MYTH: "Your signal-to-noise performance is determined by the first gain stage."

[GK.]

And like you are failing to understand - is that the introduction of the attenuator negates the need for the prior gain stages. If you are going to attenute thier ouput, you don't need to include them in the first place. They become irrelevant. The first gain stage is the one that actually provides amplification into subsequent stages. Due to your attenuator, yours does not.


Simply put:
If you are designing a system with multiple stages of amplification, it is likely that given your standard amplifier design, the stages will have a similar level of noise. If the resulting noise level is too high - you are guided to first focus on the first stage, because the noise out of that is amplified by all the subsequent stages. The effect of the work you do at the front end is multiplied by the total system gain through all stages.

The universal necessity of some form of volume control attenuator most certainly does not negate the need for prior gain stages and there is nothing particularly atypical in my example of a MM phono amplifier with 40dB gain proceeded by a volume control potentiometer.

If on the other hand the second stage generates 20 times the noise of the first stage, but only has gain of ten - then yes - you are going to want to fix that one first.

20 times? What is with the hyperbole? If the second stage generates just one tenth the noise of the first stage and there is a passive volume control attenuator between the two, then at just 20dB attenuation both stage are essentially contributing to total system noise more or less equally.

Take that generous phono amp example with its 350nV rt/Hz of self-generated output noise. When then volume is turned down by just 20dB that is reduced to 35nV rt/Hz.
If the first stage is to remain noise dominant then the the second stage really needs to be at least 10dB better than that - so in the vicinity of only 11nV rt/Hz of input noise!

If the second stage is something like a pot wiper buffer followed by something like an active Baxandall tone control circuit or some other kind of active equalization, then 11nV rt/Hz is a very stringent requirement to satisfy!

Furthermore, not all signal chains these days require as much as 40dB of up-front gain. You might have a first stage with just 15dB of gain, which makes the situation far more difficult. This would be typical of the amount of gain required to satisfy old-school sources with nominal 250mV output levels.

Say you make a particularly low-noise 15dB first stage with just 5nV rt/Hz of input-referred noise. With 15dB of gain, were up to only 28nV rt/Hz at the output.
So now to complete your signal chain design, you need to add some form of volume control, tone control, equalization, pan-pot buffer stages or whatever else needed.

In this situation you have practically zero prospects of maintaining a prominent first-stage noise dominance no matter what you do.

 

[antcollinet]

In this situation you have practically zero prospects of maintaining a prominent first-stage noise dominance no matter what you do.

Of course. Because you have already focussed all your efforts on creating a "particularly low-noise 15dB first stage". Which is exactly what this guideline is guiding you to do.

No one is saying that if you have horribly dirty later gain stages, compared with the first that they can't be dominating system noise.

Just that if you need to reduce system noise - start at the device which will give you best bang for the buck - which in a chain of gain stages with similar levels of self noise, will be the first one. In a chain of stages where one has 10dB higher noise than the rest - that one might be the bottleneck - though you'd have to look at the total gain staging to be sure.

 

[antcollinet]

When then volume is turned down by just 20dB that is reduced to 35nV rt/Hz.

And the point I was making, is that in real world listening, 20dB is a massive reduction. 1/10th the voltage. 1/100th the power. Actually 1/400th the power, if you compare it against typical digital sources putting out 2V rather than 500mV.

For my vinyl set up it represents "background music" levels of volume, if that. I never listen to vinyl at that level. If I'm going to the trouble of vinyl then I will be listening at least 10dB higher than that.

 

[GK.]

No one is saying that if you have horribly dirty later gain stages, compared with the first that they can't be dominating system noise.

I am wasting my time here. I am trying in vain to point out just how clean the successive stages actually need to be (so as to usefully maintain first stage noise dominance) to people who clearly have no real comprehension of the noise voltages that I am quoting in the practical sense.

 

[GK.]

And the point I was making, is that in real world listening, 20dB is a massive reduction. 1/10th the voltage. 1/100th the power. Actually 1/400th the power, if you compare it against typical digital sources putting out 2V rather than 500mV.

For my vinyl set up it represents "background music" levels of volume, if that. I never listen to vinyl at that level. If I'm going to the trouble of vinyl then I will be listening at least 10dB higher than that.

