Page 32 of 50

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

Posted: Thu Feb 21, 2019 9:26 pm
by mediatechnology
Have you actually built this or just sim'd it?

I never seriously considered injecting servo into the collector load.
There's another circuit out there that I've seen that servos Vc to bring the inputs into balance.

A lot of ways to do that other than sending currents into the bases.
I've also servo'd emitter current in a design that used current sinks. (Ie not provided by the op amps like this one.)

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

Posted: Thu Feb 21, 2019 11:59 pm
by terkio
I calculated it. Simply using basic EE laws and a transistor doing Ic=Beta x Ib, Vbe=0.7V. I think this very simple transistor is valid because they work at Vcb = constant, and Ic = fairly constant
I will sim it to make sure my calculations make sense. I have LTSpice but on another machine, I do not have with me presently.

I do not see many ways to inject correction into the Cohen design.
_At the transistor bases as done with a trim in Samuel Groner's Monte Generoso
_At the transistor collectors, directly or indirectly as done in your last version.
_At the transistor emitters, I did not look in this way yet.....looks promissing
_At the THAT1240 as done in your first version.

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

Posted: Fri Feb 22, 2019 6:00 am
by mediatechnology
terkio wrote:
Thu Feb 21, 2019 11:59 pm
_At the THAT1240 as done in your first version.
The last one doesn't quite describe it correctly since the injection is done one stage ahead of the 1240s at the bias node.
The basic DC model is right but there are at least one post in here (not yours) were sims were done that led to wrong conclusions.
I've done three of the 4 you listed using real parts not sims.

The method which uses the op amps to provide Ie, similar to Cohen, doesn't lend itself to being used with a gain controller.
The gain controller IC, for reasons of distortion and clicking, doesn't want Ie in the Ra and Rb gain legs.
For that topology I used differential and common mode servo'ing of Ie and it works.

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

Posted: Wed Feb 27, 2019 9:54 am
by terkio
I think the various ways to inject servo correction do not make much difference in the case of a fixed gain preamp, it's only a matter of convenience.
In the case of variable gain, that is an all different story and this is an excellent paper showing the issues. http://www.thatcorp.com/datashts/AES919 ... ifiers.pdf

I looked at injection at the emitters. It works great to achieve servo action independant of gain. This is why: A key point of the "Cohen topology" is that Ic is constant thanks to the op amps that enforce this to be. Vcb is constant too. The results are that Ie is constant, Ve is constant. So when injecting a current at the emitter, it can only flow into Rf the feedback resistance, regardless of the preamp gain. Furthermore, current injection is accurately done, simply with a resistor from the servo since Ve is at a constant voltage and near zero.
I checked with LTSpice, it does work as said. The current injection all flows into Rf with near no other effect in the preamp. The servo action does not depend of the preamp gain.

Wether one wants servo action independant of gain versus dependant of gain is another story. We have means to cleanly achieve one or the other.

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

Posted: Wed Feb 27, 2019 10:34 am
by mediatechnology
terkio wrote:
Wed Feb 27, 2019 9:54 am
I think the various ways to inject servo correction do not make much difference in the case of a fixed gain preamp, it's only a matter of convenience.
In the case of variable gain, that is an all different story and this is an excellent paper showing the issues. http://www.thatcorp.com/datashts/AES919 ... ifiers.pdf

I looked at injection at the emitters. It works great to achieve servo action independant of gain. This is why: A key point of the "Cohen topology" is that Ic is constant thanks to the op amps that enforce this to be. Vcb is constant too. The results are that Ie is constant, Ve is constant. So when injecting a current at the emitter, it can only flow into Rf the feedback resistance, regardless of the preamp gain. Furthermore, current injection is accurately done, simply with a resistor from the servo since Ve is at a constant voltage and near zero.
I checked with LTSpice, it does work as said. The current injection all flows into Rf with near no other effect in the preamp. The servo action does not depend of the preamp gain.

Wether one wants servo action independant of gain versus dependant of gain is another story. We have means to cleanly achieve one or the other.
I agree on all points.

Having fc fixed in a fixed gain preamp really doesn't help much since once you know gain you can move fc if you want to keep it constant.
If gain can be varied by 20 dB, and you're using servo fc as a first-order derived high-pass, then it might matter since the HP fc would move by a 10:1 ratio.
The derived first-order high-pass might be useful for simple warp reduction for instance.

