A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

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: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by mediatechnology »

In retrospect it might be possible to eliminate two resistors (R15 and R16) since the servo will be pinned in common mode to ground by the 1240 outputs.
I'll try that on the bench.
I seem to recall trying that but may have not.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by mediatechnology »

OK, just checked it and R15 and R16 can be eliminated.
R17 and R18 become 10KΩ.

The servo attenuation is now "2."

Good catch!
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by terkio »

I was not interested in T bias, not aware of it, actually. The preamps I know have two 10K resitors to bias the two transistor bases to ground, without a common resistor. I know this gives a poor common mode rejection, however, if my understanding is right, this is trouble only when the mic has poorly balanced impedances. So I assume good quality mics will be used. If not, junk it or if you badly want to use it, insert a transformer. I know, too I need tight matching of the two caps required for phantom power, those are 33uF, are quite expensive and don't come with tight tolerance.

However I am getting interested in the T biais, and I was wondering about the boostraping of the common mode impedance technique, is the boostraping compatible with a servo to balance the preamplifier ? Your link to a T bias example comes handy.
Indeed the T bias opens a door, but brings more complexity.
Last edited by terkio on Tue Feb 19, 2019 3:56 pm, edited 1 time in total.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by mediatechnology »

T-bias' benefit is the low frequency common mode rejection improvement.
It is a capacitor value mismatch improver.
I've never tried bootstrapping the CM in a mic preamp similar to InGenius.

About the only complexity T-bias adds in a servo situation, where servo is injected into the inputs, is the requirement for a differential servo.

I went back to my notes and remembered where the extra two bridge resistors came from. (R15 and R16).
During my initial experiments I had the servo sense the INA output prior to the THAT1240 stages.
There was nothing to pin the floating Deboo servo to ground so the added two resistors were required.
Once the servo sense was moved to the final output, and the servo input became pinned to ground by the 1240s they became redundant.

If you visually take apart the differential Deboo you see an instrumentation amp at the core.
The common mode gain of the differential Deboo is "1."

I use the floating and common mode gain of 1 properties of the differential Deboo here: viewtopic.php?f=6&t=598

viewtopic.php?f=6&t=559
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by mediatechnology »

I went back to my notes and remembered where the extra two bridge resistors came from. (R15 and R16).
During my initial experiments I had the servo sense the INA output prior to the THAT1240 stages.
There was nothing to pin the floating Deboo servo to ground so the added two resistors were required.
Once the servo sense was moved to the final output, and the servo input became pinned to ground by the 1240s they became redundant.
I'm re-thinking this.
Removing them may not be such a good idea from a troubleshooting perspective.
With R15 and R16 in circuit the preamp will run, for troubleshooting, with the OPA2277 servo removed from its socket.
If, for instance, there was a DC problem someone could pop the servo op amp out of its socket to isolate which stage was bad.
Though Ic wouldn't be correct the front end would bias up enough to make DC measurements.

With R15 and R16 in place, the front end will run with both THAT1240s and the OPA2277 pulled.
It may have offset, but it will run.
That allows the front end to be tested in isolation.
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by terkio »

1) Indeed the differential Deboo is a very interesting circuit. I saw a Deboo integrator needs resistor matching. Some R1*R2/R3 that makes a negative resistor that must equal a fourth resistor R4. Does that need accurate resistor matching ? What about stability ? Negative resistors are a way to make oscillators !

2) About correction not subject to gain, by injecting at an appropriate place, is there a patent, a paper that explains it ?
How come Thatcorp does not use such a technique here: http://www.thatcorp.com/datashts/AES919 ... ifiers.pdf. From that paper, where they discuss a wacky micro controlled work around, I thought there was no simple way to avoid the servo fc shift with the gain.

3) Troubleshooting with servo popped out. There should be a way to desengage servo action, by means of jumpers.

4) Balancing could be done not only by the servo. A two levels balancing. First level, with a trimmer added at an appropriate place, servo desengaged. Second level, with the servo engaged and fine trimming to center the servo for maximum headroom.
Last edited by terkio on Tue Feb 19, 2019 5:33 pm, edited 1 time in total.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by mediatechnology »

Negative resistors are a way to make oscillators !
You'll note that the positive feedback is rolled off at high frequencies (above fc) by virtue of it being an integrator.
You can simulate the servo if you wish.
I recall Gary Hebert of THAT doing a sim for me.

I would use 1% resistors.
The 2C integrator requires matched capacitors to be monotonic to step changes.
I'd rather use 1% resistors and a single C.

I've only found one other reference to the Differential Deboo.
There may be more now.

