Should phantom power be supplied by a linear supply?

[Bo Deadly]

No, I DO connect pin 1 directly to the chassis. I also connect the phantom 0V to pin1 (not to the analogue 0V).

Attached is my approach documented.

Ok. I understand. So the key difference is that you're NOT connecting phantom 0V to the chassis bolt at the supply like HT. You're running phantom 0V out to the XLR pin 1s directly. And I guess you have a bit of buss wire touching each of multiple pin 1's or is phantom 0V and chassis connected in the backplane?

Are you not concerned with RF being radiated into the enclosure on that phantom 0V line? I must admit RF is not my wheelhouse but even with pin 1 connected to the chassis at entry, that phantom 0V is still potentially an antenna. No?

 

[ruffrecords]

Ok. I understand. So the key difference is that you're NOT connecting phantom 0V to the chassis bolt at the supply like HT. You're running phantom 0V out to the XLR pin 1s directly. And I guess you have a bit of bus wire touching each of multiple pin 1's or is phantom 0V and chassis connected in the backplane?

I bus mic inputs XLR pins 1 together with thick copper and connect the floating phantom 0V to its centre. Each XLR pin 1 is also connected to chassis at the connector. Note these are not on a back plane but screwed directly to the rear panel.

Are you not concerned with RF being radiated into the enclosure on that phantom 0V line? I must admit RF is not my wheelhouse but even with pin 1 connected to the chassis at entry, that phantom 0V is still potentially an antenna. No?

RF is not my gig either but I think this is no different to all the screens of internal cables connect to pins 1 of XLRs. I think the important thing for RF is the length of the wire from pin 1 to chassis because that's where the rf current flows.

Cheers

Ian

 

[abbey road d enfer]

This Jensen schematic looks very clear about it; trafo's phantom tap goes to psu, then both wires (+-) connect to R1,C1 filter on a switch without touching ground before it. Ground wire from C1 minus pole is connected to XLR pin 1 which is bolted to chasis at the screw this XLR. Then + wire from joint of R1,C1 go to joint at 6k8 resistors which are then connecting pins 2,3 of the same XLR.

OK, this is doing it by the book for a single-channel. What about multi-channel, and particularly a large-ish mixer?
You can't have a separate 48V supply for each input, so that means one has to find a workable compromise.
The reason for connecting pin 1 to chassis is for EMI/RFI protection. Whatever you do, the phantom power negative rail cannot be connected to all pins 1 at a single point.
One has to accept that the phantom zero-volt cannot be the same for all inputs.
Then what?
Creating a separate rail joining all pins 1 is identical (maybe worse) than using connection to a unibody chassis.
The voltage developed by EMI/RFI across a pice of copper is almost identical to a piece of steel or aluminium the same length, whatever the thickness or width. It can be easily checked with inductance calculators.
However, in order to save Ian from protesting , it helps when used as an addition, particularly with a modular rear panel chassis.
So basically, whatever parasitic voltage appears between the different pin 1 can hardly be improved.
The final performance is in the hands of the mic's PSRR and the global CMRR.
I advocate the use of decoupling the phantom power with a cap connected to the actual relevant pin 1. Then the parasitic voltage will be "dumped" to the mic's reference.

 

[Newmarket]

RF is not my gig either but I think this is no different to all the screens of internal cables connect to pins 1 of XLRs. I think the important thing for RF is the length of the wire from pin 1 to chassis because that's where the rf current flows.

There is a case for saying that screens of internal cables should connect to signal ground or other low impedance source rather than chassis (pin 1) given that they are already within a shielded enclosure and (presumably) their primary function is to eg attenuate capacitive cross talk between relatively Hi-Z signals.

Yes - the length and impedance of the pin 1 connection is the major factor wrt rf shielding so the direct to chassis approach wins.

As for internal rf emissions - it gets complicated but a thick busbar type thing will likely be a less efficient antennae than a thin round wire.

 

[sahib]

This RF issue was raised before. I am also not an RF person but it is not really a concern.

The phantom 0V wire is connected to XLR pin1 which is also connected to the case chassis. As the case chassis is (mains safety) earthed at the chassis stud point, the interference/noise currents arriving through the microphone cable screen will flow into safety earth through the chassis, not into that ground wire, as now the case chassis is the low impedance path to safety earth for the noise currents.

 

[abbey road d enfer]

As for internal rf emissions - it gets complicated but a thick busbar type thing will likely be a less efficient antennae than a thin round wire.

Yes, but it must be put into perspective. The inductance (which is what matters regarding EMI/RFI) of a 4 inch piece of 0.2" diameter wire is only half that of a 0.04" diameter. Very steep law of diminishing returns.

