Mic Preamp DC-Servo: The "backwards" way

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
JR.
Posts: 2418
Joined: Sat May 24, 2008 7:21 pm

Re: Mic Preamp DC-Servo: The "backwards" way

Post by JR. » Fri Oct 24, 2014 6:29 pm

ricardo wrote:
JR. wrote:OK I have roughed up a cut and paste using an app note.
I'm not sure 1570/1573 is the best demo of your floating feedback network.
  • The 1570/1573 has its own anti-click strategies which allow direct connection between the i/p emitters to Rg
  • The 2 separate feedback scheme is very similar to how I used to implement high order rumble filters so you would need to check LF response carefully .. both for cut off as well as peaking at various gains. You may need my 1mF CGO/NPOs in 0603 size after all :D
_____________________

Wayne, I seem to remember, your final Servo used 2 x OP07 instead of the single you show on this thread.
viewtopic.php?f=6&t=256

Where did you put the final circuit?
I mainly used the THAT parts because it was easy to cut and paste a few caps.

Did you get to read the AES paper Gary Hebert just presented, before AES made them take it down?

I am just showing another way to skin the click cat without a servo.

JR

User avatar
mediatechnology
Posts: 3276
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mic Preamp DC-Servo: The "backwards" way

Post by mediatechnology » Sat Oct 25, 2014 7:26 am

The 1570/1583 has its own anti-click strategies which allow direct connection between the i/p emitters to Rg
But only if you use an external servo or the companion THAT5170/5173 gain controller with its internal servo.

What JR drew, in the context of a standalone 1570 or 1583 makes sense.
It would seem to work with any IC or discrete implementation that gives you the differential outputs.
(The 1510/1512/INA217/SSM2019 do not have external differential outputs so the AC and DC feedback loops can't be separated this way.)

I'm still pondering a combined bootstrap/offset servo using Rbias for injection...

User avatar
mediatechnology
Posts: 3276
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mic Preamp DC-Servo: The "backwards" way

Post by mediatechnology » Sat Oct 25, 2014 12:04 pm

I have a couple of ideas for a combined bootstrap/servo using the bias resistors for injection for you to consider Olaf.

One of the "problems" with using the Differential Deboo, aka DiffBoo, in a combined servo/bootstrap is that it will pass CM signals through with a gain of "1" at DC in addition to integrating the offset.
As Olaf has pointed out that CM gain can't extend to DC and also be used as bootstrap.

I see two potential ways to eliminate the DC CM component from the servo path:

Method A:

Instead of having the offset servo look at the gain stage op amp outputs, move the servo sample point one stage right after the preamp's common mode rejection stage.
That eliminates the AC/DC Vcm signal from the servo inputs and outputs.
Sample the common mode voltage at the gain stage op amp outputs (e.g. 1570 pins 2&3) and HP filter it to use as an AC Vcm signal.
Combine the +/-Vservo errors and Vcm bootstrap and feed them to the bias resistors.
I think the DiffBoo can be modified to re-inject AC Vcm.

Method B:

Sample the preamp DC error at the gain stage op amp outputs (e.g. 1570 pins 2&3) and allow Vcm to enter the servo +/- inputs.
Sample the common mode voltage at the same gain stage outputs and low pass filter the Vcm signal.
Subtract the low pass-filtered Vcm (which includes DC) from the servo outputs to derive a high-pass Vcm response to be used as AC bootstrap.
If the DiffBoo can be modified to re-inject low pass-filtered Vcm, then the sign of Vcm injection needs to be "-1" to provide a derived high pass response.

olafmatt
Posts: 32
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

Re: Mic Preamp DC-Servo: The "backwards" way

Post by olafmatt » Sat Oct 25, 2014 12:33 pm

Hhm, this thounds all rather complicated. Especially considering the precision neded so that no imbalance (due to not so perfectly matched resistors) is added which would mess up the effect of the CM bootstrap.

Since both alternatives have in common that they do not sample the DC error across Rg, what about this one:

Method C:

Sample the DC error at the diff-amp's output (i.e. CM signals already rejected and assuming differential output). Use a Diff-Boo as servo and control the bias voltages at the non-inverting inputs of the opamps if the instrumentation amp.
If no differential output is available, use some other servo circuit and invert the servo's output to get differential correction voltage.

This would mean no need to inject anything into the bases. This would also make marketing guys happy, because the hypesheet could say "DC servo does not act on your precious audio signals directly". Of course this is technically as incorrect as saying the DC servo is not part of the signal path.

User avatar
mediatechnology
Posts: 3276
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mic Preamp DC-Servo: The "backwards" way

Post by mediatechnology » Sat Oct 25, 2014 1:07 pm

A and B really aren't complicated but if you're not interested I won't defend them.
I'm not crazy about CM bootstrapping anyway.
They are simply suggestions.

Neither A or B are that significantly different than C when it comes to offset correction.
The only difference is using the bases or the collectors and option C doesn't appear to bootstrap the bias resistors.

In IC-based preamps the collectors are not available.
I don't know (or remember from your OP) if your preamp is discrete or not.