That is your system and good for you.

A 20dB reduction in signal voltage is still only a 20dB reduction in power (log 20 for the former, log 10 for the latter) and that is hardly a "massive reduction" by any metric. My run-of-the-mill system comfortably does north of 100dB average SPL with the wick turned up and I don't generally listen to it at any anywhere near that level.

 

[Hayabusa]

That is your system and good for you.

A 20dB reduction in signal voltage is still only a 20dB reduction in power (log 20 for the former, log 10 for the latter) and that is hardly a "massive reduction" by any metric. My run-of-the-mill system comfortably does north of 100dB average SPL with the wick turned up and I don't generally listen to it at any anywhere near that level.

a 20dB reduction in signal level is NOT a 20dB reduction in power level..
10log10(power) gives you the level ratio..

 

[antcollinet]

real comprehension of the noise voltages that I am quoting in the practical sense.

You really are putting yourself on a pedestal aren't you, when you consider the level of engineering knowhow of the membership here.

I am trying in vain to point out just how clean the successive stages actually need to be

They only need to be clean to the extent that their noise is less than or equal to

(noise-level-of-first-stage x cumulative-gain-of-prior-stages-including-any-contrived-passive-attenuation) in order for the noise of the first stage to be dominant.

So to take your Phono pre example, if you attenuate by the same as the 40dB gain of the preamp (though why you'd do that is a mystery) then yes if the input noise levels of each stage is the same, they will contribute equally.

However if you only attenuate by 10dB (as I do for real world listening at a sensible level) then the second stage input noise has to be 30dB higher to become equal or dominant.

But here is the real kicker - how many vinyl play back systems do you know of where there is a passive volume control between the Phono Pre, and the input to the next stage whatever that might be. Most phono preamps have input, output, and sometimes selectable gain for MM or MC cartridges. The next stage will normally be a preamp, or an integrated preamp - that will receive the full 40dB amplified input noise from the Phono Pre to be added to its own input noise.

In your diagrams above you should simply cross out your passive attenuator. It (almost) never exists. It is an artificial construct you are using to bolster your argument.

 

[antcollinet]

a 20dB reduction in signal level is NOT a 20dB reduction in power level..

Yes it is. That is why doubling of signal/voltage is 6db where doubling of power is 3dB. It is why we use 10log (p1/p2) for power, and 20log(v1/v2) for voltage (or current)

Think about it.

A 20dB reduction in signal voltage is still only a 20dB reduction in power

(my bold)

20db reduction in power is a 100x reduction. It is huge. It reduces perceived volume by roughly 4x.

 

[Waveform Fidelity]

Are there any qualified electronic engineers on this discussion? Ive read a lot of non-sense

 

[xanalog]

I'm not going to retype my opening post.

You may not wish to retype it but you need to correct a nasty error that permeates your otherwise worthy note to others.

and an input-referred voltage noise of 3.5nV rt/Hz

The proper term "3.5nV/rt(Hz)" refers to a voltage noise spectral-density of 3.5nanoVolts per square-root of bandwidth in hertz.

OT: Such noise-voltage over a given bandwidth is calculated by multiplying the noise spectral-density (in V/√Hz) by the square root of the BW (in Hz).

 

[a4eaudio]

I am wasting my time here. I am trying in vain to point out just how clean the successive stages actually need to be (so as to usefully maintain first stage noise dominance) to people who clearly have no real comprehension of the noise voltages that I am quoting in the practical sense.

Considering almost 100% disagreement by others, have you considered (1) you are wrong, or (2) you aren't communicating your points clearly?

 

[antcollinet]

Are there any qualified electronic engineers on this discussion? Ive read a lot of non-sense

Puts hand up (though retired).

Care to point out any nonsense I've stated? I'm always happy to learn if I'm getting specifics wrong.

 

[solderdude]

Are there any qualified electronic engineers on this discussion? Ive read a lot of non-sense

Best to specify what is nonsense and what's not instead of handing out a blank statement that could apply to anyone who stated anything.

 

[Waveform Fidelity]

Puts hand up (though retired).

Care to point out any nonsense I've stated? I'm always happy to learn if I'm getting specifics wrong.

Best to specify what is nonsense and what's not instead of handing out a blank statement that could apply to anyone who stated anything.