For troubleshooting a board that someone has assembled incorrectly having them remove a bunch of ICs is a better option.
I live in the real world and have to deal with obvious and not-so-obvious mistakes on the part of builders.
Having them remove a jumper for an IC plugged in backwards or of the wrong type doesn't fully disconnect it from the board.

WRT servo-injecting at unity gain into the non-inverting inputs forum member Olaf suggested it.
It was his idea. I just built it.
I thought it to be rather brilliant.

There's simply no point in a trim.
You have to understand this circuit in the larger context of a mic preamp that is directly-coupled to phantom potentials.
The Vbe mismatch offset is de minimus compared to what the mic produces (or can).
The servo needs to have a large range for offsets that are dependent on what's connected to the input. We can't trim that in the lab.
The servo has to have range.
Or maybe someone doesn't want to go to the trouble to closely-match their transistors.

In the context of the MC preamp why put in a trim. What does it really gain you?
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by terkio »

A trim would take care of gross transistor mismatch. I agree, it would be nice to have no trimmer at all if this can be done without tight transistor matching. Here this depends of the user. Hobbyists who have time to match pairs from hundreds of transistors. Companies that has no time to do that. Repairmen who are not transistor pair matchers.

With a centered servo, I guess there will be less crackling noise when switching gains. Shall we have the same servo correction when switching the gain ? Or will the servo readjust the correction at every gain change ?
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by mediatechnology »

I haven't experimented using this servo with switched gains but I think it has advantages.

Olaf alerted me to this technique in a mic preamplifier thread.
Since I had a similar topology on my bench in the form of the MC preamp I decided to try it there.
It worked so well I decided to use it even though the MC preamp is essentially fixed-gain.

It's something I definitely intend to try in a variable gain mic preamp.

In a stepped gain mic preamp the fact that servo fc doesn't move allows more freedom in choosing what fc is.
A preamp with a gain range of 20-60 dB has a 100:1 change in fc. This allows us to work around that.
It can be made fast enough to respond to step changes quickly but won't have the crazy-long recovery times at low gain that a servo subject to gain does.
Whether it can be made optimum without switching fc I don't know.
Regardless, the fc needs to be chosen it doesn't form a peaking response with the input network if its set too high.

I've already shown that I can float an SPI-controlled gain controller so my interest is in mechanically-switched gain.
The "slew rate" from low to high gain is limited by how fast we can turn the knob so mechanically-switched gain doesn't place the same "catch-up" demands that a gain controller does.
We'll just have to see how it does.
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

Re: A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851

Post by terkio »

mediatechnology wrote: Tue Feb 19, 2019 7:03 pm I haven't experimented using this servo with switched gains but I think it has advantages.

Olaf alerted me to this technique in a mic preamplifier thread.
Since I had a similar topology on my bench in the form of the MC preamp I decided to try it there.
It worked so well I decided to use it even though the MC preamp is essentially fixed-gain.

It's something I definitely intend to try in a variable gain mic preamp.
Great.

I studied injecting the servo correction at the inverting inputs of the op-amp versus the non inverting inputs.
I came to the conclusion it makes no significant difference. I convinced myself this design choice does not make a difference about fc shifting with the gain.
Using all symetric circuits, no additional 11V rail, just using the +15V -15V rails, here is what I come up with.
In both cases I have a +1mA / -1mA at the transistors collector current from 3V servo output. Gain here is a transconductance 0.3333mA/V
In both cases, Ic is 6mA, 5mA, 4mA for servo drive 3V, 0V, -3V.
Case A: Injection at the non inverting inputs is an easier load on the servo that must drive like 15K to ground.
Collector resistor is 1K to the 15V rail.
Case B: Direct injection at the transistor collectors puts a heavier load on the servo that must drive like 1.8K to ground.
Collector resistor is 0.6K to the 15V rail.

This +1mA -1mA collector current -assuming infinite high beta transistors- will result in the same +1mA -1mA current throught the Rfeedback resistors. So with Rf =1K it gives +2V at the op-amp outputs.
Finally: When the outputs of the Deboo send +3V -3V to skew the mic preamp,
it results in +1V -1V at the op amp outputs,
which results in +2V -2V at the THAT1240 outputs

The good news is that this does not depends of the chosen preamp gain. Indeed fc doesn't shift with changing the mic preamp gain.
BUT there is a catch: This is only true if the transistors have extremely high beta. The bad news is that this is wrong in the case of not so high beta together with high values of transistor base resistors ( I plan on using two 10K where you have two 0.499K ) and this becomes terrible at high preamp gains like 60dB. The trouble comes from the voltage drop at the base resistors.
I am working on this, with some compensation scheme, not easy because Beta is not well defined and has tempco.
Well, in order to have fc not shifting with gain, I must solve this trouble or compromise with not so ambitious expectations.
Post Reply