chassis panel and connect it to pin 1. That is the phant ground. You do not need to (maybe even should not) connect pin 1 to the mic pre circuit or anywhere else if pin 1 is grounded through the chassis. Next to the power input there should be a bolt and toothed washer on the chassis that grounds it to the power supply ground with an extra stout wire. The bolt is also where earth should be connected.

Of course with plastic enclosures and floating supplies a lot of modern gear doesn't actually do this anymore. But if you're building a one-off metal box with phantom and mains, do the above.

Note: There was a discussion recently about weather or not to connect transformer screens to pin 1 or the circuit ground but I don't recall the details.
 
This may sound like a dumb question, but isn't the entire signal of the microphone, referenced to phantom ground, not the signal ground?

Just because you DC block it on the way in - doesn't mean it loses it's reference?

I'm really confused on this topic. It's way easier with an isolating input transformer, but with IC's, and DC blocking caps, my mind "blacks out" trying to work out what's referenced to what.
 
Rochey said:
This may sound like a dumb question, but isn't the entire signal of the microphone, referenced to phantom ground, not the signal ground?

Just because you DC block it on the way in - doesn't mean it loses it's reference?

I'm really confused on this topic. It's way easier with an isolating input transformer, but with IC's, and DC blocking caps, my mind "blacks out" trying to work out what's referenced to what.
No because a mic input is almost always differential. With a differential input (either a transformer or coupling caps w/ transistors in front of an op amp), the signal is recovered entirely from debalancing pins 2 and 3 which renders the 0V "reference" to be not so important. Common mode noise of phant ground is rejected. The primary function of phant ground is to 1) ground the body of the microphone to shunt electrostatic noise and 2) return current for 48V.
 
Perhaps search pin 1 problem...

Of course pin 1 is the phantom current return path, but common mode to audio so IxR not that critical, should be bonded to chassis ground for RF suppression.

BTW Ground is a concept not a voltage node.

JR
 
Phantom power is the exception to the rule. It is a bit of a kludge really.

Rule 1: is pin1 = chassis = screen = a way to conduct interference safely away

Rule 2:  (in a balanced system) Pin1/chassis/screen is not analogue signal 0V and the two should only ever be connected together in a single place (usually the bolt next to the IEC inlet to which safety earth is also connected)

The problem with phantom its its 0V analogue is connected to the mic cable screen which is connected to pin1 then chassis in violation of Rule 2. So the question becomes how to minimise its potential degradation of the audio.

If you common phantom 0V with analogue 0V at the power supply then interference current can potentially flow in your 0V line. Do you want that? If not you should run a phantom power 0V from the PSU to the mixer and connect it to pin 1 of one of the mic inputs. In my early mixer builds, as well as connecting each mic input pin 1 to chassis at the connector, I also wired an 18 gauge bare copper bus across all the mic input pins 1 and connected phantom 0V to the centre of it. Probably overkill but I always run phantom power up its own dedicated pair in the PSU cable (and the PSU cable is screened).

Cheers

Ian
 
Indeed Pin 1 should be connected to chassis for EMI protection.
And yes, having the P48 supply  floating and ist negative connected to the Pin(s)1 is better. But how much? The only issue there is the differential voltage between the chassis and the 0v reference at the mic preamp stage. Soundcraft proposed a clever solution, which is to reference the mic pre to the Pin 1; which results in displacing the noise voltage at the output of the preamp, not at its input, so the S/N ratio is increased by the preamp gain. Actually, following the mic pre with a diff amp gest rid of the noise introduced by the loop. In practice the advantages were not obvious and they reversed to the former arrangement, which relies on somewhat brute force, with heavy wires.
And, as has been mentioned earlier, this loop noise is attenuated by the CMRR of the preamp, so concentrating on optimizing CMRR seems to be a promising endeavour.
 
abbey road d enfer said:
Indeed Pin 1 should be connected to chassis for EMI protection.
And yes, having the P48 supply  floating and ist negative connected to the Pin(s)1 is better. But how much? T
It does not matter. The incremental cost of one wire is tiny and there is an improvement.
The only issue there is the differential voltage between the chassis and the 0v reference at the mic preamp stage.
Is it? The issue is interference currents in the mic cable screen flowing in the common analog 0V wire to the psu which add a noise voltage to the analogue 0V feeding every circuit in the mixer.
Soundcraft proposed a clever solution, which is to reference the mic pre to the Pin 1;
What does this mean 'reference' the mic pre to pin1. Surely you are not suggesting connecting the mic pre analogue 0V to pin1??