 

[moamps]

If the piece of wire connecting pin 1 and the case is 1cm long for example, all frequencies below approximately 7GHz (1cm = lambda/4) will be shunted into the housing. If someone is still worried about the spread of EM interference through the 0V busbar between multiple XLR pins 1, they can insert small RF chokes in series between each pin 1 and the phantom power 0V busbar. Using the case as a 0V phantom power reference is wrong and engineering suboptimal, IMO, as I explained before.

 

[Newmarket]

Yes, but it must be put into perspective. The inductance (which is what matters regarding EMI/RFI) of a 4 inch piece of 0.2" diameter wire is only half that of a 0.04" diameter. Very steep law of diminishing returns.

Yes. Diminishing returns indeed. But nevertheless can be useful in getting that EUT under an EMC limit. fwiw I don't think I've ever replaced a wire with a thicker wire. But have insisted on a braid connection or beefing up a pcb OV connection with copper bar.

 

[abbey road d enfer]

But have insisted on a braid connection or beefing up a pcb OV connection with copper bar.

Beefing up the reference (0V) connection is always good practice, particularly in mixers.

 

[ruffrecords]

As I have said before, phantom power is a kludge so any implementation of it is not going to be perfect. The best solution is not to have it and to use externally powered tube mics.

 

[Bo Deadly]

This RF issue was raised before. I am also not an RF person but it is not really a concern.

It's kinda funny that everyone admits that RF is not their specialty and then makes proclamations about what the right thing to do is. I think it is easy to underestimate how RF can get into places you might think it shouldn't. RF does not behave like conventional AC signals. Impedance of wires and capacitors and the effects of parasitic capacitance of parts is different at RF frequencies. At least I would not be shocked if a little piece of wire hanging off of some pin 1 has just the right inductance and capacitance to pickup your neighbors outdoor wireless speaker transmitter, get amplified, demodulated by the grid of a high gain pre and then suddenly you're tube pre is playing Smash Mouth.

I'm sure there's lots of pro-sumer gear at least that does not use chassis for phantom return but that is mostly because the chassis is frequently plastic or, perhaps more likely, it was just easier (cheaper) to solder the XLRs directly to the PCB and then just stuff the whole thing into the enclosure. But just for one data point, the Mackie VLZ mixer, which is actually a very well designed mixer, does appear to use the chassis for phantom power return:


and then it gets tied together at one point:



All of the jumpers are probably for debugging I guess.

 

[FIX]

When I made the API consoles, I would make sure that the hot, cold and shield from the mic panel, through the patchbay, into the console never saw any other ground except itself until it literally hit the back of the Mic Pre in the console. This insured minimal ground issues, and made sure that the 48V was referenced only to it.

The correct way to do a 500 rack is to simulate the same thing. PIN 1 of the 15 pin module connector is CHASSIS, and that should connect to the chassis, not any other ground in the module or the XLR jacks. They have a chassis connection to the shield of the connector. That should go to chassis. Then in the power supply, you connect the power supply DC ground to the backplane and to the chassis (with a removable jumper or switch if desired) in only one place. This insures that everything is grounded according to code or CE, and minimizes any ground loop issues created from multiple points.

IN the US, the ground lift is useful, in Europe, for the most part, the earth is actually quiet and can be used as a mic ground and tied to pin 1 of the XLR. I learned this the hard way when I shipped a rack of mic pres to Manor Mobile to be used to record Oasis at The Wembley Arena. It was humming right up until show time, when we realized that pin 1 needed to be earthed. And the show went on.

 

[Newmarket]

It's kinda funny that everyone admits that RF is not their specialty and then makes proclamations about what the right thing to do is. I think it is easy to underestimate how RF can get into places you might think it shouldn't. RF does not behave like conventional AC signals. Impedance of wires and capacitors...

Well - yes - you don't need to be an rf engineer. Basically we are trying to keep it out - and avoid it causing highly non-linear effects in semiconductor junctions etc. - rather than wanting to build wireless networks. (Unless, of course, you are designing a wireless system or similar !)
Basically we are doing 'EMC' whether for quality / signal integrity or compliance reasons.

And yes, it does help to have some understanding of the differences to our 'normal' low frequency electronics.
Impedances can be huge and so sources will often approximate current sources rather than voltage sources.
Get those ferrites out and pass the copper tape

 

[Matt Syson]

There a good reason why Neutrik (may be others and the 'proper' EMC guys go to such extremes to make enclosures 'EMC tight'. From that jensen document i would add that there should be a 'C1 for each mic input (probably with a good VHF bypass cap as well) to minimise all loop distances.
since carrying portable 'multi GHz' transmitters seems to be all the rage nowadays, sloppy RF immunity of older gear needs addressing and designing out of any new builds. Bear in mind 'balancing' and CMMR is near non existent at frequencies of even 200KHz and above. I remember having to 'tweak' common mode rejection by twisting, or untwisting bits of solid core wire a cm or two long 40 years ago for one manufacturer before acceptance by the customer.

