Small Signal Circuits vs Discrete Opamp

[user 125886]

I have been studying small signal amp circuits in general. All the examples I have come across are the typical common emitter ones with the voltage divider at the beginning.
I would not even be studying these if it weren't for the classic Neve small signal circuit. I wanted to know why the neve is the only pro preamp that uses these with such amazing success.
It seems that all the other big name preamps uses either tube or discrete opamps. They both seem to have your typical transistor network like in these pictures below, except the opamps use positive/negative rails, and the neve uses the single 24V rail. Other than the transformer differences, Which one of these typologies makes for a better way to amplify a small signal? And why?

 

[Bo Deadly]

It's not correct to compare them as if they were all equally accessible options. There was an iterative progression from tubes to Germanium amps, 2-3 transistor circuits and then discrete op amps. Once IC op amps were good enough, they took over and then CFP and then finally today we have CFIA in just about everything right down to the cheap mixers.

Why would you want to use one over the other? Many folks believe that the 2-3 transistor circuits contribute desirable harmonics or that the discrete op amps sound better for some reason. Personally I think that's all a bunch of hooey because there's enough open loop gain and feedback that the harmonics are going to be really far down and once it starts clipping it's going to be instantly harsh.

 

[user 125886]

Would it be safe to assume that, as long as a preamp design is class A and has quality transformers, it's probably going to sound "pretty darn good". No matter if it's tube, opamp, or 2-3 transistor?

 

[JohnRoberts]

Would it be safe to assume that, as long as a preamp design is class A and has quality transformers, it's probably going to sound "pretty darn good". No matter if it's tube, opamp, or 2-3 transistor?

Not safe at all...

JR

 

[ruffrecords]

When I was at Neve back in the mid 70s the R&D department was trying out a new op amp from Philips (which later became the 5534). At that time Neve was transitioning from the 3 transistor circuit to its own DOAs (BA440 etc). Not many years later opamps appeared in all Neve products.

Cheers

Ian

 

[abbey road d enfer]

I wanted to know why the neve is the only pro preamp that uses these with such amazing success.

It's a marketing success, capitalizing on the number of hit records.
Actually, there are many more hits produced with an MCI console that uses IC's. But MCI is not there anymore and Sony have never tried to capitalize on the MCI name.
Today's Neve have chosen to capitalize on the recreation of vintage units, but Rupert Neve himself had long chosen using more modern designs.

It seems that all the other big name preamps uses either tube or discrete opamps.

Big names to whom? Do not confuse marketing presence with actual industrial success.

They both seem to have your typical transistor network like in these pictures below,

There are not that many basic implementations of a transistor-based amplifier. Typically there's an input stage, a gain stage and an output stage. Sometimes one transistor does two things, and sometimes two transistors are used to do the job of one, and there are many differeing details, so it may seem to you that there is an almost infinity of designs, but actually there are more common than differing aspects.

the neve uses the single 24V rail.

Single rail is a remnant of the tube era. When designers found out they could use bipolar rails, they fully understood the advantages. Many did not look back.

Other than the transformer differences,

Transformers were another remnant of the tube era, where they were necessary for adapting impedance, because tubes are high-impedance devices, that need impedance adaptation to drive the typical 600 ohms load of lines and the nominal 8/16 ohms of loudspeakers.

Which one of these typologies makes for a better way to amplify a small signal?

Depends on your expectations.
If you want neutrality and transparence, a 1073 or 1081 is not the best choice.
And there are many ways to achieve neutrality and transparence, but it's generally not what people want; they want colour, mojo, larger-than-life. Since these are subject to individual choices (google "acquired taste"), there are as many "best" than there are individuals.

 

[ruffrecords]

Of course, the other very well known three transistor preamp is the one by Helios.

Cheers

Ian

 

[ruffrecords]

There is an old adage from the world of Unix software applications that states an application should do one thing and do it well. The simple Neve circuits largely follow this adage. There is the standard low noise pre for mic inputs, the 2N3055 based power stage for outputs and the odd little op amp used in EQ circuits.

One could argue that DOAs subscribe to the "one size fits all" philosophy. They have incredible output power, which is not needed for a mic pre input stage, they have excellent noise properties which are not essential for an output stage and they get used as op amps in EQs.

I think Calrec had the right idea; they had a common 4 transistor op amp circuit and PCB and simply changed driver, and output transistors and their bias depending on the application.

Cheers

Ian

 

[rackmonkey]

A couple of thoughts for consideration beyond the fine answers above.

I think you’re focusing on the wrong module of the Neve boards referenced. The output module - the BA283 - is key to understanding the appeal of the old Neve consoles (which is real, but pales in comparison to the sociological power of the internet in contributing to its mythical status).

The 283 is a single-ended design coupled to the outside world by a steel core transformer. It’s a bit of a rarity when compared to the later transistor era and the world of ICs, where various push-pull output stages tend to dominate. P-P cancels out even order harmonics to a degree, while SE does not. Steel core transformers tend to be more colored than those with 49% or higher nickel content, and tend towards a more appealing distortion profile when lightly pushed (transformer distortion is ugly across the board beyond the earliest overload point). These two factors more than anything else contribute to what people perceive as the sound of the 1073 and its siblings.

obviously, the amount of feedback can cancel output stage harmonics, and if enough is applied, make the differences mostly moot. But that’s not the case here.

2 cents from the gallery.

 

[user 125886]

Fine answers indeed. Much to think about.

