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. » Wed Oct 22, 2014 4:49 pm

olafmatt wrote:
mediatechnology wrote:So I understand correctly where you're headed with this what is the value range of C1 and C2 you propose?
2u2 film caps.
JR. wrote:I don't see a servo in your schematic forwards or backwards.
Don't worry, your eyes are doing just fine, it's my schematic which is not showing a servo. I didn't draw the servo because there are so many ways to skin the cat. The idea was to have the servo monitor offset across Rg and then adjust the voltages at the opamps non-inverting inputs.

The reason I didn't draw the servo and the way the voltages get controlled is that there are so many ways to do that and I wanted to avoid discussion about the not-so-important parts. Sorry, I guess I'm traumatized by other forms where people would have complaint that my opamps are not connected to the power supply when being presented the schematic linked in above... :oops:

Olaf
Several of us here have a great deal of experience with that mic preamp topology.

The largest problem with putting your servo across Rg is that you will still have the full offset voltage of your servo op amp across Rg. At gains of 1000x, not uncommon for a mic preamp you will experience 1000x that DC offset at the gain stage output.

JR

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

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

Post by olafmatt » Wed Oct 22, 2014 4:58 pm

mediatechnology wrote:
It seems like combining the functions of both servo and bootstrap using a combination of differential and common mode drive feeding bias resistors is a possibility.
So would that work?
Maybe... but isn't that almost what my circuit a few posts above does? Except that I have separate resistors for CM and DM signals.

The idea was that a diff. Deboo servo will have a CM gain of 1x. So the output "control" voltages of the servo will move up and down with CM signals. That way R11 and R12 see a constant voltage across them when looking at CM signals. The servo's differential output gets added to the input through a differential voltage divider (i.e. no ground reference, so not hurting Rcm, hopefully).
mediatechnology wrote:Do you have an example of that? It may help me to visualize it.
It's done in the Green Pre (which I think is a copy of the Amek Mozart). The first schematic I could find is here: http://www.mhumhirecords.org/DIYpics/Gr ... e_cct4.gif
JR. wrote:The largest problem with putting your servo across Rg is that you will still have the full offset voltage of your servo op amp across Rg. At gains of 1000x, not uncommon for a mic preamp you will experience 1000x that DC offset at the gain stage output.
I'm totally aware of that and would probably add another servo to the output stage / diff. amp.

Actually, this is what got me started on looking for other servo injection points. Boths ways of doing it (either monitoring Rg or the output) have severe drawbacks.
I was hoping to find some point in the circuit, where the servo correction voltage could be injected without gain being applied to it. There seems to be none. But at least I had the idea with altering the DC voltage at the non-inverting opamp inputs which at least solves the problem of the servo injection hurting the common mode impedance of the input (between caps and transistors) because that offset can be set "from the back".

Olaf

User avatar
JR.
Posts: 2418
Joined: Sat May 24, 2008 7:21 pm

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

Post by JR. » Wed Oct 22, 2014 7:18 pm

olafmatt wrote:
JR. wrote:The largest problem with putting your servo across Rg is that you will still have the full offset voltage of your servo op amp across Rg. At gains of 1000x, not uncommon for a mic preamp you will experience 1000x that DC offset at the gain stage output.
I'm totally aware of that and would probably add another servo to the output stage / diff. amp.

Actually, this is what got me started on looking for other servo injection points. Boths ways of doing it (either monitoring Rg or the output) have severe drawbacks.
I was hoping to find some point in the circuit, where the servo correction voltage could be injected without gain being applied to it. There seems to be none. But at least I had the idea with altering the DC voltage at the non-inverting opamp inputs which at least solves the problem of the servo injection hurting the common mode impedance of the input (between caps and transistors) because that offset can be set "from the back".

Olaf
Adding a servo post the mic preamp can correct for DC but it will not mitigate scratchy gain pot, or clicky gain switches.

JR

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

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

Post by ricardo » Wed Oct 22, 2014 7:54 pm

olafmatt wrote:Actually, this is what got me started on looking for other servo injection points. Boths ways of doing it (either monitoring Rg or the output) have severe drawbacks.
I was hoping to find some point in the circuit, where the servo correction voltage could be injected without gain being applied to it. There seems to be none. But at least I had the idea with altering the DC voltage at the non-inverting opamp inputs which at least solves the problem of the servo injection hurting the common mode impedance of the input (between caps and transistors) because that offset can be set "from the back".
I think we need to look at what evils we are really interested in avoiding.

0.5V DC on the output of a mike preamp is no big deal ... provided it's consistent and repeatable. You can bias a nice electrolytic or even a Golden Pinnae film.