 

[sahib]

It's kinda funny that everyone admits that RF is not their specialty and then makes proclamations about what the right thing to do is. I think it is easy to underestimate how RF can get into places you might think it shouldn't. RF does not behave like conventional AC signals. Impedance of wires and capacitors and the effects of parasitic capacitance of parts is different at RF frequencies. At least I would not be shocked if a little piece of wire hanging off of some pin 1 has just the right inductance and capacitance to pickup your neighbors outdoor wireless speaker transmitter, get amplified, demodulated by the grid of a high gain pre and then suddenly you're tube pre is playing Smash Mouth.

I am assuming you are directing this at me since you have quoted me.

I do not mean to offend you, and with all due respect, but I can clearly see that you lack basic electrical principals.

What is a conventional AC? Is there an unconventional AC?

With your above statement you are talking about a dangling bit of wire emitting RF. Here we do not have a bit of unterminated dangling wire.

How can a bit of wire emit RF when its RF receiving end is tied to the safety earth and essentially at 0V potential?

The noise current developed at XLR pin 1 is also the noise current developed on the case chassis, as pin 1 is tied to the chassis, and the chassis is at safety earth ground potential. This is the low impedance path for the noise current.

The wire that is connected between pin 1 and phantom 0V reference point (at the PCB) is a high impedance path to the noise current. Moamps took the trouble to work it out that, the (RF) noise current will have to have a frequency of >7GHz before it can leak into this wire. At that point, that wire is the last of your concerns because that 7GHz noise current will also leak into the analogue ground through the chassis stud point as the analogue ground is tied to it.

 

[moamps]

........ But just for one data point, the Mackie VLZ mixer, which is actually a very well designed mixer, does appear to use the chassis for phantom power return...

I have a good knowledge of RF theory and technology but I will not comment on your first part of the post, but I would like to know exactly from which VLZ mixer you took that schematic.
Here I have a schematic of a 1604 VLZ mixer where it is clearly seen how the grounding in the mic preamplifier and power supply circuit is made .

 

[Bo Deadly]

I have a good knowledge of RF theory and technology but I will not comment on your first part of the post, but I would like to know exactly from which VLZ mixer you took that schematic.
Here I have a schematic of a 1604 VLZ mixer where it is clearly seen how the grounding in the mic preamplifier and power supply circuit is made .

The schem fragments I posted are from the Mackie 1202 VLZ3. Your fragments are from two series prior. So they changed the circuit from using analog 0V to using the chassis for phantom return and I think it's safe to assume that the Mackie engineers knew a lot more about what works in practice and in theory than either of us combined.

 

[moamps]

The schem fragments I posted are from the Mackie 1202 VLZ3. Your fragments are from two series prior. So they changed the circuit from using analog 0V to using the chassis for phantom return

Unfortunately I don't have a schematic of that mixer but I have one from his younger brother (402VLZ3) and here's what grounding looks like. This one was obviously designed by some novice engineer. Can you post a 1202VLZ3 power supply schematic?

.... I think it's safe to assume that the Mackie engineers knew a lot more about what works in practice and in theory than either of us combined....

I think that way you can only speak for yourself.

 

[Bo Deadly]

This one was obviously designed by some novice engineer.

I don't think that is a fair criticism. The ground scheme might seem too simple but realize that that particular unit is a 2 channel + 1 stereo compact $89 line mixer. I'm surprised it has phantom at all.

Can you post a 1202VLZ3 power supply schematic?

Mackie (AKA Loud Audio) posts many of their schematics on their website including the 1202 VLZ3 and most of the VLZ series minus the latest VLZ4. We really shouldn't be posting full schems of a living company without permission. Especially when they actually have a enough spine to post schems on their site.

 

[JohnRoberts]

Unfortunately I don't have a schematic of that mixer but I have one from his younger brother (402VLZ3) and here's what grounding looks like. This one was obviously designed by some novice engineer. Can you post a 1202VLZ3 power supply schematic?

Greg Mackie, and IIRC his senior engineer at the time, Rick Chen (?) both worked at Tapco before Mackie... Not exactly novice engineers.

[edit- some of their mic pre designs were were known for poor RFI rejecton /edit]

JR