Crosstalk in phantom switching

[JAY X]

Hi!.

I'm currently building a 500 series preamp. For switching the circuits i use alps dpdt switches and relays powered between +/-15v rails. Got no problems switching everything this way...except phantom power.
The phantom rail i have comes from an ssl9k psu preamp kit i built years ago.
I got 43,5volts and 3,89ma.

The switches are alps spuj dpdt rated at 30v/100ma. In theory good enough?

The phantom rail to the preamp inputs is filtered with 100R/100uf63v.
Resistors are 6k81 melf. The phantom relay is an omron g5v1 24v.

This is a crosstalk problem. How to minimize? I will upload a picture asap.
Thank you for your advise!
Jay x.

 

[Khron]

Crosstalk between what and what?

 

[JAY X]

Hi!,

I suspect to have crosstalk between supply rails and 48v rail... But perhaps it is another problem, could be ground noise...

When i switch on the phantom, i got "pop" noise, and when i switch Phantom off, i hear a short blowing noise, that sounds as an expiration. Maybe is the electrolytic cap discharging...

 

[JohnRoberts]

Phantom power is provided common mode so as long as 6.81k resistors are precision that should not be a source of crosstalk. If worried you can use parallel decoupling, additional 100 ohm and 100 uF caps.

JR

 

[JAY X]

Hi!,

Ok, i will check the values of these resistors. What do you think about poping and breathing noise when switching on (poping), and off (breathing). When i switch on the pad or the phase relays i also get poping noise. When i pull out the 48v phantom cablefrom the psu, these noises disappear...

 

[JohnRoberts]

Some amount of transient noise is normal when switching on phantom. If the "breathing" noise after power down is with an active mic connected, that could be normal too... some active circuitry inside mics can become unstable at below normal operating voltages.

JR

 

[JAY X]

Ok...so, when phantom is active, switching in the pad and/or the phase switch, can also produce those transients?.

 

[JohnRoberts]

Ok...so, when phantom is active, switching in the pad and/or the phase switch, can also produce those transients?.

Not because of phantom power, but switching polarity (phase?) and/or pads while signal is playing may cause clicks.

When you first energize phantom power you are applying tens of volts to the input of a high gain stage looking for mV level audio signals. The common mode rejection can keep that huge CM DC step down to a mild tick/thump but it is rarely completely silent.

JR

 

[JAY X]

Ok!. The 48v rail filter cap is tied to the analog ground. May I gain something (reduce noise), tying the cap to chassis gnd instead, like in soundcraft consoles.?

 

[FIX]

To get the quietest noise figures, try these things:

1. Have the mic gnd/shield pass through the patchbay isolated only its jack. If it's normalled to the jack below, then connect the grounds together.
2. For the mic pre input jack, have the gnd/shield go all the way to the mic pre, without connecting to chassis or any other ground. If no patchbay, then have the end/shield go all the way to the mic pre. Then in the mic pre, have it join the system ground.
3. The phantom power filter cap must be on that same ground reference, not chassis. 10uf to 100uf is plenty, 63V minimum.
4. The 48 volts should go to the switch directly, then to a 510 ohm resistor, then to the cap, then to the resistor (6K19 to 6K98 depending on where you grew up) split to + and -. These two resistors need to be as matched as possible if you want to reduce clicks and pops, though better than 1% is probably a waste of time. The 510 ohm resistor will ramp the 48V into the cap, lower the thump a bit and minimize the current surge on the switch. In the old days, we used 10K, but when the TLM mic series came out, they drew too much current.
5. After that, if you are using a solid state input stage, make sure you have 10 ohms in series with both lines going to the DC blocking caps feeding the stage, and use diode protection from after the caps to the voltage rails of the amp. The current at this point can be quite high when the 48V switch is engaged, as the input caps are like a direct short until they charge, the diode prevent this surge from hitting the preamp input.

The thump or click, like John said, is almost impossible to eliminate, as the 6K19 resistors have to match, the input caps have to match, and the impedance of the unit stage has to match, which is unobtanium. When I designed the Sunset Sound Tutti mic pre, is have a delay circuit for both 48V and Line in that waits, as it mutes the amp, and then comes back on.

 

[abbey road d enfer]

4. The 48 volts should go to the switch directly, then to a 510 ohm resistor, then to the cap, then to the resistor (6K19 to 6K98 depending on where you grew up)

Can you elaborate on that?
I've never seen anything different than 6.81k for P48.
If you install a 510r in series before, you have to substract 1020r to the 6.81k, for an actual 5790r. I don't know what's the closest normalized value, but anyway a difference of even 200 ohms won't make any difference to the practical operation.
What I'm really curious about is 6.19k vs. 6.98k (and why not 6.81)and the relation with nationality.

