Use a single-section potentiometer to control gain on two analog circuits

Thread Starter

ttshaw1

Joined Jan 24, 2016
14
I'm designing a headphone amp, and one thing I need is volume control. I'm shooting for excellent fidelity to the digital signal, so left-right channel matching is keeping me from using a two-section potentiometer. I'm also hoping to avoid needing to rely on resistance matching on silicon, so those audio amp ICs with gain controllable by one pot are out. What I'm hoping to do is come up with an analog circuit that will match the gain of two channels based on the position of one potentiometer. I found this article on how to accomplish what I'm after, but the schematic has 404ed and I'm having a hard time understanding what's going on from the text only. My thought process is this:
  1. Need to use the wiper as common ground to get two linked resistors
  2. Common ground requires that each of the resistors be used as the resistor from inverting input to ground of a non-inverting amplifier, as no other configuration makes sense
  3. Turning the pot increases the gain of one non-inverting amp (A) and decreases the gain of the other (B). Use a differential amplifier to do something like subtract the signal from A from the maximum gain of the stage. For instance, maximum gain is 2, minimum is 1, with the potentiometer at halfway it's 1.5. Turning it a quarter turn gives 1.75 and 1.25, subtract the input signal from the 1.25 and subtract that from twice the input signal.
Here's a circuit I was kicking around. Unfortunately, it doesn't work.
upload_2019-7-27_23-35-56.png

The gains of the first stage don't change at the same rate with potentiometer turns, as you can see in this graph.
upload_2019-7-27_23-39-41.png
So I'm stumped. I don't see any way to do some analog work on them and wind up with their gains in lockstep. But it seems like it must be possible, as Woodward wrote an article on how to do it. Any thoughts?
 

crutschow

Joined Mar 14, 2008
34,280
The problem I see with such a scheme is that it gives a linear change in volume with pot position, but that gives a non-linear change is sound level to the ear since the ear is logarithmic in loudness perception.
That's why pots for volume control are audio taper (logarithmic) not linear.

Below is an illustration of the difference.
A linear taper will initially give a very large change in apparent loudness as you start from the zero position, and then the apparent change gets smaller and smaller.
For example the 50% to the 100% point is a doubling the audio power, but that's only a 3dB change in apparent loudness (it takes 10dB for an apparently doubling of the loudness).



I suggest you take a closer look at the gain controllable amps.
They may make some that stereo with matching change if volume levels.
 

Alec_t

Joined Sep 17, 2013
14,280
so those audio amp ICs with gain controllable by one pot are out.
Have you actually tried one? I would expect any slight mismatch in gain to be undetectable by the normal human ear, given its logarithmic response to audio levels (but then, I don't have golden ears :) ).
 

Thread Starter

ttshaw1

Joined Jan 24, 2016
14
The problem I see with such a scheme is that it gives a linear change in volume with pot position, but that gives a non-linear change is sound level to the ear since the ear is logarithmic in loudness perception.
That's why pots for volume control are audio taper (logarithmic) not linear.
You're right about that. I used the numbers in the OP as an example to be easy to understand; in actuality I'll be using a logarithmic pot. That said, the taper of the pot is immaterial. There will always be some resistance R on one section and (total resistance - R) on the other. The intention of identically controlling the gain of two independent signals with those resistors remains no matter what kind of pot I use.

Hello,

As I understand, you are looking for an audio pan circuit.
Have a look at this page:
https://www.analog.com/en/analog-dialogue/articles/ultralow-distortion-panpot-amplifier.html

Bertus
Nope, that looks like it's for changing the balance between left and right channels, so that turning the pot would make left louder and right quieter or something like that. What I want is to keep the balance identical and control the volume to both with one pot, so that turning the pot brings the gain for both channels from, say 2x to 10x.

Have you actually tried one? I would expect any slight mismatch in gain to be undetectable by the normal human ear, given its logarithmic response to audio levels (but then, I don't have golden ears :) ).
I agree that it probably won't be a practical issue, but I'm trying to design this amp to keep the output as faithful as possible to the input; ideally around 115dB to match my DAC.
 

ci139

Joined Jul 11, 2016
1,898
Last edited:

Thread Starter

ttshaw1

Joined Jan 24, 2016
14
I'll clarify what I'm really interested in. I want to recreate the circuit W. Stephen Woodward wrote about in this article, in which a single section potentiometer controls the gain of two channels. The schematic attached to the article is lost, but the description of the circuit is still there. Is anyone able to parse how he's using only op-amps, one potentiometer, and passive components to accomplish this?
 

