Following on from the article by Pano on gain structure and Nelson's comments in the original B1 article...
The standard approach appears to be a voltage divider on the input to a preamp, however, assuming a preamp might be stable over a wide range of gain, wouldn't we be better off in terms of SNR to control the output voltage via variable gain and use all of the available ~2V rms from a DAC input all of the time..?
The standard approach appears to be a voltage divider on the input to a preamp, however, assuming a preamp might be stable over a wide range of gain, wouldn't we be better off in terms of SNR to control the output voltage via variable gain and use all of the available ~2V rms from a DAC input all of the time..?
All roads are leading to Rome
my Rome is having a Name Sign showing "Signal Energy Preservation" ........ meaning that I' always trying to avoid any form of resistive attenuation
that leads to nasty ( read - expensive and space demanding ) solution of using inductive attenuators, naively thinking that signal is happy being "just" transformed, changing shape in realms of voltage and current, rather than just being thrown to South Pole in form of wasted heat through resistor bodies
anyhow, variable gain stages are also specific can of worms, and only way of deciding what to do is - make it, then compare with solutions you did before, then choose what's warmer to your heart
incidentally, one of my ideas is to try I/V conversion in role of attenuating block, but we'll see in which decade I'm going to that line of list
regarding I/V - now I remember that Yamaha had few (mass produced) integrated amps with that....... they didn't use it for long
still regarding I/V - and I remember some boutique manufacturers had fancy name solutions in that spirit, pretty much only benefit was for them - you couldn't combine any other manufacturer's component with their precious ones
you'll also find references (mainly in Pro World) for VC (Voltage Control) gain stages/volume control; I mean - you'll find integrated circuits for that and also references for entire gadgets using that ..... say mixing consoles of Yore, each channel having faders governing DC voltage going to corresponding functional blocks
go figure
anecdote - Musical Fidelity A1 ( early,. original one) had OS constructed by no other else than Tim de Paravicini; little jewel, no doubt
now - phono stage was so-so decent (showing entirely different mental approach), while line stage** was catastrophe, I hope rather constructed by someone else.... in case that even that was Tim's work, I can bet he had nasty headache these days, or he was using some nasty weed
**OP having variable resistor (pot, labeled as Volume on front) in NFB loop, so stage was variable gain stage; it sound as utter rubbish, and also eats ("volume") pot in few years, due to some stubborn DC in NFB loop
my Rome is having a Name Sign showing "Signal Energy Preservation" ........ meaning that I' always trying to avoid any form of resistive attenuation
that leads to nasty ( read - expensive and space demanding ) solution of using inductive attenuators, naively thinking that signal is happy being "just" transformed, changing shape in realms of voltage and current, rather than just being thrown to South Pole in form of wasted heat through resistor bodies
anyhow, variable gain stages are also specific can of worms, and only way of deciding what to do is - make it, then compare with solutions you did before, then choose what's warmer to your heart
incidentally, one of my ideas is to try I/V conversion in role of attenuating block, but we'll see in which decade I'm going to that line of list
regarding I/V - now I remember that Yamaha had few (mass produced) integrated amps with that....... they didn't use it for long
still regarding I/V - and I remember some boutique manufacturers had fancy name solutions in that spirit, pretty much only benefit was for them - you couldn't combine any other manufacturer's component with their precious ones
you'll also find references (mainly in Pro World) for VC (Voltage Control) gain stages/volume control; I mean - you'll find integrated circuits for that and also references for entire gadgets using that ..... say mixing consoles of Yore, each channel having faders governing DC voltage going to corresponding functional blocks
go figure
anecdote - Musical Fidelity A1 ( early,. original one) had OS constructed by no other else than Tim de Paravicini; little jewel, no doubt
now - phono stage was so-so decent (showing entirely different mental approach), while line stage** was catastrophe, I hope rather constructed by someone else.... in case that even that was Tim's work, I can bet he had nasty headache these days, or he was using some nasty weed
**OP having variable resistor (pot, labeled as Volume on front) in NFB loop, so stage was variable gain stage; it sound as utter rubbish, and also eats ("volume") pot in few years, due to some stubborn DC in NFB loop
A voltage divider followed by a voltage gain stage where the gain is fixed and set to give the capability to hit the maximum desired value, means that for much of the usage, (whenever not being operated at maximum level), the input signal is being divided down and the amplified back up. In this case and in the scenario where there is some minimal fixed amount of noise, then after it is amplified (along with the signal), the SNR is not as good as it could be.
But as you might have suspected my approach is much more simplistic...
Assuming for a moment one can trust the output from LTSpice...
Take the DIY FE 2022 circuit
https://www.firstwatt.com/pdf/art_diy fe 2022.pdf
Gain as drawn is 10x, but change R3 and R4 to 5k and the gain changes to <1.
Replace R3 and R4 to 5k ohm resistor and add a variable resistor 0-95k in series (2-gang pot per side or 4-gang pot for stereo), and this would give you variable output over a range which makes the front end something from a buffer to a voltage gain stage. Assuming that this covers the range over which you would listen, then all else you need is a mute button. You can pump the entire signal into the unit without loss, and so maintain the best SNR..
I figure my assumption are flawed and the gain stage is not equally stable with gain, or that the fixed component of noise in the system is vanishingly small, or something else. Voltage dividers are after all ubiquitous.
But as you might have suspected my approach is much more simplistic...
Assuming for a moment one can trust the output from LTSpice...
Take the DIY FE 2022 circuit
https://www.firstwatt.com/pdf/art_diy fe 2022.pdf
Gain as drawn is 10x, but change R3 and R4 to 5k and the gain changes to <1.