 

[FIX]

I've seen both, the comment was a joke, and installing one 510R won't subtract 1020R. At the maximum load of 7 ma, which is what the TLM drew it drops a max of 3.5 volts, which is well within the phantom power specs required to power mics, except maybe the older AKG C451E or Earthworks mics, which draw 10ma. Most mics are in the 3ma range, which drops only 1.5V, keep in mind that a 3ma mic will drop the 48V to 43V, including the series 510R. I have seen 6K19 when a series of 510R is used. I have also used 310R. The reality is that the total load of the phantom supply is 14K across the mic, which will drop the signal level no more than .1dB, so if it's 12K, it might drop it .12dB. The Earthworks mics have a lot of headroom, so the lower value would assist in the extra current draw.

I've done it since I owned API when they came out with the higher current mics. I also put the LED indicator after the switch, so it shows actual 48V and not a switch contact hooked to another supply. The other up side of this is that the LED will light up if the mic pre is being powered from another unknown source. That saved our butts one night while recording Dave Mathews Band at RFK stadium, and we noted that the LEDs were dim on many channels, (Ultrasound mults mic lines instead of using a transformer (loss of low end VS loss of high end with a splitter) and as the band played, some channels lost headroom. By simply turning on the 48V on the record pre's, we solved the problem.

I have done the same from 1985 with API, when I owned Tonelux, and with every of the 10 mic pre's I've designed, and currently with FIX audio. There are probably 25,000 preamps out there with it, so I don't see the problem. Having the series resistor will ramp the charge up, softening the POP, and will spare the switch contacts of hitting a dead short with maybe an amp of 48V...

 

[abbey road d enfer]

I've seen both, the comment was a joke, and installing one 510R won't subtract 1020R.

Yes it does, if you want to be in accordance with the standard.
Now I understand it doesn't make much difference in practice.
At Barclay Studios, we had an API. When I designed the phantom powered DI's there, I found that what worked right with the Neve didn't with the API. I found that the 100R resistors that fed the 6.81k were actually 10k! Nobody had noticed since this studio used mainly dynamics and valve mics. The U47FET didn't seem to work well.
I know it's gross; it's the equivalent of replacing the 6.81k with 26k!
Even the 510r with no compensation on the 6.81k could result in a false trigger with those Neumann mics that use(d) variable phantom feed to select remotely gain and/or pattern.

I have done the same from 1985 with API, when I owned Tonelux, and with every of the 10 mic pre's I've designed, and currently with FIX audio. There are probably 25,000 preamps out there with it, so I don't see the problem. Having the series resistor will ramp the charge up, softening the POP, and will spare the switch contacts of hitting a dead short with maybe an amp of 48V...

Of course a low value resistor is good practice. 100r results in a deviation of about 3%, which is acceptable and within the 4% of the standard, but 510r without compensation is about 14%.

 

[Winston OBoogie]

...

 

[JAY X]

Hi!

I take note of all the good advise in the posts!. Now that I know a bit more about phantom, my main worry is noise ripple in my old 9k psu. With more preamp gain, more psu noise ripple gets amplified... So I will build a better psu in the next days, with Lm317/337 and a voltage doubler with a TL783. Despite, if the preamp ends up in a 500 series rack, I hope these racks have well filtered rails...

 

[Khron]

What about CMRR (common-mode rejection ratio)? Shouldn't that virtually nullify whatever minor ripple there might be on the phantom power rail?

 

[JAY X]

What about CMRR (common-mode rejection ratio)? Shouldn't that virtually nullify whatever minor ripple there might be on the phantom power rail?

Hi!

If the psu is noisy, you will have noise and ripple in all rails. And in a Mic preamp the noise gets amplified with more gain... This is what i think, and why I will build a new psu better filtered. Of course good opamp rail decoupling is always necessary. And also capacitor filtering at the supply rails near the edge connector...

Jay x

 

[abbey road d enfer]

What about CMRR (common-mode rejection ratio)? Shouldn't that virtually nullify whatever minor ripple there might be on the phantom power rail?

That's how it is in principle, but a number of factors come into play to ruin a good CMRR. Remember that the actual CMRR is that of a complete connection, i.e. the source, the receiver and the wiring.
CMRR tends to decrease with frequency, so higher harmonics of the mains frequency (which are more audible than lower ones) are less attenuated than the fundamental and 2nd-3rd.