Cheers

Ian
 
ruffrecords said:
What does this mean 'reference' the mic pre to pin1. Surely you are not suggesting connecting the mic pre analogue 0V to pin1??
Actually yeah. It's just moving the 0V branch point from the PS to pin 1 and moving the differential step to after the mic gain. So the CMRR is effectively applied twice. But I would think once is enough. I never thought about using the chassis as 48V return as a problem that needed to be solved. Also since the mic pre is already differential, adding another differential step just requires more circuitry in which case I think a ground cancelling bus would be a better place for that circuitry.
 
ruffrecords said:
It does not matter. The incremental cost of one wire is tiny and there is an improvement.Is it? 
It's not only the cost of one wire, there's the cost of having a separate transformer winding. As you know, it sometimes mean adding a separate xfmr.

The issue is interference currents in the mic cable screen flowing in the common analog 0V wire to the psu which add a noise voltage to the analogue 0V feeding every circuit in the mixer.
I don't understand that. Draw a circuit and tell me where this supposed current flows.

What does this mean 'reference' the mic pre to pin1. Surely you are not suggesting connecting the mic pre analogue 0V to pin1??
But that's exacty what I mean. The reference voltage of the preamp could be in theory attached anywhere, as long as it's output is connected differentially. Indeed there are issues with CMR, but here we are dealing with minute voltages.
 
Thank you so much for everyone's input!  Plenty to digest.  I'll post back on results after using Pin1's for P48 returns.
 
abbey road d enfer said:
It's not only the cost of one wire, there's the cost of having a separate transformer winding. As you know, it sometimes mean adding a separate xfmr.
I partly agree. The wire is already there if you take the care to screen the dc supply cable (which you should)
I don't understand that. Draw a circuit and tell me where this supposed current flows.
No need. If you use analogue 0V also as phantom common then by definition the analogue 0V is connected to chassis at the mixer end. This means all the chassis/cable screen borne interference currents flow down the analogue 0V connection  from the mixer to the power supply and hence into safety ground. The correct way to do it is to screen the power cable and connected the screen of this cable to chassis at both ends. Connect phantom 0V to the screen at the power supply end. Now any chassis/cable screen induced interference currents flow down the screen of the cable. You really do need to isolate the phantom supply to do this.

It is all in the Rane notes.

Cheers

Ian
But that's exacty what I mean. The reference voltage of the preamp could be in theory attached anywhere, as long as it's output is connected differentially. Indeed there are issues with CMR, but here we are dealing with minute voltages.
[/quote]
 
ruffrecords said:
I partly agree. The wire is already there if you take the care to screen the dc supply cable (which you should).
I said the cost of wire is irrelevant. It's the cost of having a floating 48V source that matters.

  If you use analogue 0V also as phantom common then by definition the analogue 0V is connected to chassis at the mixer end. This means all the chassis/cable screen borne interference currents flow down the analogue 0V connection  from the mixer to the power supply and hence into safety ground.
Sorry, this makes no sense to me. You may think you don't need a schemo, but I do. Interference currents don't come out of thin air.

  The correct way to do it is to screen the power cable and connected the screen of this cable to chassis at both ends.
  Really? Screened power cables?

Connect phantom 0V to the screen at the power supply end. Now any chassis/cable screen induced interference currents flow down the screen of the cable. You really do need to isolate the phantom supply to do this.
I don't understand what you mean. Where do these currents come from? It seems there's a mix-up between RFI and ground loops here.

 
ruffrecords said:
If you use analogue 0V also as phantom common then by definition the analogue 0V is connected to chassis at the mixer end. This means all the chassis/cable screen borne interference currents flow down the analogue 0V connection  from the mixer to the power supply and hence into safety ground. The correct way to do it is to screen the power cable
If pin 1 is connected to analog 0V, then there would have to be a differential out for that channel. Meaning each channel would have a local 0V that is not connected to the mix bus. I think that might be where your circuit thoughts are different from what abby is describing.

Although I don't know why you guys are burning posts on some experimental soundcraft topology that will probably never be repeated. IIUC we're all in agreement that pin 1 should not be directly connected to analog 0V.
 
abbey road d enfer said:
I said the cost of wire is irrelevant. It's the cost of having a floating 48V source that matters.
There is certainly a cost associated with a floating phantom supply but in a professional mixer it is a tiny fraction of the overall cost.
Sorry, this makes no sense to me. You may think you don't need a schemo, but I do. Interference currents don't come out of thin air.
In fact they do. Why else screen mic cables?
  Really? Screened power cables?
Really, I use them all the time.
I don't understand what you mean. Where do these currents come from? It seems there's a mix-up between RFI and ground loops here.
I said interference currents. Their source does not matter. RFI or ground loops they are still interference. And by the way, RFI does not have to be megaHertz. Plenty of low frequency stuff around from mains cables, transformers and fluorescent lights.

Cheers

Ian
 
ruffrecords said:
There is certainly a cost associated with a floating phantom supply but in a professional mixer it is a tiny fraction of the overall cost.
Not all mixers are "professional". Even then there are always cost considerations. The picture presnted by the OP is a clear example of that.