But a cap in series with Rg has to be huge. The physical size makes them very good aerials for RFI and good layout for LN becomes really difficult. Even a huge electrolytic is likely to introduce some THD cos it will be affecting response at 20Hz.

But what we are really trying to avoid is clicks when switching gain or scratchy noises if you are using a pot for Rg. The right solution is 1mF 25V NPO/CGO caps in 0603 size. :D

Wayne has an excellent thread on the use of servos to replace these mythical beasts (out of stock at Mouser & Digikey) with servos. He explicitly checks the very thing we want to avoid .. clicks & pops on gain switching.
____________

I have to disagree with JR on the EVILs of getting rid of clicks & pops by putting yucky caps in the signal path. The broadcast people often want to change gain etc with mikes live. I'm happy to give them imperceptible switching at the cost of the EVIL distortion from my caps. :o

User avatar
JR.
Posts: 2418
Joined: Sat May 24, 2008 7:21 pm

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

Post by JR. » Wed Oct 22, 2014 8:13 pm

ricardo wrote:

I have to disagree with JR on the EVILs of getting rid of clicks & pops by putting yucky caps in the signal path. The broadcast people often want to change gain etc with mikes live. I'm happy to give them imperceptible switching at the cost of the EVIL distortion from my caps. :o
Argue with something I actually believe... The customers punish clicks and pops by not buying the products.

Yes I do not appreciate caps in series with Rg, but if you put two caps between the feedback network and gain stage - inputs they don't need to be as large as in series with Rg . You will need another pair of caps in series with the gain stage outputs and a separate feedback resistor for DC feedback. Since the entire feedback net is no longer in the DC path, no more scratch or clicks.

Note: Even though I do not appreciate caps in series with Rg I have done it in many value designs because the customers dislike clicks and pops more.

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 » Wed Oct 22, 2014 8:44 pm

I agree with ricardo and JR that offset across Rg and the resulting clicks from the DC not being 0V is a bigger problem than offset at the output.
What I've found with the 1510/1512 and 1570-series however is that sampling the output to indirectly measure the DC error at Rg and correcting it at the input works pretty well despite the cutoff frequency varying with gain.
It's low on clicks and I set the worst-case cutoff at the maximum gain.
The slower settling time at low gain doesn't seem to be an issue.

If the goal is to use film input caps and they're 2.2uF/leg (or 1.1uF) then I would work backwards from that C value and the desired differential cutoff frequency to establish the values of R1 and R2 so that the cutoff established by Cin/2 and R1+R2 is where you want it to be.

If you're bootstrapping these bias resistors, then their value in terms of LF CMRR and capacitor mismatch isn't that critical since their effective value is multiplied by several orders of magnitude due to bootstrapping.

I suspect that with a bipolar input topology the high-ish Rbias values will produce a lot of Ios X Rbias error to deal with in addition to the Vbe mismatch and other errors.

The bias current requirements of the servo amplifier and the resulting Vos when its sampled from the emitters produce large errors: The Vos and IbiasXRin errors are as big as the error being measured.
Using the gain of the preamp to magnify the DC error and sampling the output just seems to work better. That's been my experience.

The circuit you drew show both differential servo correction and CM bootstrapping.
I think you can however combine both functions into the servo and have it produce both differential DC correction and common mode bias resistor bootstrapping.
Both components appear to be injectable into the bias resistors.

I remember now why I sampled Vcm from the THAT1512 emitters: It's the last place on the IC that you have access to Vcm.
The op amp outputs for the two preamp half-circuits are not accessible.
The final output of the 1510/1512 is after the differential to single-ended stage and the common mode component has been rejected.
Only the inputs and the emitter connections contain the Vcm signal.

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

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

Post by olafmatt » Thu Oct 23, 2014 9:35 am

Ok, please ignore the question where to sample the DC offset for a while...
mediatechnology wrote:The circuit you drew show both differential servo correction and CM bootstrapping.
I think you can however combine both functions into the servo and have it produce both differential DC correction and common mode bias resistor bootstrapping.
Both components appear to be injectable into the bias resistors.
I get your point, but I'm not sure whether it would work.

The CM bootstrap is not a CM DC level shifting thing. Due to C3 the effect of the bootstrap drops at some low frequency, so that at DC we're back at "normal" operating conditions (i.e. same as with no bootstrap being there).
Trying to combine the CM bootstrap signal with the servo's correction voltages would require a cap in series with the outputs of the servo. But then the servo would stop working at very low frequencies.

OTOH, I think my proposed circuit woudn't work either. Due to the CM gain of 1 of the servo (which extends down to DC), for CM signals it would look like large value resistors (2M2) being connected from the emitters to the bases of the opposite transistor.