Thread Starter

ttshaw1

Joined Jan 24, 2016
14
That seems like a good IC if I decide to go that route. I'm sure there are plenty that are pretty good. But like I said above, I want to recreate the circuit W. Stephen Woodward wrote about in this article, in which a single section potentiometer controls the gain of two channels. And specifically, controls the gain of two op-amp circuits with discrete components.
 

bertus

Joined Apr 5, 2008
22,270
Hello,

The link to the schematic is dead.
I already looked in the archive, but also there it can not be found.

Bertus
 

Thread Starter

ttshaw1

Joined Jan 24, 2016
14
Yep, that's exactly the problem I'm running into. I tried to recreate what the schematic looks like from his textual description, but it's difficult for me to understand. I want to see if anyone on here is able to figure it out, or come up with an alternate circuit that works using only
  • op-amps
  • one single-section potentiometer
  • discrete passives
 

AnalogKid

Joined Aug 1, 2013
10,986
I'm also hoping to avoid needing to rely on resistance matching on silicon, so those audio amp ICs with gain controllable by one pot are out.
No, they're not.

Even without laser trimming, resistance value *matching* and tracking on ICs is very accurate. The absolute value might vary 10% from one part to the next, but the matching can be under 1% without trimming. This is the core design concept behind the much adored/maligned 555. If you post some part numbers, we can review them.

I agree that a dual pot is the worst option (unless you can throw serious money at it). Does the control have to be a pot? Maxim et al have digital pots with excellent tracking, click-less changing, log taper, and simple up-down button control. Some come with non-volatile memory so they remember the last pot setting when powered up.

ak
 

Thread Starter

ttshaw1

Joined Jan 24, 2016
14
Alright, so I spent a few minutes looking through audio DAC datasheets and found 0.1dB channel mismatch for the TLV320AIC3104, which is around 1.1%. Unless that's particularly high, a PGA with 1% matching should work. Do you know if I'm likely to find a 24bit/192kHz DAC with better matching?

I seem to remember that PGAs have other undesirable characteristics like nonlinearity. I'm shooting for 115dB SNR so any non-ideal behavior could sink me. What do I need to keep an eye out for in terms of signal integrity with PGAs?
 

AnalogKid

Joined Aug 1, 2013
10,986
Slew rate as a function of gain. Based on your bandwidth requirements you can calculate the required slew rate, increase it for margin, and compare that to the datasheet worst case for all areas of the gain range.

ak
 

danadak

Joined Mar 10, 2018
4,057
At 1.1% seems like matching could be done w/o sophisticated techniques.

But if .1% I would start thinking about a control loop, maybe use a sub audible
signal that can be added to both channels and compared/measured for amplitude
and in turn control a simple JFET in a voltage divider to achieve closed loop
accuracy.


Regards, Dana.
 

ci139

Joined Jul 11, 2016
1,898
how to get level wide bandwidth curve for the j-Fet 192kHz
i haven't studied j-Fet amplifiers too much though
 
Last edited:

RPLaJeunesse

Joined Jul 29, 2018
252
Below is an idea I came up with. Top stages are gain of 0-0.5x & 2x for 0-1x combined range. Lower stages are 0.5x-0 (in that direction) followed by a +1/-2x. When the U4 stage runs at 0.5x the U3 stage cancels it out for zero output. As the U4 stage gain decreases the U3 stage output increases, reaching +1 as R9 is fully to the right. Note that for each stage resistor ratio matching is critical. Although it may simulate nicely (didn't try) it is an impractical circuit because you will almost never get two input resistors to exactly equal the +/- 5% resistance tolerance of a real pot. For something more realistic take a look at using a pair of factory trimmed THAT Semi 2180s: http://thatcorp.com/2180-series_Pre-Trimmed_Blackmer_IC_Voltage-Controlled_Amplifiers.shtml
Stereo_Vol_Ctrl.jpg
 
Top