"Interference currents don't come out of thin air" In fact they do. Why else screen mic cables?
Please tell me how grounding the 48V at the PSU or at the chassis makes a difference for EMI/RFI. You don't believe a schemo is useful, but I do. Please look at it. The only difference is how voltages developed across the PSU wires are interfering with signal. As was discussed before, this is supposed to be taken care of by CMRR.  If commoning 0v's was so damnable, why did so many did it? They can't all be idiots.

 
"Really? Screened power cables?"Really, I use them all the time.
The fact that you use them does not make them indispensable nor necessary. Shielded power cables are a necessity for high-power motors and smps that generate a lot of EMI, but for a typical mixer, even one that has smps, I think it's overkill. Modern switching amplifiers with nominal power of about 20kW don't need a shielded power cable.

 
  I said interference currents. Their source does not matter. 
  Yes it matters. Conducted and emitted interference need to be treated differently. Emitted deals only with the receiving end of interference, conducted has to deal with the entire loop. Emitted interference is essentially voltage, conducted is essentially current.
In order to treat them, an equivalent circuit is a great help.
 

Attachments

  • 48V seaprate vs commoned.jpg
    48V seaprate vs commoned.jpg
    53 KB · Views: 30
abbey road d enfer said:
Not all mixers are "professional". Even then there are always cost considerations. The picture presnted by the OP is a clear example of that.
Please tell me how grounding the 48V at the PSU or at the chassis makes a difference for EMI/RFI. You don't believe a schemo is useful, but I do. Please look at it. The only difference is how voltages developed across the PSU wires are interfering with signal. As was discussed before, this is supposed to be taken care of by CMRR.  If commoning 0v's was so damnable, why did so many did it? They can't all be idiots.

  The fact that you use them does not make them indispensable nor necessary. Shielded power cables are a necessity for high-power motors and smps that generate a lot of EMI, but for a typical mixer, even one that has smps, I think it's overkill. Modern switching amplifiers with nominal power of about 20kW don't need a shielded power cable.

    Yes it matters. Conducted and emitted interference need to be treated differently. Emitted deals only with the receiving end of interference, conducted has to deal with the entire loop. Emitted interference is essentially voltage, conducted is essentially current.
In order to treat them, an equivalent circuit is a great help.
+1  several good points.

JR
 
OK, here is a schematic:

mixerpsuconnection.jpg


which shows a mixer connected to a power supply using an unscreened cable. The power supply creates a positive rail and 48V which share a common 0V. To meet safety regulations the chassis of the power supply is connected to the mains safety earth. At the mixer end the 48V supplies the mic inputs and in order for this to work the common 0V needs to be connected to the chassis because the mic XLR pins 1 are connected to chassis. In order for the mixer chassis to be also be connected to safety earth (essential), the 0V at the power supply is also connected to safety earth.

With all  mixer pins 1 connected to chassis, any and all interference currents in the mixer chassis, howsoever generated (by ground loops or external RFI) are routed to safety earth via the chassis and the common 0V connection in the power supply to mixer cable. An unscreened cable may also be subject to external RF interference.

A better way to ensure interference currents are kept away from analogue 0V as much as possible is to screen the power supply to mixer cable and connect it to chassis at both ends. This way interference currents are not flowing in analogue 0V and their is effectively one big chassis surrounding both the mixer and its power supply.

I hope that clarifies my meaning.

Cheers

Ian
 
ruffrecords said:
OK, here is a schematic: which shows a mixer connected to a power supply using an unscreened cable. With all  mixer pins 1 connected to chassis, any and all interference currents in the mixer chassis, howsoever generated (by ground loops or external RFI) are routed to safety earth via the chassis and the common 0V connection in the power supply to mixer cable.
We have a very different conception here of what earthing does. Indeed voltages created by electrostatic fields are "dumped" to earth. Actually they are not, they are prevented to develop by the equipotential surface. The currents that circulate in the chassis are only partially diverted. For example, the return currents of any stage are nor modified in anyway by earthing the chassis, unless a gross design error has been made in mixing audio "ground" and earth. The currents that circulate in the chassis may well be the result of a loop. Again how do these currents interfere with the operation of the mic, and how can changing the grounding of the phantom supply change the issue?

An unscreened cable may also be subject to external RF interference.
There's (almost) no debate regarding the need to shield audio cables. The mechanism is well-known and well documented. It's a matter of mutual capacitance between the "interferer" and the "interferee". Susceptibility to that type of interference is governed by this mutual capacitance and the impedance of the "interferee". For mic or line, we're dealing with impedances of a few hundred ohms. For PSU connections, the impedances are a fraction of ohm. Not comparable.
Regarding the interference magnetic field, the conductors are very close, so whatever voltage resulting is developed equally in all conductors.
 
Back
Top