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.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

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

Post by billshurv »

Gold star for you there Wayne. It's also interesting that almost no one discusses the distortion that can occur from underloading. I wonder what is going on there. I assume this is with the amp right next to the TT so only 20-30pF from the cabling?
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

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

Post by JR. »

I don't think I've ever seen a published MM preamp with less than 100pF across it's input... (but haven't looked).

I never seriously tried less, even with my switch able input termination kit that could go down to zip. I only offered 25 pf resolution, maybe we need to use trimmer caps. :lol:

JR

PS: even shorty RCA cables will probably overwhelm 10 pf of stray C.
Cancel the "cancel culture", do not support mob hatred.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

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

Post by billshurv »

I think the Barney Oliver design is the most well known with very little capacitance added (30pF).
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 that experiment there was about 3 feet of 2C shielded cable so the cart had about 100 pF of cable load.

My hunch was the distortion was underdamped mechanical/electrical resonance in the 2-10 kHz region where we also see the impedance hump. Maybe around 6 kHz.

At the moment, with my Stanton 681, I can't recall if I'm using 100 pF or 200 pF Cterm.
billshurv
Posts: 436
Joined: Thu Feb 01, 2018 6:07 am

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

Post by billshurv »

It can't be mechanical damping, and in theory low pF would give an overdamped response so this is something I haven't seen in the literature before. I'll try and replicate that when I get a chance.
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'm a few steps closer to getting back on the ZTX851 MC preamp.
Finishing a few schematics for other projects as we speak.
The ZTX851 Protoboard is actually sitting in the center of the bench.
My next move is to rip the whole thing up and rebuild it more cleanly and true to the current sch.
User avatar
terkio
Posts: 304
Joined: Thu Feb 07, 2019 3:17 am
Contact:

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

Post by terkio »

mediatechnology wrote: Mon Apr 30, 2018 5:14 pm This is a draft schematic for the latest Flat Balanced Input Moving Coil Phono Preamp using the ZTX851.

Image
Flat Balanced Input Moving Coil Phono Preamp using the ZTX851.

Printable copy: httsp://www.proaudiodesignforum.com/images/png/ ... ZTX851.png

The circuit above is virtually identical to the original in terms of noise performance.
The servo has been updated to a Differential Deboo and a TL431 shunt regulator added to improve PSRR.
The design is completely DC-coupled.
A high pass filter response is derived by a servo.
C7, a 0.47 uF film capacitor, sets the single-order High pass response 3 dB point at about 10 Hz.

The circuit is based in the "Cohen" philosophy which keeps the impedances low. http://www.proaudiodesignforum.com/foru ... 161&p=1343
R8 and R9 sink emitter current to eliminate additional noise sources.

R13, R14, R15 and R16 form a bridge circuit to set the collector current at a nominal 5.5 mA each.
R17 and R18 along with the Deboo integrator rebalance the bridge to correct offset error by skewing the collector currents.
The servo forces the emitter potentials to be equal reducing the delta-Vbe x gain error to millivolts at the final output.

I have to give credit to Demrow and Cohen for the topology and low-noise philosophy.
Horowitz and Hill, along with Brad Wood (Bcarso RIP) alerted me to the low rbe ZTX851.
"Olafmatt," here on our forum, handed me the servo on a silver platter for which I'm very grateful.
Thanks also to Hans for the sims and JR for the input.
Back to the circuit on April 30, 2018.
1_ R15 and R16 could be omitted with R17=R18=10K ( instead of 20K )
2_Isn't it easier to inject the servo correcting signals at the op-amps inverting inputs, instead of the op-amps non inverting inputs, then there would be no need to split the two non inverting inputs.
3_Before adding a +11V rail made from a TL431, did you try large caps at the op-amps non inverting inputs. With such caps enforcing DC across R13 and R14, all op-amps inputs would be referrenced to the +15v rail, all op-amps inputs would carry about the same ripple as the +15V, so rejecting this ripple.
4_The preamp is a balanced symetric circuit, however I think there is about no benefit to use a symetric correcting circuit. Injecting a correcting signal at one leg of the preamp circuit or at both legs of the preamp with half the signal amplitude and opposite sign does about the same. There is no need for a servo that monitors both out+ and out-. Monitoring either one is enough because whenever this one goes at 0V the other one goes accuratly at 0V too, thanks to how the two THAT1240 make them. Then instead of a differential symetric Deboo integrator a simple integrator ( 1 op-amp, 1 resistor, 1 cap ) could do the job.
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 »

Back to the circuit on April 30, 2018.
1_ R15 and R16 could be omitted with R17=R18=10K ( instead of 20K )
2_Isn't it easier to inject the servo correcting signals at the op-amps inverting inputs, instead of the op-amps non inverting inputs, then there would be no need to split the two non inverting inputs.
3_Before adding a +11V rail made from a TL431, did you try large caps at the op-amps non inverting inputs. With such caps enforcing DC across R13 and R14, all op-amps inputs would be referrenced to the +15v rail, all op-amps inputs would carry about the same ripple as the +15V, so rejecting this ripple.
4_The preamp is a balanced symetric circuit, however I think there is about no benefit to use a symetric correcting circuit. Injecting a correcting signal at one leg of the preamp circuit or at both legs of the preamp with half the signal amplitude and opposite sign does about the same. There is no need for a servo that monitors both out+ and out-. Monitoring either one is enough because whenever this one goes at 0V the other one goes accuratly at 0V too, thanks to how the two THAT1240 make them. Then instead of a differential symetric Deboo integrator a simple integrator ( 1 op-amp, 1 resistor, 1 cap ) could do the job.
In order..

