Preamplifiers, gain and volume control - is the usual architecture really clever?

[Peter A]

The usual architecture: we have some kind of a source, producing a fixed level of signal. Volume control is achieved by reducing the level of source signal to some mV. There is a fixed gain power amplifier which amplifies this tiny level signal again.

In terms of total chain SINAD a more logical approach appears to create a variable gain power amplifier. I am not an electronic engineer, to my opinion, some amplifier topologies would allow this approach.

1. Any kind of amplifier with feedback allows to increase the feedback down to unity gain
2. voltage amplification of any kind followed by unity gain power buffer. same as 1
3. Class D amps with switching PSU for the output stage. Digitally controlled voltage for the output stage

If we have the technology to build R2R DACs measuring as good as a Holo Audio MAY DAC, it should be possible to realise a digitally controlled variable feedback, solving all possible problems with instability, RON etc.

Is there any experience with this approach?
Any ideas, thoughts, feedback?

Taking this one step further, why not create a discrete delta sigma DAC with a beefy output stage capable of driving loudspeaker. In combination with the variable gain approach this results in one unit combining a DAC, pre amp and power amp in one architecture.

 

[Ropeburn]

I guess the reason this isn't done or rarely ever, is that the current convention lends itself much more to a "building blocks" approach that is very practical and convenient during the design process. It's much like car platforms: different engines and trim levels and extras and even body types, all relatively easy to integrate into a specific product. The preamp + fixed gain poweramp concept is nicely flexible, also because any processing is better done on signal, hence preamp level.

Investing development time and money into a power stage volume control seems like wasted effort; because the existing approach is already so good in every aspect, there would be little to nothing to gain from it.

 

[MaxwellsEq]

The usual architecture: we have some kind of a source, producing a fixed level of signal. Volume control is achieved by reducing the level of source signal to some mV. There is a fixed gain power amplifier which amplifies this tiny level signal again.

In terms of total chain SINAD a more logical approach appears to create a variable gain power amplifier. I am not an electronic engineer, to my opinion, some amplifier topologies would allow this approach.

I feel you are not wrong - this could be better.

The issue comes from history. In the 70s and 80s before CD, everything was reasonably stable. Most people probably had integrated amplifiers rather than pre/power combinations. Professional balanced gear was rarely used at home. People knew with confidence that 1V was the line level standard on unbalance RCA. This meant that power amplifier gains (for those with pre/power combos) were necessarily high. Preamplifiers would expect to amplify and RIAA eq turntable feeds and for everything else, would accept 1V in and produce 1V (necessarily simplified). CD came out at 2V. This meant some preamplifiers and integrateds were overloaded, but even if they weren't, there was a big volume step when switching between a cassette deck at 1V and a CD player at 2V. Then DACs started adding variable attenuation (acting a bit like a preamplifier). Then people started using balanced gear at 4V. If your entire system is balanced at 4V you need a lot less gain in the power amplifier.

A example problem is that if people have a big mix of balanced, SE, 1, 2 and 4V, what is the correct gain of a power amplifier? High 20s and low 30s for 1V operation, 15ish for 4V operation. So power amplifiers could have gain controls, and many do. Mine has 3 fixed setting, adjusted by DIP switches on the input board. Professional power amplifiers often have gain controls which can be "locked off" to prevent tampering once set.

 

[ppataki]

Correct me if I am wrong but why not simply use digital volume control upstream (64-bit precision, like the one in Jriver that I am using) and leave everything else downstream in the signal chain on max volume (if there is volume control on those devices)

I think this way you maximize all the key performance indicators

 

[JeffS7444]

Correct me if I am wrong but why not simply use digital volume control upstream (64-bit precision, like the one in Jriver that I am using) and leave everything else downstream in the signal chain on max volume (if there is volume control on those devices)

Fear of digital volume control randomly going to 100% (yes it happens).

 

[ppataki]

Fear of digital volume control randomly going to 100% (yes it happens).

I have been using it for 10 years and when that happened it only happened because of my fault
Typically when I have Tidal open in Exclusive mode and trying to play something in Chrome - in that case Jriver's engine (WDM driver) gets bypassed and I get 100% volume to the speakers
But this is the only scenario (for me) when that happens

 

[unpluggged]

Fear of digital volume control randomly going to 100% (yes it happens).