This is a point that I've been wanting to make for a long time and it is a point the THAT paper makes in a different way:

"A servo, which is a low-pass filter, produces a derived, subtracted, high-pass response."

This is why second-order servos peak and why inadvertent second-order servos illustrated in the THAT paper also peak.

An "analog" is the derived response of a second-order, uncorrected, elliptic equalizer. viewtopic.php?f=6&t=828

I've written an xls calculator based on Gary's half-circuit equations and have used them here and elsewhere.
Not sure I ever posted it.

Injection at the emitters is something I've tried and used with success.
For some applications, such as digital gain control, current flow back into the feedback resistors may be problematic.
What I did was split the emitter current sinks in this circuit.
I then applied a common mode correction signal and a differential one.
See: http://www.thatcorp.com/datashts/dn109.pdf

Image

The bases of Q1A and Q1B can be considered summing nodes.
Splitting the base connections and summing a shared common mode correction into both bases along with differential correction allows both transistors to control input stage Ie. (4 resistors 2X/transistor base.)

The advantage of controlling Ie or Ic (Vc) over base correction is that low impedance at DC sources don't differentially shunt the base currents used for correction. (Think ribbon mic or MC phono pickup directly to bases.)

In the MC preamp correction of Vc or Ie are both viable options.
Since the op amps are supplying Ie we're not concerned about current in Rfa or Rfb.

Thanks for doing the sims.
I've "simulated" all three in 3D on the bench so its good to know sims do relate in some way to reality.

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

Posted: Wed Feb 27, 2019 2:01 pm
by terkio
I understand the servo, sensing the common mode at the op amp outputs, is to center these at zero, to achieve a best headroom.
However I do not undestand the need for a servo, because I see no need for an accurate centering at zero.
I woul simply use two resitors ( instead of two current sources and an integrator ). The 1/2 4570 would better be used to provide XLR balanced outputs.
May be I am missing something. I wonder about noise from resistors versus current sources.

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

Posted: Wed Feb 27, 2019 4:59 pm
by mediatechnology
terkio wrote:
Wed Feb 27, 2019 2:01 pm
I understand the servo, sensing the common mode at the op amp outputs, is to center these at zero, to achieve a best headroom.
However I do not undestand the need for a servo, because I see no need for an accurate centering at zero.
I woul simply use two resitors ( instead of two current sources and an integrator ). The 1/2 4570 would better be used to provide XLR balanced outputs.
May be I am missing something. I wonder about noise from resistors versus current sources.
Its not primarily about headroom.
When you use a switched resistor digital gain controller there cannot be DC current in the resistors. (THD and clicking.)
Current in the resistors can be setup in common mode or differential.

They likely used a common mode servo in DN-109 to use the second half of the op amp.
The circuit I did used a gain controller so a CM servo (and a differential one) was a simple way to keep DC out of the IC's FET switches.

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

Posted: Wed Feb 27, 2019 5:27 pm
by terkio
Thanks to enlight me about switched resistor gain controllers. So far I only know, reading about the THAT5173.
I am very interested.
What about IC's with resistor networks and FET switches ? As an alternative, for example, to a 24 position rotary gain switch doing 2.5 dB steps in a mic preamp. Available ICs and Application notes ?

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

Posted: Thu Feb 28, 2019 4:15 am
by terkio
mediatechnology wrote:
Wed Feb 27, 2019 4:59 pm
When you use a switched resistor digital gain controller there cannot be DC current in the resistors. (THD and clicking.)
Thanks, this is an important point I had overlooked.

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

Posted: Thu Feb 28, 2019 6:20 am
by mediatechnology
THAT have told me that the internal THD reduction used in the 517X-series can't cope with DC.
Discrete FETS or transmission gates might not need that consideration.

For this topology in a mic preamp using mechanical switching Cohen got it right by switching only Rg.

THAT's switching of Rfa and Rfb in some models allows them to hit the optimum NF at various gains but the primary reason (I think) is to keep Rg from getting too low to integrate.

Summing up my thoughts:

In the MC preamp (and "Cohen") having the op amps supply Ie has a lot of advantages in terms of fewer noise sources and simplicity.
But it limits the range of Rfa and Rfb to relatively low values and reduces output headroom somewhat.
In this application simplest and lowest noise is best so I went with using the op amps as current sinks for the emitters.

The MC preamp and a mic preamp use similar front ends but in many ways they're different.
What works here may not be best for a mic preamp.
But the ZTX851 at the heart of it may be the best we can do with obtainable transistors.