1) It has to do with setting the Q-point (for Ic) since the floating Deboo is "blind" to the common mode voltages at TP2 and TP3. The lower arms of the bridge (R15,R16) set the ground reference for the voltage dividers: The Deboo injects differential currents into the arms. Note that where the Deboo is placed the common mode voltage at TP2 and TP3 can be anything and the Deboo would never see it due to DC CM rejection by the THAT1240s. The Deboo was placed at the final output, after CM rejection, so that it would servo out all differential offsets including those of the THAT1240s. The servo has no idea what the actual CM voltage is ahead of the THAT1240s. If the Deboo were placed ahead of the 1240s their offsets wouldn't get corrected. R15 and R16 tell the bias circuit where ground is.

2) If you inject correction into the inverting input, that correction is subject to gain. The objective, not so much in this circuit but more so in a mic preamp of similar topology, is to NOT subject servo correction to gain because the servo fc shifts with the gain. By injecting servo correction into a node which is unity gain (the non-inverting input), a shifting fc is avoided. That node is a voltage follower. The fc you calculate is the fc you get regardless of gain. In a mic preamp it helps avoid a second-order resonating servo response with the input network because fc is controlled.

3) Ripple reduction was not the primary objective of the shunt regulator. I could achieve ripple-reduction (and did) with a simple cap-multiplier. A cap multiplier was too loosey-goosey. Re-read Cohen's paper - the part about the LEDs. I wrote:
Ultimately I decided that I wanted a better-defined voltage for the collector loads and reference dividers than a cap multiplier could provide.
I looked at simple solutions and decided on a TL431 shunt regulator and an 11V sub-rail for the collector loads and reference dividers.
Cohen's circuit had two red LEDs to lower Vcc to about 11.3V with a 15V rail.
The reason he gave - a good one - was to limit the common mode range the following 5532 would see.
Having an 11V sub-rail seemed like a good idea not only from a noise perspective but also mindful of the following op amp's CM range. https://proaudiodesignforum.com/forum/p ... 160#p11289
4) While there is no need to sense the differential output - only one leg will do - it does make sense to have a differential servo output because it provides twice the amount of available correction voltage to a unity gain correction node. Remember #2. If there was a single-ended servo, either conventional or Deboo - it's output would have to be inverted to obtain the same equivalent servo correction. Why not then use that second op amp as part of the Deboo? Op amps come in duals anyway. What would I do with the other half? In the input-capacitorless preamp the servo floats to the mic's phantom Q-point voltage. It can't be single-ended because the servo capacitor cannot connect to ground. The floating Differential Deboo is drawn from that circuit. (Note that the actual servo attenuation from the Deboo output to the non-inverting input is "4" due to the resistive bridge formed.) Servo DR is king. I need 10V of Deboo servo output to get 2.5V at the op amp non-inverting inputs.

EDIT: One other thing I would like to point out, since your interest is mic preamps, is that a differential servo output is a requirement if you are using "T-bias" with an AC-coupled input. T-bias reduces capacitor matching requirements by using a high value shared bias resistor to reference two low-value ones to ground. This raises CM impedance. A single ended servo, injected into one input only, will have a large amount of differential to common mode conversion. The servo loop will have low gain and be unstable.
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 »

2) "fc regardless of the gain" is a very interesting feature. I firstly presumed that was the reason to split and inject at the non inverting inputs, but I was enable to calculate or prove this feature.
3) I see such a ripple removing cap on the Samuel Groner schematic ( Monte Generoso ). I thought the 2 red LED on the Graeme Cohen schematic was specific to the 5532, an input limitation not applicable to more recent op-amps.
4) In the case of using a single integrator I do not see the need to invert its output, you can inject at one leg or the other of the preamp for opposite corrective actions.

EDIT: I must do more work using LTSpice for a better evaluation of alternatives.
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 »

terkio wrote: Tue Feb 19, 2019 1:16 pm 4) In the case of using a single integrator I do not see the need to invert its output, you can inject at one leg or the other of the preamp for opposite corrective actions.
Yes, you can but you get only half the corrective action.
By using a differential output you have twice the voltage swing available.
The servo voltage is attenuated by 4.
A 1 mV offset input with 60 dB gain is going to require 4 volts of correction.

Injecting it into the non-inverting input also only "pulls" one transistor's Ic.
By pulling both transistor's Ic, you only have to pull them half as far.

Then there's T-bias. (Not used here.) Tell me how single-ended correction is suppose to work in that scenario.
Questions about T-bias? https://proaudiodesignforum.com/forum/p ... 191&p=2404 or look at figure 5 here: http://www.thatcorp.com/datashts/THAT_1 ... asheet.pdf I had that text added.

Other than use it in the other channel what would you do with the unused op amp?
Single DC-precise (or any single for that matter) op amps are disappearing...
NJM just disco'd the OP07.
Post Reply