Olaf

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 » Thu Oct 23, 2014 10:44 am

The CM bootstrap is not a CM DC level shifting thing. Due to C3 the effect of the bootstrap drops at some low frequency, so that at DC we're back at "normal" operating conditions (i.e. same as with no bootstrap being there).
Trying to combine the CM bootstrap signal with the servo's correction voltages would require a cap in series with the outputs of the servo. But then the servo would stop working at very low frequencies.
My thought was that the input to the servo that produces common mode drive would be AC-coupled and the servo input and outputs DC-coupled.
The servo would have three inputs: Servo +, Servo -, and Vcm. The Vcm input would be AC-coupled to serve as bootstrap. The + and - inputs are the integrator terminals.
The servo would have two outputs: Servo + and Servo -.

The Servo + and Servo - outputs would apply DC servo correction differentially on their outputs.
The Servo + and Servo - outputs would also carry a common mode AC bootstrap signal.

The THAT1606 or 1646 have a similar input and output for what's needed.

An example of the 1646 outputs providing a DC servo signal differentially with an AC common mode signal is the Dual Class-A amplifier.
The differential servo is the +/- Vbe voltage with thermal feedback; the common mode signal with the differential "error" riding on it are the boost transistors' base drive.
Unfortunately a 1646 wouldn't work so well as a servo since its not DC-precise and can't be turned into an integrator. But its a similar concept.

It seems at first glance that the Differential Deboo's common mode gain of +1 makes it serve both functions intrinsically except for the fact that the common mode gain isn't AC-coupled. That's a problem.
Maybe I should think about how to do that.
My choice of the Differential DeBoo in the input-capacitorless preamp was that I wanted a single floating capacitor.
The "Diff-Boo" may not be the best choice for a combo servo/bootstrap circuit.

But I think there's a differential input/differential output integrator out there than can have an AC common mode signal impressed on the outputs.
By doing that the only resistors feeding anything into the input on the right-hand side of the coupling caps are the bias resistors.
The differential impedance will be the sum of the two and the common mode impedance will be as big as bootstrapping can make them.

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

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

Post by olafmatt » Thu Oct 23, 2014 11:16 am

mediatechnology wrote:But I think there's a differential input/differential output integrator out there than can have an AC common mode signal impressed on the outputs.
One could use a Diff-Boo followed by a fully differential opamp (like OPA1632), driving that opamp's Vcm pin with the AC common mode signal. Or maybe even turn that fully differential opamp into a differential intergator right away.
But the point is that the signals going into the servo also have CM (incl. DC) on them, which needs to be removed first, and then the desired AC CM bootstrap signal added.

Olaf

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

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

Post by ricardo » Thu Oct 23, 2014 7:05 pm

JR. wrote:Yes I do not appreciate caps in series with Rg, but if you put two caps between the feedback network and gain stage - inputs they don't need to be as large as in series with Rg . You will need another pair of caps in series with the gain stage outputs and a separate feedback resistor for DC feedback. Since the entire feedback net is no longer in the DC path, no more scratch or clicks.
I've played with such circuits in Jurassic times. But the extra caps in the feedback network make the circuit a higher order LF filter and its difficult to avoid serious peaking on some gain settings.

If you've successfully used this, please post a sketch or some clue on how you did it.

I've used this trick to get higher order rumble filters into RIAA preamps.
Argue with something I actually believe... The customers punish clicks and pops by not buying the products.

Note: Even though I do not appreciate caps in series with Rg I have done it in many value designs because the customers dislike clicks and pops more.
I too am guilty of this EVIL and they weren't even 'value designs'. Mea culpa, mea culpa, mea maxima culpa. :o

It's really just my 'Broadcast' snobbery vis a vis da 'Music' guys. The unwashed masses rarely switch gain when tracking or mastering with the 'tape' running .. while the Broadcast guys often have to.

In da 60's & 70's, if the Studio Manager (BBC-speak for sound engineer) moved a fader during a Proms broadcast, there would be letters to the Times (OK HiFi News) from Colonel Auric Pinnae & Co.

"Sir, I distinctly heard you touch a fader just before the drum-roll on last night's Haydn Toy Symphony broadcast. And why are you using those evil new fangled transistorised AKG414s instead of C12s?"
______________

Olaf, you should really read Wayne's huge thread on simple servo on THAT 1510/12. It covers practically all the important points and has many 'real life' measurements.

I think its viewtopic.php?f=6&t=256 though there are other threads which expand on certain points.
______________

It appears THAT have incorporated many of these points in the system around THAT5171

Post Reply