I am designing a small power amplifier similar to the circuit below, and I would like to have a volume control on the input. My understanding is that I can't just put a voltage divider on the input, since then the spectra of the filters on the input would change. I can't just have the pot change the gain of the amp continuously, since it has a minimum gain of 10. I figured I could just precede the input with a voltage divider and a unity-gain voltage follower, but I wasn't sure if there were some more elegant solutions or techniques available. Thanks!


Typically I'd go for a logarithmic potentiometer on the feedback path of the amplifier. The difficulty here is that the non-inverting amplifier has a minimum gain of 1 (not 10).

The output will influence the node between \$R_i\$ and \$C_i\$ until the corner frequency at around \$f = \frac{1}{2\pi(R_{f1}+R_i)C_i}\approx 16mHz\$. This is far below the audible frequency range! I assume it is only used to bias the output and input at the same DC voltage. For all practical considerations, above this frequency, \$C_i\$ can be considered an AC short-circuit.

The combination \$C_f, R_{f2}\$ works at a frequency of \$f \approx 160kHz\$ which is well above the audible frequency range. Above this frequency, the capacitor \$C_f\$ can be considered an AC short-circuit as well, giving you a gain of 10.

Since the audible range lies between these frequencies, the useful gain is more or less \$A \approx 1 + \frac{R_{f1}}{R_i} = 11\$. If you don't mind having a minimum gain of 1, you can consider using \$R_{f1}\$ as a potentiometer. If \$R_{f1}\$ is turned all the way to \$0\Omega\$, you will get unity gain.

Alternatives are either using or cascading an inverting amplifier where you use a potentiometer in the feedback branch. An inverting amplifier has a gain of \$A = -\frac{R_f}{R_{in}}\$ so cranking the potentiometer all the way down will completely mute the output.

| improve this answer | |
  • \$\begingroup\$ Thanks for the suggestion - my amplifier does indeed have a minimum gain of 10, not because of the noninverting configuration, but because the LM3886 is specified to have a minimum gain of 10 in the datasheet. \$\endgroup\$ – Billy Kalfus Apr 11 '18 at 14:27
  • \$\begingroup\$ I see, I was not aware of the stability issues. In that case I would add an inverting amplifier with a logarithmic potentiometer to attenuate the input signal. You might be able to get a bit more out of your opamp instead of just buffering (which is also an option). \$\endgroup\$ – Sven B Apr 11 '18 at 15:02
  • \$\begingroup\$ @SvenB Pardon me a newbie question - What exactly would go wrong if I used a potentiometer voltage divider just before the input capacitor (Cin)? From what I see there the frequency characteristics impact should be minimal. I am just trying to avoid another opamp in the path ... \$\endgroup\$ – rudolfdobias Apr 13 at 22:50
  • 1
    \$\begingroup\$ There is no "breaking" problem with attenuating at the input so if avoiding a second opamp is your priority then go for it. The disadvantage of doing so is a worse noise performance (so it may just be alright if noise is not your major concern). Also, the corner frequencies at the input are likely to change depending on your potentiometer resistance value. \$\endgroup\$ – Sven B Apr 15 at 5:37
  • \$\begingroup\$ @SvenB Thank you. In the end I tried to go with inverting preamp stage (A=0..5x) with the log pot in the feedback. It works well but I have another concerns: [1] The main volume pot will be on the front panel. I feel concerned wiring the feedback loop all the way through the chassis and back. Is this a good idea? Also when this line accidentally breaks, the amplification would spike to infinite. [2] Should I deal with band filters at the preamp stage (and in the fb loop), or the filtering at the final stage is "good enough"? Sorry for bothering you, don't know where else to go for advice. \$\endgroup\$ – rudolfdobias Apr 16 at 10:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.