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.
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

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

Post by ricardo »

After dismissing Wayne's design as far too complicated for dis beach bum's KISS philosophy ... I'm looking with a very jaundiced eye at my (actually Professore Angelo Farina's) MOTU Traveler Mk2 which is showing leaky P48V caps after 7 yrs.

It's galling as I have a 1970's Calrec P48V PSU which is still deathly quiet after more than 30 yrs.

The Traveler uses PGA 2500 and has 1dB NF for 150R source when new. But it also has evil Schottkys as protection that go noisy. I replace the Schottkys on one noisy channel with 1n4004 but as the originals were SMD, dis blind old beach bum ripped up the PCB. :o

So I'm reluctant to attempt changing the caps. :(

It only takes a couple of such incidents to make you paranoid about P48V caps. How about you get THAT to put your Input-Capacitorless Mic Preamp on a single chip? :mrgreen:

1 +/-0.1 dB gain steps via simple control would nice too.

Also found THAT 1510/12s vary from chip to chip by up to 0.5dB .. probably cos although laser matched trimmed to very high precision for CMRR, the absolute values may vary. The solution if you need accurate 0.1dB channel to channel matching is to buy twice as many 1510/12s as you need & match them. I need 4 for an Ambisonic mike preamp which isn't too difficult and they are cheap enough.

MOTU is one of those companies where all design is farmed out so there is NO internal Tech support at all. They haven't a clue what's in their product. Traveler Mk1s seem to remain quiet. Dunno about the current Mk3
Last edited by ricardo on Fri Oct 09, 2015 6:07 pm, edited 1 time in total.
olafmatt
Posts: 44
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

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

Post by olafmatt »

mediatechnology wrote:When I get a chance I will check for this: The worst combination would be at high gain with input coupling caps.
But I should have seen it by now.
Thanks Wayne for considering my "worries". - Some years ago I also experimented with a input-cap-free preamp with a floating input stage. And it worked well with some mics and had oscillating DC offset with other (TLM 103, for example). So maybe I was having that trouble with interaction. Need to find my circuit from back then and re-do the math (and check inside the TLM103 what value the caps are).

At least the good thing is that switching the phantom caps out of the way in a conventional preamp hopefully means that only mics without output caps will be connected then. Well, unless somebody connects one of those condenser mics that can also be powered by a battery.

Olaf
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

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

Post by JR. »

The elephant in the room about eliminating caps from the mic preamp path is that many phantom powered mics have caps in their audio path inside, so this design approach is only reducing caps in the path not removing them entirely.

I have been scribbling about this approach for a very long time, but never considered it for a commercial design because of the likelihood of such interactions (not stability as much as DC offset issues. and maybe some I don't anticipate). In large scale production you will find all the gotchas and it will be difficult to deal with them in a production design after the fact.

I tried to get friends still in the business to make a premium mic preamp without phantom power for the raft of crude passive mixers that need a mic preamp stage for make-up gain, w/o any need for phantom power.

JR
Cancel the "cancel culture", do not support mob hatred.
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 Pre

Post by mediatechnology »

The servo question was a good one.
Thank you olafmatt for posing it.
It may be worth a third or fourth visit.

I've pretty much found that after building three or four different versions of this now it's pretty tolerant of just about anything including huge deliberately-induced DC imbalances.

Though microphone DC offset has been addressed ad nauseam it still serves the useful purpose of throwing occasional cold water on the project.
Its a fall ritual and I think re-hashing that one more time for sport is complete and utter bullshit.
It's been discussed.

Everyone told me the Behringer's DC-imbalance was going to be a big "problem."
It wasn't.

Next up?
The TLM-103 as deal-breaker.
We've already talked about the TLM-103...
Somewhere I have a sample TLM-103's DC offset measurements.
I duplicated it's large offset and it was a complete non-issue. Like the Behringer.
Somebody send me a TLM-103 and we'll know for sure.

Then we'll move on to some other alleged "problem" microphone.
The hypothetical and non-existent elephant in the room makes a convenient excuse for doing nothin' methinks.

And that capacitor thing:
Are we really supposed to accept the fact that because "some" microphones have caps in their outputs that eliminating them elsewhere isn't worth doing?
Come on, that's BS.
That line of thinking justifies still having output coupling caps in power amps.
It's like Walmart racing to the bottom of retail and trying to pull everyone else with them.

We build protoytpes to test things to avoid dealing with the hypothetical.
I need people who have tested this circuit with microphone closets full of examples to provide me with something real that doesn't work.
Or I need to put this on a PC board and visit that closet.
Then we'll talk about offsets again.

Servo stability with AC-coupled sources may be an issue but I sure haven't seen it.
When experimenting with using AC-coupling looking at 1/f noise I suspect if there was servo instability I would have seen it.
But I'll look again.
Someday.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

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

Post by JR. »

I am not trying to throw cold water on this (just the marketing premise) and I have been supportive for a long time (as long as I've known you). It was this exact project on another forum, that triggered my initial contact with you, since I had similar musings. I think YOU need to make and sell this. IMO it is a merchantable selling hook, as long as you ignore the caps hidden inside some mics.