Replace R3 and R4 to 5k ohm resistor and add a variable resistor 0-95k in series (2-gang pot per side or 4-gang pot for stereo), and this would give you variable output over a range which makes the front end something from a buffer to a voltage gain stage. Assuming that this covers the range over which you would listen, then all else you need is a mute button. You can pump the entire signal into the unit without loss, and so maintain the best SNR..
I figure my assumption are flawed and the gain stage is not equally stable with gain, or that the fixed component of noise in the system is vanishingly small, or something else. Voltage dividers are after all ubiquitous.
Been in Rome twice... the airport. Mid 60s, on taxi in an Alitalia DC6, in the rain, I could see from my window the exhaust from the prop engines.... it was on fire! I had never flown a prop before, only jet Caravelles.
Little Tony yelled out "Fuego! Fuego!". The stewardess came quickly and told me to sush, it was just the exhaust... Nice lady... Thank God the passengers were Italians and did not understand Castillian. My poor grandmother was mortified but she understood.
Anyhow... if we are not going to throw away energy via resistance... how about variable Vcc? Is that what you mean? How about a TVC like set up in the power supply? Control the gain by modulating the power supply output voltage via the transformer.
No power/energy loss anywhere. Just the required voltage and run the preamp wide open throttle all the time.
I've been thinking about transformers a lot lately... thanks.
Just return from Rome yesterday, beautiful city!
Back to the topic, it reminds me those fancy terms "Variable Gain Transconductance" from Ayre and PS Audio's Gain Cell...
If you use digital source the attenuation can be done in the digital domain. If up-sampled enough it can be done "mathematical perfect".
In digital domain it is "just" a multiplication with a factor.
Apart from that a solution could also be just put the full signal from source directly into power amp and then have different power amps with different gain so depending on how loud the recording is you can choose the right amp. We build so many power amp here......from source followers to some having 10-15 dB of gain. Having many speakers with sensitivity from 80 to 100 dB can also be a solution and then just select the right set of speakers.
Right now I play using the kit VFET amp which has an Edcor transformer for the voltage gain at input. My source is a DAC where I use the digital vol control of it. So right now I waste not much energy in signal transfer. If amp had no Edcor (just a follower) I may be close to be able to play just using fixed output from DAC.....but I think it still will be to loud for most tracks. A "variable" Edcor could be a solution......
I am building a source follower amp in the moment and interesting to test if using fixed output could be a solution for some low volumen recorded tracks.
I am also waiting for a preamp kit called Iron Pre. It has a bit of it as the output gain can be set via a transformer to a few fixed values......but still has a vol pot.
If the output transformer had "enough" fixed values they could be used instead of vol pot. I know this solution exists in another pre but I can't remember the details.
To just have the gain needed will also give the best SNR.
In digital domain it is "just" a multiplication with a factor.
Apart from that a solution could also be just put the full signal from source directly into power amp and then have different power amps with different gain so depending on how loud the recording is you can choose the right amp. We build so many power amp here......from source followers to some having 10-15 dB of gain. Having many speakers with sensitivity from 80 to 100 dB can also be a solution and then just select the right set of speakers.
Right now I play using the kit VFET amp which has an Edcor transformer for the voltage gain at input. My source is a DAC where I use the digital vol control of it. So right now I waste not much energy in signal transfer. If amp had no Edcor (just a follower) I may be close to be able to play just using fixed output from DAC.....but I think it still will be to loud for most tracks. A "variable" Edcor could be a solution......
I am building a source follower amp in the moment and interesting to test if using fixed output could be a solution for some low volumen recorded tracks.
I am also waiting for a preamp kit called Iron Pre. It has a bit of it as the output gain can be set via a transformer to a few fixed values......but still has a vol pot.
If the output transformer had "enough" fixed values they could be used instead of vol pot. I know this solution exists in another pre but I can't remember the details.
To just have the gain needed will also give the best SNR.
my entire semi-rant was nothing else than - whatever idea you have, try and compare with already tried
pretty much all (clever and some non clever) solutions are already made and tried, with various areas of implementation
what would suit you most, regarding sound and greensize, that's up to you
anyway, when you opt for some specific approach, best to chat about in dedicated thread
not much benefits of general discussions, except getting general info
@zm... I agree. And that is why I finished with the comment: " I figure my assumption is flawed..."
The gain structure article by Pano appears well founded and the arguments definitive and yet voltage dividers followed by excessive gain appear almost everywhere.
The choice of capacitors, wire etc. result in differences that in some cases may be vanishingly small, yet there are those who are evangelical about these decisions. One might think that the same people might worry about these SNR arguments.
I was really looking for `general info' and hoping that someone would point out the pragmatism that is the basis for " Voltage dividers everywhere". It might be as simple as a question of convenience versus a minor degradation in signal which is inaudible (but then again some people still worry about `wire'...).
The gain structure article by Pano appears well founded and the arguments definitive and yet voltage dividers followed by excessive gain appear almost everywhere.
The choice of capacitors, wire etc. result in differences that in some cases may be vanishingly small, yet there are those who are evangelical about these decisions. One might think that the same people might worry about these SNR arguments.
I was really looking for `general info' and hoping that someone would point out the pragmatism that is the basis for " Voltage dividers everywhere". It might be as simple as a question of convenience versus a minor degradation in signal which is inaudible (but then again some people still worry about `wire'...).
I like the idea, but have never seen a real life implementation of it.
Can somebody tell what the downside is of this?