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

[mediatechnology]

I have been repeatedly told by a lot of people that an input capacitorless phantom-capable "active" mic preamp cannot be done.

I'm here to tell all of them that it has been done. Twice already.


Input-Capacitorless DC-coupled Preamp 2012 Protoboard

I promised to re-build the prototype and make a few improvements.
Here it is.
It works, I've used it with real microphones, and it doesn't blow up.
I decided to take a pic of it while I clean up the bench.
I don't like having a messy bench with almost 100V floating around. (The supplies are +72, +48 and -20.)

One of the first comments I always get is "Well, some microphones pull unequal current out of each leg. You'll be amplifying all that offset. XYZ tried that and gave up."
Excuse me, but isn't that what servos are for?
The differential servo is capable of correcting - as it's currently built - about 100 uA/leg or 200 uA of total error. It can easily be made to provide more correction current.
The common mode servo keeps things from blowing up by flying the supply rails based on the microphone's DC quiescent point.

It's particularly fun to switch phantom on and off: The recovery time with an actual microphone is about 1 sec from off to useable, monitorable, audio.
Shorting legs to ground to create phantom faults is also a thrill.

Microphones with transformer outputs will typically equalize the offset due to the low DC source resistance.
Microphones with direct-coupled outputs e.g. emitter followers will also have fairly low voltage offset errors.
Microphones with capacitively-coupled outputs current errors will usually be based on resistor tolerance.
The only microphone I've found that pulls all it's current out of one leg - and doesn't meet any standard - is the unbalanced Behringer ECM8000 measurement microphone. This preamp is not for it.

Pardon my temporary use of carbon film resistors while I tweak values.

From left to right are the passive common mode integrator (orange cap), 1510 preamp (buried under the gain switch leads), OPA2272 "Deboo" non-inverting integrator (orange cap), LME49720 low-gain offset trim stage and common mode servo and finally the flying rail power stage. The heatsink is way oversized. The big blue cap is the output coupling cap.

I'll draw up schematics after I clean up the bench and play some more.

[JR.]

Surely nobody "here" told you it can't be done...

This is still unproved until it is tested by the unwashed in the marketplace, in numbers. I wouldn't underestimate the ability of the market to come up with strange black boxes loading down the input phantom supply. i think one company powers a DSP from that input, some just pretty LEDs. It's probably Ok to not support the worst offenders, of good phantom power behavior. In a console one could add a channel or two of old school conventional inputs, JIC.

In practice you probably don't need the full 72V rail... a real phantom mic will draw down the input, and nominal AC 0V to below 48V. If it doesn't then it doesn't really need to be up there in the first place. I guess it's not worth the trouble to scrimp for only a few volts.

When do you start production... ?

JR

PS Don't tell the customers that there is a cap in the servo. Let me know if you want me to design a microprocessor based servo (no caps baby).

[mediatechnology]

Surely nobody "here" told you it can't be done...

No, not here. But I'm amused at the number of times I have heard it after I have already done it.

Not too worried about anyone wanting to use phantom as a power supply for their DSP gizmo.

When do you start production... ?

I'm on SN002.
As soon as I order more protoboards and some potting compound.

In practice you probably don't need the full 72V rail.

I'll take 64V. Or, in another version, 54V.
I'm working in multiples of 24V right now because the math is simple.

a real phantom mic will draw down the input, and nominal AC 0V to below 48V

Yeah maybe 41+18 for 59V at 1 mA/leg.

The limits of the 6K81 are everyone's problem.

[JR.]

When do you start production... ?

I'm on SN002.
As soon as I order more protoboards and some potting compound.

This is one of the worst kept trade secrets ever...

I was liking my discrete front end for little less stuff flying around up at voltage, but nowadays I kind of like the THAT digital gain control chip, so I am flip flopping around. Luckily this is just mental masturbation for me, so I can flip flop around painlessly.

I am getting closer and closer to going over to the dark side... (DSP). Of course even that requires a clean front end capture.

JR

[mediatechnology]

This is one of the worst kept trade secrets ever...

The potting compound is just to keep things from shorting out on the protoboard during shipping..

I was liking my discrete front end for little less stuff flying around up at voltage, but nowadays I kind of like the THAT digital gain control chip, so I am flip flopping around.

You can still do that. Just need to opto the SPI...

I did do a discrete front end for the THAT5171.

so I can flip flop around painlessly

That's why I'm not clipping the leads on the 1% resistors yet. Or the 15.6V Zeners.