Then ensure that even when it happens the SPL still remain safe. Adjust the gain on the amp or use fixed attenuators. Done.

 

[rdenney]

I dunno. It seems to me that all devices are a series of gain stages—amplifiers—and the boundary between one part of the series and another is arbitrary for convenience. A classic preamp (standalone, integrated with an amp, or integrated into a receiver) has amplification stages around every control or filter. It isn’t just attenuation. And some add in amplification rather than attenuating an amplified signal.

And the so-called power amp has stages, too.

Each stage involves some sort of device that has limited gain all by itself, so they have to be strung in series to convert the source signal (which may be millivolts) to the tens of volts needed to move the speaker cones. And then there’s the balance between impedance and current capability, which changes throughout the gain stages, too, requiring different strategies.

But as it turns out, providing a line-level interface greatly contributes to flexibility, allowing products to be mixed to match a much wider array of use cases. That counts for a lot—probably more important than wringing another few dB from a system that is already orders of magnitude cleaner than the transducers on both ends of the production/playback chain.

Rick “6 dB here, 6 dB there, pretty soon you’re talking about a lot of gain” Denney

 

[Peter A]

Correct me if I am wrong but why not simply use digital volume control upstream (64-bit precision, like the one in Jriver that I am using) and leave everything else downstream in the signal chain on max volume (if there is volume control on those devices)

I think this way you maximize all the key performance indicators

It's about optimising performance. With high attenuation the digital control makes the DAC work outside of optimal parameters. this might still be better or on the same lewvel with best analog preamps.

The variable gain approach lets the DAC work full range and just amplifies the signal amplitude as much as needed

 

[Peter A]

I dunno. It seems to me that all devices are a series of gain stages—amplifiers—and the boundary between one part of the series and another is arbitrary for convenience. A classic preamp (standalone, integrated with an amp, or integrated into a receiver) has amplification stages around every control or filter. It isn’t just attenuation. And some add in amplification rather than attenuating an amplified signal.

And the so-called power amp has stages, too.

Each stage involves some sort of device that has limited gain all by itself, so they have to be strung in series to convert the source signal (which may be millivolts) to the tens of volts needed to move the speaker cones. And then there’s the balance between impedance and current capability, which changes throughout the gain stages, too, requiring different strategies.

But as it turns out, providing a line-level interface greatly contributes to flexibility, allowing products to be mixed to match a much wider array of use cases. That counts for a lot—probably more important than wringing another few dB from a system that is already orders of magnitude cleaner than the transducers on both ends of the production/playback chain.

Rick “6 dB here, 6 dB there, pretty soon you’re talking about a lot of gain” Denney

Yes - mixing and matching.

I am all about optimising an all in one. A fully integrated system from the source to the speaker with optimised noise floor, gain.

Somebody tried one of this ideas?Variable gain amp or discrete delta sigma dac with power output?

 

[boXem]

1. Any kind of amplifier with feedback allows to increase the feedback down to unity gain

Absolutely not, a lot of them will become unstable

1. Class D amps with switching PSU for the output stage. Digitally controlled voltage for the output stage

As soon as feedback is introduced, doesn't work anymore. And class D without feedback is just bad or overly complicated.

Taking this one step further, why not create a discrete delta sigma DAC with a beefy output stage capable of driving loudspeaker. In combination with the variable gain approach this results in one unit combining a DAC, pre amp and power amp in one architecture.

It has already been done, doesn't work well. Doing the analog conversion at very high frequencies then amplifying in the analog domain provides much better results.

That said, most amplifiers have several stages, and the power stage is quite often a relatively low gain one. To stay with high performance class D, Hypex and Purifi modules have gains comprised between 12 and 14 dB.