Option C however does not bootstrap your input resistors.
A and B might if you want bootstrapping but if you don't then they are complicated.
Since both alternatives have in common that they do not sample the DC error across Rg
Been there, done that, have fun with it.
This would also make marketing guys happy
Whose marketing guys?

olafmatt
Posts: 32
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

Re: Mic Preamp DC-Servo: The "backwards" way

Post by olafmatt » Sat Oct 25, 2014 1:24 pm

mediatechnology wrote:Option C however does not bootstrap your input resistors.
No, obviously not, because it just explains how to servo the circuit without having to touch the bases. So in order to have a bootstrap for Rbias, you'd still need a cap and two matched resistors for that. And yes, I am working with a discrete circuit.
mediatechnology wrote:Been there, done that, have fun with it.
Yeah, in the meantime even I realized that I was asking the wrong question initially. And I also realize that I'm at least 20 years to late becasue I'm the last dinosaur using a discrete circuit instead of a chip.

User avatar
mediatechnology
Posts: 3276
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mic Preamp DC-Servo: The "backwards" way

Post by mediatechnology » Sat Oct 25, 2014 2:19 pm

So in order to have a bootstrap for Rbias, you'd still need a cap and two matched resistors for that. And yes, I am working with a discrete circuit.
Well, it occurred to me that if one were to use a THAT1206 InGenius as the common mode rejector stage the common mode bootstrap output (OA4 out, pin 5) normally used to bootstrap its own resistors could be used to provide Vcm for the input bias resistors.
Rather than return the 1206's Cbootstrap to it's internal bias resistors, the bootstrap output could be extended all the way back to the preamp's input Rbias.
That's not a complicated thing if you use a 1200/1206.
And I also realize that I'm at least 20 years to late becasue I'm the last dinosaur using a discrete circuit instead of a chip.
I'm not so sure of that.

WRT to the servo...
What I've found using the servo taken from the output and fed back into the input is that the slowest servo recovery is conveniently at the lowest gain settings.
That's because at high gain the cutoff is higher - I chose 7 Hz at 60dB.
I've done a lot of switching at deadly monitoring levels without signal and it just works well.
You have to violently rotate the gain switch (or ramp the THAT5173) to get a pip.
With signal present it's not audible - the change in gain and signal masks everything.

Taking the error from the emitters just requires too much DC precision - the required op amp is more expensive than the preamp - and you risk having the servo imbalance the AC CMR of the preamp.
I also found emitter sensing more sensitive due to errors from HF and VHF rectification.

I'm not sure it really matters too much where you introduce correction: Base or collector.
As long as it produces the desired result, doesn't affect balance and introduce noise.

olafmatt
Posts: 32
Joined: Wed Oct 22, 2014 11:33 am
Location: Europe

Re: Mic Preamp DC-Servo: The "backwards" way

Post by olafmatt » Sat Oct 25, 2014 2:39 pm

mediatechnology wrote:That's because at high gain the cutoff is higher - I chose 7 Hz at 60dB.
The cuttoff frequency of my caps and Rbias is about 0.7 Hz (since I need to record signals down to 8Hz*). So the servo cutoff can only be below that at max. gain. And with a gain range of 60dB this results in the servo being so incredibly slow at min. gain that it takes ages for the DC offset to go away.
Maybe I should use a gain switch with two decks and change the servo's frequency after every 20dB of gain change, or so. Or distribute the gain across the instrumentation amp and the diff amp. I.e. have a "coarse" gain setting on the diff amp (+0dB, +20dB, +40dB), so that the gain range on the instrumentation amp is not that large (and thus the cutoff doesn't change that much). But then the diff amp needs another servo since the remaining (error) offset of the first one would get amplified 100x when at the +40dB setting. But that's easier to implement with a servo frequency that gets switched along with the gain. A drawback of having a diff amp with more gain is that CMRR of the complete circuit gets lower by the same amount.

*(I'm totally aware that I've now started another discussion on the topic why on earth I need to be able to record 8Hz since this will make granny's speakers jump from the shelve.)

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

Re: Mic Preamp DC-Servo: The "backwards" way

Post by ricardo » Sat Oct 25, 2014 3:21 pm

Wayne, I seem to remember, your final Servo used 2 x OP07 instead of the single you show on this thread.
viewtopic.php?f=6&t=256

Where did you put the final circuit?

User avatar
mediatechnology
Posts: 3276
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: Mic Preamp DC-Servo: The "backwards" way

Post by mediatechnology » Sun Oct 26, 2014 6:53 am

Wayne, I seem to remember, your final Servo used 2 x OP07 instead of the single you show on this thread.
viewtopic.php?f=6&t=256
Are referring to the Deboo?
I think that for single-ended servo inputs and outputs I settled on one OP07.

Could you be thinking about this one?
This single-input Deboo using two op amps has balanced outputs to drive preamps using T-bias:
viewtopic.php?f=6&t=14&start=114
(T-bias requires a "push-pull" servo.)

Or this one?
The Differential Deboo, for differential output preamps, uses two OP07's and looks similar to an instrumentation amp.
viewtopic.php?f=6&t=559&p=6320

Post Reply