Old: A Direct-Coupled Input-Capacitorless Active Mic Preamp

[mediatechnology]

I've rebuilt the prototype and have a bunch of updates to provide soon but stumbled upon an interesting thing about microphones while poking around this circuit with the 'scope:

Some phantom-powered mics have a significant low frequency common mode component due to internal dynamic signal currents. Other's don't.

Handling noise in particular increases current draw significantly in some mics to the point you can see it modulate the Vcm servo slightly.
Both phantom legs get pulled closer to ground together - in common mode - as signal current increases.
It does seem to follow...

It's of absolutely no consequence in this circuit since Vcm doesn't do anything now and the LF common mode rejection is as good as it is at mid-band.
The Vcm servo only drives the floating rails and the ICs riding them have significant PSRR.
As to CM rejection I'm seeing typically -80 dB at 40 Hz low gain and I suspect it can be better on a real PC board.
CM signals are passed through this preamp at unity gain up to the cross-coupled differential output where common mode rejection is realized in a "double-balanced" configuration.

But I have to wonder if active Mic preamps with capacitor inputs that have either low-value or mis-matched input capacitors convert some of this mic-induced LF common mode signal to differential.
It appears to mostly be sub-sonic in the sample I have that produces it.

[ricardo]

[bSome phantom-powered mics have a significant low frequency common mode component due to internal dynamic signal currents. Other's don't.

Wayne, will you share which mikes do this?

Gotta admit us evil mike designers sometime ASSume supa dupa preamps ... though not usually as evilly as ECM8000

We gotta assume supa dup transformer inputs too which would be seriously upset by mismatched currents on pins 2 & 3

[mediatechnology]

Ironically the ECM8000 doesn't do this.
Though it does pull unequal static current the dynamic current looks fairly constant.

The Audio Technica AT3525 does have significant CM "wiggles."
It has a low DC resistance from pins 2 to 3 and appears to have a transformer-coupled output.

I suspect the the lower the DC output resistance the more the dynamic current appears in CM.

I need a full Mic closet to check this. I may have one...

Makes the case for Mic Preamps with excellent LF CM rejection.

[JR.]

Interesting... I used to do a lot of tone burst testing, because it would reveal things that static testing didn't. For a mic/preamp i can imagine other obscure things like popping plosives, wind noise, etc.. but these are quickly getting esoteric and arguably gray area off-sheet for expected performance. Better is always better tho.

JR

[emrr]

We gotta assume supa dup transformer inputs too which would be seriously upset by mismatched currents on pins 2 & 3

Geez, yeah, old school phantom through the transformer center tap versus pair of 6K8's. Very different end result.

[mediatechnology]

My thinking is that mics that have internal transformers, and thus a low DC resistance between pins 2 and 3, are more likely to show this effect in common mode since the winding equalizes the LF currents between legs.
Also, these mics may tap phantom off of their output transformer center tap as well (producing CM modulation) rather than resistors to each leg.

I hope to post pics today of the preamp protoboard.

Not including output drivers (2X THAT1646 in the dual class-A) there are only six ICs.
The signal passes through only two 470 nF PP film caps; four total with 40 Hz -12 dB/oct HPF engaged.
The preamp is fully-balanced input-to-output.

THAT1570
5534 Vcm Servo
OP2277 Differential Deboo Servo
LME49860 AC-coupled line receiver/HPF/Opt post-preamp +10dB "ribbon" gain stage
2X THAT1246 Double-Balanced Common Mode Rejector

I figure the 5534 and HV complimentary pair are about the only gizmo that's being added to eliminate input caps.
So it's really not a complicated thing...

[mediatechnology]

I have another mic, a LDC that is bigger than the 2535 that I am going to send Wayne. Don't know if it is SS or xfmr out though. I am sending is as ransom/bribe for the Leader lMV185A Wayne sent me
Things have been absolutely nuts here, and I meant to get this out a few weeks ago.

Thanks Roger! I'm just now getting back online today.

[ricardo]

My thinking is that mics that have internal transformers, and thus a low DC resistance between pins 2 and 3, are more likely to show this effect in common mode since the winding equalizes the LF currents between legs.
Also, these mics may tap phantom off of their output transformer center tap as well (producing CM modulation) rather than resistors to each leg.

This is the favoured method if the P48V mike has a transformer.

I suspect my old Calrec stick mikes would be truly evil in this respect as they moved into Class B at very high spls and had extended LF response too. Mea maxima culpa.

[mediatechnology]

I made a couple of recordings. I think this thing is pretty quiet.

The first is the sound of a 22 pF monolithic ceramic capacitor being dropped from workbench height at 75 dB gain using the phantom-powered AT-3525.
Think the Sprint pin drop commercial. Beware that it reaches full-scale.

http://www.waynekirkwood.com/Content/DC ... B_Gain.wav

The second file is the gain being swept from +15 dB to +75 dB with the AT-3525.
The differential Deboo servo works like a champ.
The clicking you hear is primarily acoustic.

http://www.waynekirkwood.com/Content/DC ... Change.wav

[mediatechnology]

Another Smoking Gun

I modified the "DC-Coupled Input Capacitorless Preamp" to be a conventional preamp by adding 47 uF/63V coupling capacitors.
Otherwise the circuit is identical to the Input Capacitorless version.
I turned off the common mode servo: The supplies no longer float since it's AC-coupled.
The differential servo was not modified - in an AC-coupled preamp the injection resistors could be made 1M-2M2 vs. 100K.
The gain is ~65 dB.
The FFT was averaged over 50 samples.

Sure enough folks there's a huge difference in 1/f noise with Phantom applied.
I'd been pretty certain that leakage current played a role in noise performance and that seems to be the case below 100 Hz.
At 10 Hz, it's about 25 dB quieter with phantom power off.

I'd be willing to bet that almost every active preamp is spec'd (though not necessarily QC'd) with phantom off.


AC-Coupled Preamp Phantom Off 47uF Per Leg 150R Source Impedance


AC-Coupled Preamp Phantom On 47uF Per Leg 150R Source Impedance

In the Direct-Coupled Input-Capacitorless version the noise performance with phantom on or off is virtually identical.