I have been bit in the ass enough times when interfacing with other peoples products to remain wary, and my discussions with others in the console business reveal a low tolerance for risk.

In modest volume selling singles or duals even a handful of customer issues will have minimal effect. And you could charge enough of a premium over normal preamps to provide good customer service.

This could be cheaper to implement in a full console with common supplies etc, while the potential for interface issues increase when using 10x-20x at a time.

Just do it.... In fact the rise of Digital consoles suggests that perhaps a DC coupled stage box front end that outputs to established digital mixers might be a product. In that case you could fly the A/D convertors up to phantom supply too, then use optical to shift the digital data down to the outside world. You might be able to hack into an existing stage box to grab their digital controls over your analog front end.

JR
Cancel the "cancel culture", do not support mob hatred.
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 Pre

Post by mediatechnology »

Just do it.... In fact the rise of Digital consoles suggests that perhaps a DC coupled stage box front end that outputs to established digital mixers might be a product. In that case you could fly the A/D convertors up to phantom supply too, then use optical to shift the digital data down to the outside world. You might be able to hack into an existing stage box to grab their digital controls over your analog front end.
Thanks John. I think the "Universal Input," modified for DC-Coupled Mic input, would be ideal as it only requires one socket (or a GPI-switched input to support two connections) for Digital Mixers and Stageboxes.

Universal Input: viewtopic.php?f=6&t=514

The mic output DC offset issue is dead.
Servo tuning may not be but I have the flexibility that if I find some particular mic topology problematic I can just fix it.

I think I recall the TLM-103 output structure from olafmatt's previous post was op amp output > 47 uF > 47 Ohm per leg.
The AC output impedance is also lowered by the phantom pick-off resistors internal to the mic.
We don't know what the DC resistor values are.
At 7 Hz (the highest servo fc) the AC-coupled TLM-103 has an AC output impedance around 1K if not lower.

The lower the total termination impedance across tip/ring, the more stable the servo.

A 1K mic Zout in parallel with the preamp 5K Rterm is a combined 830 Ohm termination.
In previous servo tests I found stability with Rterm values as high as 10K.
Also, the lower the DC Rterm, the greater the self-equalization of developed offsets.
olafmatt
Posts: 44
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

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

Post by olafmatt »

mediatechnology wrote:The mic output DC offset issue is dead.
Yes I know. Sorry for having mentioned the TLM103... in the context of some other potential problem.
mediatechnology wrote:I think I recall the TLM-103 output structure from olafmatt's previous post was op amp output > 47 uF > 47 Ohm per leg.
The AC output impedance is also lowered by the phantom pick-off resistors internal to the mic.
We don't know what the DC resistor values are.
5k6 per resistor. And the mic I was talking about when mentioning the component values is the Gefell MT 71s. But I don't have access to that mic and I haven't build your proposed circuit... that is why I was asking whether you were at some point looking for that problem.
mediatechnology wrote:At 7 Hz (the highest servo fc)
So assuming a 60dB gain range the lowest fc is then 7/1000 or 0.007Hz. And if the microphone's output caps together with whatever resistance comes after them (phantom pick-up, phantom insertion, Rterm, Rbias...) gives you an fc somewhere between 7Hz and 0.007Hz then you will get interaction / servo oscillation at some gain setting (where the resulting fc of the servo is too close to the fc of the caps/resistance combination).
I'm not saying that the servo will not work or that some mics will produce too much DC offset... I know it's a beaten dead horse. I'm just worried about the DC servo's correction voltage "leaking" out of the front of your preamp. Please note that I put "leaking" in inverted commas... I'm not saying you are wrong. It is just that I get this interaction with "normal" preamps with phantom blocking caps. And I thought why doesn't he get them when in effect the caps are still there, just inside the microphone (at least for some microphones, depending on their internal circuit).

Olaf
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 Pre

Post by mediatechnology »

I'm just worried about the DC servo's correction voltage "leaking" out of the front of your preamp.
OK. Are we talking about the DC component or an AC component?
I realize that you said DC servo but it's output has both.

The AC component from a servo exist to similar degrees whether the preamp input is AC or DC-coupled.
I don't know if anyone has explored the subsonic effect of this but it is there in both topologies.
Depending on the amount of servo range needed one approach may have more attenuation than the other, but not necessarily so.
If correction is injected into the base, emitter or some more creative place downstream, there's still that pesky AC component.
What matters is that its stable, works and doesn't create bad response errors.

Then there's the question about DC current leaking out the input.

Assume for the sake of discussion that the preamp front-end itself has no offset to correct.
If a bridge error is measured resulting from a mic DC imbalance - and that error is corrected by injecting an opposing (low noise) current, what is the harm?

If that error is, for example, due to simple resistance mis-match in the mic, how is correcting the resulting DC error any worse than having a mic with perfectly-matched resistors?
Or precisely balancing the emitter currents of a Schoeps output?
Transformer output mics, dynamics and ribbons self-equalize input imbalance owing to their low DC source resistance. No DC mic correction is applied to them except for the very small amount required to correct preamp DC offsets.