 

[Fledermaus]

Correct me if I am wrong but why not simply use digital volume control upstream (64-bit precision, like the one in Jriver that I am using) and leave everything else downstream in the signal chain on max volume (if there is volume control on those devices)

I think this way you maximize all the key performance indicators

In my understanding, this way of doing things tends to maximize the noise component, compounding it from each stage to the next. In that respect, the best would be to attenuate as downstream as possible - including padding of speaker drivers, like it is sometimes suggested to reduce hiss in sensitive (horn) speakers when all the available amp power is not needed. In practice, high level signal from source (meaning higher SNR, through preamp if need be) into lowish-gain amp is to be favored, at least that's what I aim for.

 

[ppataki]

64-bit digital volume control has theoretical noise floor at approximately -384 dBFS
When you digitally reduce volume by let's say 40 dB in a 64-bit system, your effective noise floor is still around -344 dBFS - vastly below what any physical system can reproduce (or what humans can hear)

 

[dualazmak]

64-bit digital volume control has theoretical noise floor at approximately -384 dBFS
When you digitally reduce volume by let's say 40 dB in a 64-bit system, your effective noise floor is still around -344 dBFS - vastly below what any physical system can reproduce (or what humans can hear)

Essentially agree with you, and this is one of the reasons for my utilization of top-upstream JRiver MC 64-bit as "master volume/gain" controller, as shown in the total signal path diagram under the below spoiler cover.

Spoiler

Ref. post #931 on my project thread:

My stance/policy in relative gain/volume control in multichannel multi-SP-driver multi-amplifier active system, however, is to utilize the "best combination" of "DSP configuration in digital domain" and "analog domain gain controls using HiFi-grade preamplifiers and/or integrated amplifiers".
What would be the rationales for this my stance/policy, at least in my project case?
In my post #931 on my project thread, I wrote as follows under the below spoiler cover, if you would be interested...

Spoiler

.... as for safe and flexible tone controls (or I can say "relative gain controls among the multiple SP drivers"), my stance (policy) at least, is that we are encouraged to utilize the "best combination" of "DSP configuration in digital domain" and "analog domain tone controls using HiFi-grade preamplifiers and/or integrated amplifiers".

We need to note (and to respect for) that analog domain tone controls (relative gain controls among the multiple SP drivers) give no effect nor influence at all on the upstream DSP configuration (XO/EQ/Gain/Phase/Polarity/Group-Delay). I believe that this is a great merit of flexible tone controls in analog domain. We know well, on the other hand, in case if we would like to do the "tone/gain controls" only within DSP configurations, such DSP gain controls always affect more-or-less on "XO" "EQ" "phase" and "delay" of the DSP settings which will leads you to possible endless DSP tuning spirals every time; within DSP configurations, XO EQ Gain Phase and Delay are always not independent with each other, but they are always interdependent/on-interaction.

Just for your possible reference, my DSP-based multichannel multi-SP-driver multi-amplifier active system has flexible and safe analog level on-the-fly relative gain controls (in addition to upstream on-the-fly DSP gain controls) for L&R subwoofers, woofers, midrange-squawkers, tweeters, and super-tweeters, all independently and remotely. My post here shows you a typical example case for such safe and flexible on-the-fly analog-level tone controls. This my post (as well as this post) would be also of your interest.

Of course, I know well that I (we) can also perform such relative gain control using DAC8PRO’s 8-channel output gain controllers. I do not like, however, to change the DAC8PRO’s output levels frequently on-the-fly (while listening to music) due to safety and inconvenience concerns; I like to keep DAC8PRO’s analog out gain level always at constant -4 dB which should remain to be usually “untouchable” in my case.

One of the very unique aspects/features of my multichannel audio rig is that I fully utilize four HiFi-grade “integrated amplifiers” plus L&R active subwoofers, each of them have its own gain (volume) controller for safe and flexible relative gain (tone) control in analog domain even on-the-fly i.e. while listening to music.

In this perspective, my posts #438 and #643 should also give you better understandings. Furthermore, my posts #317(remote thread), #313(remote thread) would be also of your reference and interest.

Talking further about "safety measures" for SP drivers, my preference and actual implementation are "protection capacitors" in SP high-level signals before going into SP drivers, as well as safe "start-up/ignition sequences" plus safe "shutdown sequences" as shared in the bottom portion of my post #931 on the project thread.