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

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by mediatechnology »

olafmatt wrote:At least 30 minutes, but that's not because of the preamp, it is because of the mic and all the electrolytic caps in there.
OK, I was going to go out on a limb and say 15 min - because of the electrolytics.
In some of the tests I did a long time ago, with only the input coupling caps, it didn't even begin to get quiet until 5 minutes.
15 minutes was where I felt comfortable saying it was quiet.

One thing I've noticed with the input-capacitorless topology is that phantom turn-on doesn't pack the expected sonic wallop. That certainly doesn't apply to a capacitor output mic which to my knowledge I don't have.
The pain is over more quickly.

Gaskell's measurements do seem to suggest that PP and PET are workable based on THD. But they seem to undermine themselves by concluding with:
Listeners seemed to perceive a shift in the overall character of the sound due to the presence of capacitor distortions. Although the experiment was not designed to quantify preference or categorize the differences heard, it is clear that certain capacitors can produce a sonic signature. Further listening tests would be necessary to properly explore the relationship between capacitor distortion and perceived sound quality. It is generally felt that THD alone is not a good indicator of whether listeners will be able to perceive distortion [8, 9, 10]. This is particularly true in the case of capacitors’ frequency dependent non-linear behavior.
I don't really know what conclusions to draw from Gaskell but its interesting.

I'm pretty sure some of the big 47µF/250V Dayton PP caps used in speaker crossovers (inexpensive and available) would measure and perform as well as an input-capacitorless direct base connection. I'm not sure about PET caps. Never knowingly tried them.

I bought some of the Dayton's to test. They're huge: About 30% larger than a D cell battery. Lovely and impressive to look at.
My thought was input-capacitorless until the freak microphone came along (triggering a servo red light) where a switch could be flipped to insert the big monster PP caps.
The big PP would be off-board (since they're big) and we might eventually find out they're so infrequently needed as to not be worthwhile.
My goal is a two channel 1 or 2U gizmo and not a large multi-channel device so there would be room for battery-sized input C's.

So let me recap and see if I basically have this right:

You built at least one of these beginning around 2008 that was DC-coupled end-to-end and recorded with it.
You haven't found a mic in the wild that doesn't work with it.
The current modulation at the input, from the servo, hasn't been found to be a problem in the field.
Its quiet and doesn't sound bad.

Do you still use it?
Would you do it again? (Possibly using an updated/improved circuit.)
olafmatt
Posts: 44
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by olafmatt »

Some caps are microphonic, so even if you don't find a lot of THD on a silent lab bench, you'd probably get some if you used them in a passive speaker crossover. Probably less of a problem in a micpre, but that really depends on physical placement of the micpre.
What I find amazing with those listening tests is that people can hear the addition of a single cap, even though there are proabably more than 10 already in the signal path of their test setup.

As for the DC-coupled preamp, I basically only did a few test recordings with it. Due to that burning-off of the common-mode offset, distortion was higher than it could have been (i.e. too much load for the opamps). And the servo was before I came up with the new circuit, so the time it took to servo away the offset at low gains was pretty long.

I sometimes think I should do it again, with the new servo and some other changes. What's stopping me is that I don't want to "leak" servo correction out of the input. - What I don't like about that servo is probably also that it does two things at the same time: It servos away internal offsets basically caused by mismatch in Vbe and on top of that it servos away the offset from imbalanced phantom current draw. That's why conceptually I'd prefer a circuit where a servo balances the circuit (i.e. counteracts Vbe differences) and either another servo takes "somehow" care of the offset that we get from imbalanced phantom draw or (even better) the circuit would simply not care about that imbalance. But as long as preamp gain is flat down to DC this is of course not going to happen.
Maybe we can come up with a curcuit that has two sets of input transistors that get summed in the gain stage. Then you could LP filter the mic signal (basically leaving only the DC offset), invert it and connect it to the second input. The gain stage would then cancel out the DC offset. I do that in the line input of my power amp (one could call it a "feed-forward servo")... but at mic level you would just add too much noise. And the LP filter would probably need to be a mic-pre in itself.
olafmatt
Posts: 44
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by olafmatt »

Just thinking aloud now: With that current imbalance spec of +/-4%, we'd need to deal with a max. of 2V DC offset of either polarity. So if we would split the gain into two stages, we could have a first gain stage (with differential output) with a fixed gain of 20dB and not need a servo. Assuming that thing is nicely trimmed (i.e. doesn't add any offset by itself) we would end up with a differential DC offset of 20V worst case (so 10V "excursion" per amplifier half). That could still (just barely) fit into +/-18V rails for low-level mic signals, since we only apply 20dB of gain (this 20dB gain would need be lowered for close-micing situations and/or hot mics, of course).

Then cap-couple to a second amplifier stage to add the remaining amplification. This second stage would then use a servo to make sure it has 0VDC at it's output.

