loading

Smart amplifier design building blocks help? (making an amp w/ auto-volume adjust by controlling gain with running avg)

I am currently working on a simple headphone audio amp that automatically adjusts volume to my preference, regardless of what video I click on when it is plugged into a computer. I am tired of some video's being so quiet that I have to crank up the volume to overdrive to hear it, while other video's blast out sound and burst my eardrums. And of course when watching photonicinduction videos, he often can be hard to hear his voice at really low volume at like 12 at night when everyone is asleep, but then the sparks and hammers come into the picture, it is often too late to crank the volume back down, and thus makes your ears bleed or waking everyone up in the neighborhood! X-(. (Luckily he has included sound warnings, but I am lazy, and will engineer things that do the hard work of adjusting volume for me! :)   )

In a way, I understand that this will minimize the dynamic range, which I suppose hardcore "audiophiles" will go NUTS over me actually WANTING to do that, but none the less, That is what I want.

My design currently is just a simple emitter follower with a NPN and PNP transistor is the class AB operation, and to minimize crossover distortion and give a definite gain, I have negative feedback op amp from the output of the totem pole transistor arrangement back to the inputs that feed the base of the low and high side transistor. (obviously, the transistors by themselves are low impedance output, (CV mode) and high impedance input, and have a gain of like 1.) Sorry, no schematics, but you capable of googling it if you want to see what I am describing.

As it is, it works fine. HOWEVER, it does not do the automatic volume, or gain control. That is set in stone by a resistor divider network between the feedback and ground (the basic non inverting amp feedback config.) I do not want a fixed gain. I want a electrically adjustable gain. The thing I plan to use to do the automatic adjustment would be a peak detector on the input, which is probably going to be a simple diode and capacitor with a buffer, and a discharge resistor (or constant current drain) across the capacitor to gradually lower the max output, or possibly an integrating op amp that effectively takes a running average. The output of that needs to control GAIN of the primary amplifier, and I want to be able to control the PID of that thing, maybe by tuning the a LCR circuit, or what have you. It is all nice and analog, just the way I like it! :)

If ONLY there was some magical (and common) component that I could control the resistance anywhere in the feedback loop, of with a given voltage w/ respect to ground... I was thinking of like a MOSFET, but I would have to deal with the non linearity of the input vs output impedance, which probably changes with respect to everything from the current flowing through the channel, the voltage relative to ground at the source, the temperature, the alignment of the planet's, what mood god is in, etc etc etc. That is too many factors for an open loop control on to model in discrete stuff! And digital potentiometers are out of the question since, well they are digital, I want an elegant simple solution, and the are $$ and difficult to deal with. They have so many requirements that need to be addressed. 

I prefer a elegant discrete transistor circuit, but op amps and other basic building blocks are acceptable, (this thing I want to be built over a weekend with with jellybean parts from by junk bin.) if you happen to know of a specific audio chip that does this for me, please mention it, but I will probably not use it, but maybe analyze the internal circuitry and learn how it works.

Found it.

Demo of the circuit here:

-max- (author)  mpilchfamily2 years ago

Interesting, I cannot believe I never saw that video on his channel, I am not a big fan of using a bunch of germanium diodes though, I do not have them in my junk bin. :( Such a beautifully simple solution though!

I may even consider adding a bandpass filter on the input to the peak detector section so that it responds to the most audible frequencies better than one's that we can barely hear anyway. Though that may get too hairy too quickly. I'll see.