Ok you make a good point - a problem of making a general comment in a large forum

 

[DonH56]

I have rarely seen noise figure used in the context of an audio system, not that it matters. There is this thread discussing noise and distortion propagating through the signal chain: https://www.audiosciencereview.com/...opagate-through-my-system.33358/#post-1165118

 

[DVDdoug]

For many volume controls it is desirable to have a maximum attenuation which would exceed the stability threshold of the amplifier circuit.

I learned that lesson decades ago trying to make a variable gain preamp! This was in the early op-amp days when op-amps were often not zero-gain stable. My preamp turned-into an RF oscillator. It was super quiet so I was impressed with myself until I heard a little pop and one side of my power amp was blown! I didn't have an oscilloscope so I've only deduced what happened. I don't remember if I got a signal through it before the power amp blew but there was zero (audible) noise.

It was a valuable (and expensive) lesson but I remember it decades later and now I'd be very careful about using a potentiometer in a feedback circuit. My design wasn't around long enough to have a dirty-pot problem but that would be an important consideration. (And if my memory is correct, I was thinking it could be an issue... I might have put a resistor in parallel with the pot to minimize the effects of that but I don't remember. Even with the resistor it could be a "bad problem" but not as bad.)

I don't remember any noise problems once I re-designed it the usual way with an attenuator but I was thinking I would be better when the circuit was operating at lower gain, or as an attenuator. It was a low-gain line-level preamp so attenuation wouldn't be unusual. But I was almost always "fighting noise" when I built phono preamps. Thankfully, I also haven't played records in decades!

 

[tmtomh]

The best thing you can say for the first stage is that it sets an absolute limit to the achievable noise performance.

I would respectfully suggest that a good deal of what appears to be technical or engineering disagreement here is actually semantic.

"The first stage... sets an absolute limit to the achievable noise performance" is to many folks functionally equivalent to, "Your signal-to-noise performance is determined by the first gain stage." To be clear, I understand these two statements are not 100% identical. But the issue, I think, is that most folks read the second statement to mean, "The limit/maximum potential of your signal-to-noise performance is determined by the first gain stage." Again, to be clear, of course the part I've bolded is implicit.

Nevertheless, because most folks understand the statement to mean that - and because of how most folks understand the word "determined" - it strikes them as essentially equivalent to say the first stage determines the noise performance, or to use your own statement that "the first stage... sets an absolute limit on the achievable noise performance." Setting an absolute limit is precisely what "determines" means in this context.

I think everyone understands that subsequent gain stages can further degrade the noise/S-N performance from that absolute limit. But the idea that the first stage determines the performance precisely means (to most folks, anyway), that you can't exceed the performance of that first stage - you can only equal it or get worse than it.

So it's not that folks don't understand your point - it's that they think your point is obvious and therefore by no means a "myth-busting" insight. To put it another way, they think you're straw-manning the so-called myth - you've mischaracterized what the so-called myth is claiming.

 

[Cbdb2]

Your numbers work because 40db of preamp attenuation is a huge amount. With a 100 watt amp your putting .01 watts into a speaker. What are those numbers at a realistic 20db?

 

[TurtlePaul]

Your numbers work because 40db of preamp attenuation is a huge amount. With a 100 watt amp your putting .01 watts into a speaker. What are those numbers at a realistic 20db?

Instead of power math, it is better/more intuitive to do the gain math.

0 dB is 2Vrms on my DAC
+ 86 dB is the sensitivity of Revel towers at 2V
+ 26 dB is the gain of my amp for RCA input
+ 6 dB for stereo speakers playing in phase
- 6 dB for dynamic range of modern music
- 6 dB for sitting at 2 meters

The result is that full blast on my system is ~103-107 dB in room on modern recordings from my CD player.

I absolutely use -40 dB attenuation which gets me into the 60s dB for background music, TV, etc. And I don’t even have particularly sensitive speakers nor do I use my amps high gain switch.

 

[FrantzM]

I have rarely seen noise figure used in the context of an audio system, not that it matters. There is this thread discussing noise and distortion propagating through the signal chain: https://www.audiosciencereview.com/...opagate-through-my-system.33358/#post-1165118

O.T

Hi @DonH56 long time no post. Glad to read you!