I'm calling this a proof of concept. Nothing more.
At the current rate of change I may have boards sometime around 2016. I started this nonsense in 2007.

When you get down to it, one of the outboard dual op amps and half of the second one allows you to eliminate Cgain.
Half of the second op amp and the complementary BJT pair fly the rails and eliminate Cinput.
There's not a lot of parts overhead to eliminate the input caps; most of the "stuff" is to eliminate Cgain and deal with the resulting offsets irregardless of their origin.

[mediatechnology]

I found this quote from Professor Brad Wood aka "bcarso" from 2007.

In the midst of thnking about servos again I did a bit of work on the idea of replacing the differential integrator with a noninverting integrator (apparently first suggested by Deboo, it's basically a Howland current generator with a cap to common). I've done this before but never analyzed it in detail. It turns out that with suitable choice of resistors it has almost as low of noise gain as the differential integrator, and if the input bias current of the opamp used is negligible can have lower noise overall when part of the negative immitance converter at the core of the Howland circuit has lower R's in the inverting portion. It requires accurate resistor values but has the advantage of needing only one capacitor.

In the diff integrator the effect of cap mismatch is a longer settling time with a nonmonotonic step response, so there is something to be said for a one capacitor circuit.

The 1C Deboo/Howland current pump integrator's requirement for only one capacitor vs the 2C differential integrator is indeed a big plus. I did choose to make the inverting leg's resistors much lower for the very reason he cited above.

[JR.]

I miss the regulars exchanges with Brad. He was one of a small handful out there who really understand stuff. It's nice on a noisy forum to have a critical mass of informed consensus. I occasionally get the random email from him but he doesn't post much anywhere I frequent these days.

I rather like this investigation into improved servos. The last servo I used back in the '80s in a phono preamp just pushed the pole(s) super LF.

I am percolating my idea for a digital servo... not completely crazy if there already is digital control content in the vicinity (like gain control). The best cap is no cap....

JR

[mediatechnology]

I do miss Brad too and still occasionally exchange e-mail.
I sometimes see posts on the EET Times/EDN forums from him but not often.

I rather like this investigation into improved servos.

Thanks.

I did try putting a passive pole on the output but it seemed to prefer a first-order response.
I can't see any significant noise contribution by not having a passive output pole.

I am percolating my idea for a digital servo

Auto-zero of the low gain offset at turn-on would be a good start.

[JR.]

I am percolating my idea for a digital servo

Auto-zero of the low gain offset at turn-on would be a good start.

Sorry for the veer...

The beauty of non-linear digital processing is we can use situational input to decide how and when to trim the DC offsets.

Not only can a "brain onboard" selectively decide when to tweak DC offset, if it is tied into and knows the gain control it can learn a DC offset curve vs, gain so it remembers history and can predict a nominal starting point for faster settling for every gain setting, and update that history on the fly. It can also turn itself off for most of the time when errors are under control.

Flash memory means it never has to start from scratch, more than one first time.

JR

PS I have semi seriously looked at this whole exercise commercially (DC coupled front end) for decades. From a marketing perspective, removing caps is IMO a merchantable high end hook, but the remaining caps in the servos, creates a dissonance with customers who do not comprehend the subtle difference between cap types and application.

EDIT_ PPS a recent thread on another forum about a new MIDAS digital console seems to be bragging on their having a soft clipper in the mic preamp stage as a useful musical effect for mix engineers to use????? frigin DJs.... soft clipping is still clipping... arghhh [/edit]

[mediatechnology]

...if it is tied into and knows the gain control it can learn a DC offset curve vs, gain so it remembers history and can predict a nominal starting point for faster settling for every gain setting, and update that history on the fly.

That's key.
But, I'm not sure it's worth the effort just to replace a dual op amp, a film capacitor and four resistors.
To hear clicks in this one - under the worst case of going from low to high gain - you really have to have monitor-destroying gain.
Knowing that comes from building it.

You'd never want to listen for clicky-ness in the presence of actual input.
I'm a little nervous every time I check it.
You'd never hear it unless you were looking to hear it.

creates a dissonance with customers who do not comprehend the subtle difference between cap types and application.

You can go for that segment John.

having a soft clipper in the mic preamp stage as a useful musical effect for mix engineers to use

I just did an EFX device protype using a set of mutlipliers to generate a non-level-dependent "soft clipping" third harmonic generator and it was very useful.
It wasn't actually clipping since it happened without regard to level.
But not in a preamp.
MIDAS may be trying the SSL VHD trick.