What jumps out at me at this point is what Mr. Abbey Road and Rackmonkey point out about the transparency of DOA's and the major sound influencing factor of the output stage. It would seem that it would make more sense to just plug in a discrete opamp for ease and simplicity for the gain stage. Then go about focusing all your attention on the output stage in order to find the right "coloration". Thus relying completely on the output stage and the output transformer for the "sonic character" of the design.

 

[JohnRoberts]

This is a fairly old topic... (consider a search).

The last time I designed a DOA was back in the 70s/80s because I could not buy an off the shelf op amp that quiet to use in a console bus combining amp. Since then IC op amps have improved a lot, DOAs not so much.

For transparency I guess linearity is the best actual objective specification. I would expect modern high performance op amps to spank the popular DOAs.

JR

 

[ruffrecords]

A couple of thoughts for consideration beyond the fine answers above.

I think you’re focusing on the wrong module of the Neve boards referenced. The output module - the BA283 - is key to understanding the appeal of the old Neve consoles (which is real, but pales in comparison to the sociological power of the internet in contributing to its mythical status).

The 283 is a single-ended design coupled to the outside world by a steel core transformer. It’s a bit of a rarity when compared to the later transistor era and the world of ICs, where various push-pull output stages tend to dominate. P-P cancels out even order harmonics to a degree, while SE does not. Steel core transformers tend to be more colored than those with 49% or higher nickel content, and tend towards a more appealing distortion profile when lightly pushed (transformer distortion is ugly across the board beyond the earliest overload point). These two factors more than anything else contribute to what people perceive as the sound of the 1073 and its siblings.

obviously, the amount of feedback can cancel output stage harmonics, and if enough is applied, make the differences mostly moot. But that’s not the case here.

2 cents from the gallery.

Some good points there. And the BA283 does not take NFB from the transformer secondary so it does not affect the 'sound' of the iron. Mind you, the same is true of API output stages.

It is also interesting to note that in just about every amplifier topology, the distortion is level dependent. So in the mic pre stages where the signal is tiny, the distortion is low but it is greatest in the output stages.

Cheers

Ian

 

[JohnRoberts]

It is also interesting to note that in just about every amplifier topology, the distortion is level dependent. So in the mic pre stages where the signal is tiny, the distortion is low but it is greatest in the output stages.

Cheers

Ian

Isn't the output of a mic preamp full level?

JR

 

[Heikki]

Looks like in BA283 the negative feedback is taken from the primary side.

How does taking negative feedback from the primary side of a transformer not affect the "sound" of the iron. Do all the transformer nonlinearities run and hide away and only stay at the secondary side?

 

[ruffrecords]

Isn't the output of a mic preamp full level?

JR

Not in a 1970s Neve. With a 24V power rail, to ensure there is 26dB of head room, the mic pre output is typically at about -8dBu and this is the nominal level throughout the console. The 12dB need to reach +4dBu is provided by the power output stage.

Cheers

Ian

 

[ruffrecords]

Looks like in BA283 the negative feedback is taken from the primary side.

How does taking negative feedback from the primary side of a transformer not affect the "sound" of the iron. Do all the transformer nonlinearities run and hide away and only stay at the secondary side?

It lowers the impedance driving the transformer. This does tend to reduce the distortion produced by the transformer and extends the LF response but not by as much as when the feedback is taken from the output of the transformer.

Cheers

Ian

 

[JohnRoberts]

Not in a 1970s Neve. With a 24V power rail, to ensure there is 26dB of head room, the mic pre output is typically at about -8dBu and this is the nominal level throughout the console. The 12dB need to reach +4dBu is provided by the power output stage.

Cheers

Ian

+26 dB headroom above -8dBu, peaks around +18dBu which is not exactly whisper level, but I concede driving 600 ohms will require more current and result in higher distortion. Active differential outputs adds 6 dB more level making that +18dBu a respectable +24 dBu while if you are driving 600 ohms that makes it dBm.

JR

PS: I used to run my singled ended line level signals around inside large consoles at -2 dBu nominal (-6dB below +4dBu), for consistent headroom inside and out.

 

[fragletrollet]

I recently got made some filter pcb's based on the Neve 8128 filter section with input receiver and output driver. All IC's and a very different circtuit to the 73/84 stuff. We have a (AMS) 8801 channel strip in the studio, and it sure sounds different to the 73/84. Not very strange when you look inside:



All 5532/34's and surprisingly (to my uneducated perception) lots of tl052's. Trafo's made in Taiwan, no inductors and totally different than the early stuff. I'm not sure Rupert had anything to do with AMS, but I presume this is a clone of the dynamics & eq section of a channelstrip of the original console. If you'll read on pages like Gearslutz you'll see massive opposition/disapprovement of these newer Neve designs, and everybody praising the "classics". I think it sounds great, cleaner than the 73 for sure but still has the wolly warm preamp characteristic and an eq that just sounds sweet. Very different from my SSL's that also use a ton of 553x!

I remember reading an interview with Rupert, where he emphasized how he's always been on the quest for transparency and accurate reproduction/fidelity of sound. He never intended to make a "warm" preamp in making the 1073; and he gladly welcomed modern IC's as they became available to alter his designs, in order to lower distortion and noise.


I wonder what would have happened if Rupert was born today, but still had a career in audio electronics...

 

[fragletrollet]

And I never knew that Americas former president had a part in the design...

 

[ruffrecords]

Rupert never had anything to do with AMS. When I joined Neve in 1972 Rupert had already sold the company the Bonochotd. He was retained as a consultant but had nothing to do with the day to day running of the company. Geoff Watts, the principal designer, was still at Neve and by this time there was a fully fledged R&D department. Many years later Neve was sold to Siemens and some years later they sold it to AMS.

Cheers

Ian