In the real world, the preamp itself has DC offset and some additional correction is applied to also correct preamp input offset errors.
That current gets lumped in with the offset from the mic itself.
But how much?

Mics capable of producing DC errors - and that need large amounts of correction - dwarf the input error.

The proof of that is the Behringer measurement mic.
The Behringer offset correction requirements are more than 10 times that of the front-end.
The Behringer offset, though "large" still only requires about 70 uA to bring it into balance.
That 70 uA is itself a mere fraction of it's total phantom current draw.
The mic offset current error is 10-20 times less than the phantom current.
Pushing it into equilibrium is harmless.

When phantom is on you have up to 7 mA/leg potentially leaking out of an input.
Take that 7 mA and add or subtract up to 120 uA/leg for the servo correction that can be required.
The ratio of total current available to maximum servo current is about 60:1.
Should we worry about that? No.
Servo current is a drop in the bucket compared to potential phantom current.

Consider a conventional mic preamp.
Assume there is no servo, no input coupling caps and no phantom.
Think THAT1583 DC coupled and about to be connected to a ribbon mic.

With the input open there is a 2-5K effective Rterm. Assume 5K.
The 1583 Ios is about 1 uA. 5 mV will be developed across the open input.
That's voltage leaking out the input. And its in any bipolar input preamp.
Plug in the ribbon mic and there's 1 uA leaking from the input across it's outputs.

In a conventional capacitor-coupled input with phantom powered on, capacitor leakage current is going to be a minimum of 1-3 uA.
At turn-on that current is huge and for about 5-15 minute afterward still big and falling.
That's current leaking out the input.
And its with any AC-coupled active mic preamp using electrolytics and phantom power.

There's almost always some current flowing out the input.

I think that when you put it into perspective what current the servo does inject is small and harmless.
olafmatt
Posts: 44
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

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

Post by olafmatt »

Still not exactly what I'm talking about.

Look at the AES paper "DC Servos and Digitally-Controlled Microphone Preamplifiers" by Gary Hebert and read "3.1. Frequency-Domain Effects". In that paper Gary discusses the effect of the servo (in combination with phantom blocking caps) on the frequency response. Unfortunately that peaking can not only be triggered by infrasonic input signals but also by the servo itself. So the preamp alone without any signal on the input can start to resonate at some low frequency. What you see is very low frequency "oscillation" on the servo's output and of course also "oscillating" DC offset at the amps output.

Now in your case where these caps are not in the preamp but they might be in the mic, you don't have control over how large (or small) they are. So no way to make sure you don't get that peaking and/or the potential low frequency oscillation of the servo.

Olaf
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 Pre

Post by mediatechnology »

I linked to that paper here some time ago before the AES made THAT pull it off their website. Later it came back.
So its the AC characteristics worrying you...

Lower the Q of the servo at high gain to reduce peaking if its a problem for you.
That's what I did.
As Gary points out in the paper that makes the servo slow at low gain.

In this application I didn't find the slower low gain response to be a problem.
This is simply a question of tuning and its interaction with the input network or cap-coupled mic output.
It is not a deal-breaker that cannot be changed if needed.
Why make it one?

I haven't asked him, but I suspect the reasons THAT published this is because customers are beginning to use the gain controller in VCA-like applications requiring fast gain changes.
People likely tried to speed up the existing response (resulting in a Q >0.8) and got themselves into trouble.
IIRC there's a mic preamp out there using the 5171 as a "limiter" and another one in an automatic mic mixer.
THAT typically doesn't spend much time on stuff like this unless there's a large customer needing it.

Because the servo I used is purposely made quite slow (at low gain) I focused on its recovery behavior a lot both switching and sub-sonic.
I didn't see great harm keeping the servo slow at low gains because the gain changes I anticipate ramp slowly and the clicks at low gain from it not "catching up" aren't as audible.

I've built this with both a rotary switch and a 5171 gain controller.
The "catch-up" requirement is minimized with mechanical switching, and based on the way I control the 5171 with an onscreen knob, its rate of gain change is limited too.
Perhaps I should look at the servo response with AC-coupled inputs (again) but many of my signal tests were done that way because I don't have a floating transformer-output generator.

One thing people should be aware of is that some servos - differential ones using two or more capacitors for example - aren't necessarily monotonic in response to step changes due to capacitor and resistor tolerance.
That can contribute to stability issues as well.
The differential Deboo uses one cap relying solely on resistor matching.
Gary, out of curiosity, simulated my differential DeBoo for me to check its stability, but not in the entire circuit which would paint the entire picture.

When I'm in the mood and have the time to humor everyone I'll revisit this.

Until then I ask: "Where are you headed with this?"
If you have a better idea I think it's only fair you pony it up.
I've given a lot more here (and elsewhere) than I've taken back.
Show us what you're up to...

I also anticipate the predictable arm chair response that this job should be dedicated to a microprocessor...
Been there, heard that, but no one has sent me code.

The code I could use - it would be very useful - is a rotary encoder and pushbutton/display board that has an SPI output.
That would be a very worthwhile effort similar to what Alex did at Innersonix.
Post Reply