If the first gain stage can not handle the required DC offset, then either lower the gain, use +/-24V supply voltage or even build a frontend that operates with -18/+56V supply (cascode the input transistors so that they only see a low voltage across them) and use supply bootstrapping for the opamps.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by mediatechnology »

Thinking out loud myself...
What about just servo'ing the top side of the phantom resistors?
Like we think Paul did with collector loads.
I realize that does affect the "natural" current flows into the mic.

I think I've been calling the CM servo feed-forward - I think that's a good description.

I have to wonder though if the injection of current back into the microphone - which the preamp already does to power it - is a concern similar to phono cartridge bias current.
It bugs some people a lot but no one has been able to document any ill effect from it.
olafmatt
Posts: 44
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by olafmatt »

mediatechnology wrote: Sun Jun 10, 2018 5:36 am What about just servo'ing the top side of the phantom resistors?
That would work and be quite similar to injecting the servo correction into the bases.

I think our problem is that we want zero VOLTAGE offset on the input and don't really care whether the microphone consumes unequal CURRENT. Just that with the phantom resistors these two are interlinked. - So apart from noise concerns, the "solution" would be to use an LDR (or some other sort of "adjustable" resistance) for the phantom resistors and then servo the resistor values until the voltage offset is zero. But this again assumes that the microhone has a fixed current consumption. If the current consumption is not fixed, we'll get all sorts of interactions again (which doesn't mean they would necessarily mess with the sound).
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by mediatechnology »

I tried an LDR way back (2007?) and it worked but it affects common mode rejection because the AC impedance changes.

By servo'ing the tops of the resistors we force the differential DC error to be 0V (just like base injection) but we don't upset common mode since the resistive balance isn't changed. The downside is the same with base injection: The servo Fc changes with gain.

I would like to see any evidence of harm by forcing more current into the leg that needs it to bring the input into balance at 0V.

The AC-coupled outputs would seem to be the most prone to producing offset but they would also seem to be immune from ill-effects.

The network on the right-hand side of the output coupling caps being used to derive power from the common mode 48V shouldn't care about a few mV of differential error. The common mode voltage developed by the servo modulating the differential would almost certainly be filtered and small relative to the dynamic current draw from signal.

I need to look at that MKH-20 schematic you sent in more detail. Its like one of JR's circuits with a lot of interacting parts.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by mediatechnology »

I had a thought about the servo's forcing current into the mic outputs.
Assume that the servo is providing static correction...

Pick a microhone that naturally produces a high offset.

For the input-capacitorless active topology the servo forces differential DC null at the emitters.
At the bases - connected to the mic outputs - one leg is having current sourced into it, the other current sunk out by the servo.

If that same microphone is plugged into a transformer-coupled mic preamp the DCR of the secondary will attempt to force both legs into equilibrium.
One leg is having current sourced into it, the other current sunk out by the other leg.

From the perspective of the microphone looking toward the preamp is there a difference?
Both topologies are pushing and pulling my outputs into equilibrium.
olafmatt
Posts: 44
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by olafmatt »

Yes, with a transformer input the low(-ish) DC resistance of the primary winding basically shorts the two inputs together at DC.

What I'm worried about is having an active feedback system controlling the DC offset. Because with an active system you can get things like overshoot which leads to peaking in the frequency response or even LF oscillation. Your active correction interacts with components inside the microphone, most likely with the caps at it's output. - Yes I'm aware of the fact that we didn't find this to be a problem with any mic we tested it with. I was just asking whether it wouldn't make sense to try and find a circuit topology where no servo correction action leaks out the input plug.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by mediatechnology »

Ok, so for static correction of the offset we can set aside worries about small unchanging correction currents.
LF resonance/peaking is the concern.
I think you've said that all along but for those joining us I thought that deserved some clarification.

All input-corrected(1) servos we've seen thus far have this potential whether the input is capacitorless or AC-coupled.(2)
Is that correct?

(1) Ones that let the input offset multiply and then are corrected at the output we'll call output servos. Examples are simple THAT1510 servos, the SSL 9K preamp etc.
(2) http://www.thatcorp.com/datashts/AES919 ... ifiers.pdf

Wouldn't mic preamps with AC-coupled feedback networks having a DC gain of 0 also be prone to LF resonace? (Non-interactively with the mic.)

To me this is not unique to an input-capacitorless topology - its a concern anyone should have using the input-correcting servos we seen thus far.
olafmatt
Posts: 44
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

Re: New: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by olafmatt »

mediatechnology wrote: Mon Jun 11, 2018 6:10 amTo me this is not unique to an input-capacitorless topology - its a concern anyone should have using the input-correcting servos we seen thus far.
Yes. I'm sorry if I didn't make it clear enough that this is a general concern and does not just apply to an input-capacitorless micpre. - An input (any input) is an input and if you start to inject servo correction into it to "fix" your internal circuit then your input is also an output. Any servo should do it's work internal to the circuit and should not be "visible" outside the box.

Of course we can argue whether a DC blocking cap is enough to make the servo "invisible" from the outside.
Post Reply