0
\$\begingroup\$

I'll try to be specific:

I have an audio application that has 3 inputs and 1 output. The classic answer would be to use a DP3T to switch between the inputs. However, I would like to have a continuously variable output, one that blends as you sweep a potentiometer.

Say for instance if the pot was fully clockwise it would give you the signal X, at the midway point, the signal Y, and counterclockwise the signal Z, but between noon and clockwise you would also get a blend of XY. X and Z would never blend, just meet with Y towards the middle of the pot throw. Is this something that can be done with one pot and VCAs, or a one pot mixer? I'd like to stay away from multiple gang pots, unless that is the only answer because my application requires vertical PCB mount and conformity of potentiometers.

Thanks!

\$\endgroup\$
3
  • \$\begingroup\$ Look up center tapped potentiometer. Might take a couple of buffers, but it could be done. \$\endgroup\$
    – Marla
    Dec 27, 2015 at 23:41
  • \$\begingroup\$ I think a center tap pot might be what I am looking for! I had only come across this schematic, which appeared to be what I was looking for, but I thought it must be wrong, as I was not familiar with center tap (4 pin) pots! [i.imgur.com/YbU6jyT.png] \$\endgroup\$
    – kraft
    Dec 28, 2015 at 2:00
  • \$\begingroup\$ Tapped pots used to be common during vacuum tube era. Getting very rare now. \$\endgroup\$
    – Marla
    Dec 28, 2015 at 2:04

5 Answers 5

0
\$\begingroup\$

The simplest method I know of is to use a center tapped potentiometer. Although center tapped potentiometers are very rare now days.

enter image description here

\$\endgroup\$
4
  • \$\begingroup\$ With this circuit, you might have problems if the inputs aren't perfectly matched with respect to gain. \$\endgroup\$ Dec 28, 2015 at 5:23
  • \$\begingroup\$ @philbrooksjazz: That's going to be a problem with any cross-fade system and would be addressed by fitting a gain adjustment on the individual signals. \$\endgroup\$
    – Transistor
    Dec 28, 2015 at 11:10
  • \$\begingroup\$ @Marla: You should probably add that you need low source impedance on each of the three input signals - e.g., straight from an op-amp's output. Otherwise you've got a resistive mixer and you'll never shut off any of the signals completely. For simplicity, brilliant! \$\endgroup\$
    – Transistor
    Dec 28, 2015 at 11:13
  • \$\begingroup\$ good comments on the center tapped potentiometer circuit. The strong points and weak points of this method are revealed. In my original comment (not this answer), I mentioned that buffers would likely be needed. I wasn't trying to supply the OP with a complete design. Just the sight of a center tapped potentiometer to inspire further investigating by OP. \$\endgroup\$
    – Marla
    Dec 28, 2015 at 12:09
2
\$\begingroup\$

This is certainly possible, and is fairly straightforward, but not terribly simple. Let's assume your pots are in the 10k to 100k range, and you have VCAs with a 0 to 5 volt control input. Then

schematic

simulate this circuit – Schematic created using CircuitLab will do what you want with a single pot. Note that TL081s are shown here, and will work fine, but TL084 quads will also do, and only 2 packages are needed. If you use something like this, the op amps need to be driven with +/- 9 to +/- 15 volts.

\$\endgroup\$
1
\$\begingroup\$

It sounds like you mean that you will hook up you 3 signals to the 3 inputs of the pot, and take the output between one of those connections and ground. No that won't work - if you draw out the schematic you'll see what I mean. In the full clockwise position you have the pot's full resistance between signals X and Z, so you'll probably get some odd combination of all 3 signals depending on where you tap the output.

Honestly I have never seen an application like the one you describe, most mixer designs like this use an active buffer, like a simple op-amp (741 or 301) with a pot on the feedback resistor. the outputs are tied together.

\$\endgroup\$
1
\$\begingroup\$

if the pot was fully clockwise it would give you the signal X, at the midway point, the signal Y, and counterclockwise the signal Z

Yes this can be done and what springs to mind is a mixer where the three summing resistors were somehow varied to give the mix you want. But they can't be simple summing resistors like this: -

enter image description here

You need to have electronically controlled resistors; consider what happens when you use an analogue switch that is turned on and off at a high frequency with variable duty cycle. Basically, if you control the duty cycle of an analogue switch the audio signal you put in at one end gets multiplied by the average of the duty cycle and, providing you have a reasonable (but not excessive) low-pass filter to "clean-up" the "chopped" audio signal, you get a variable resistance ranging from 0 ohms to infinite ohms.

What next springs to mind is the Linear Technology LTC6993 pulse width modulator. It can take a linear voltage demand and produce a pulse-width modulated signal like so: -

enter image description here enter image description here

So now you can chop up all three audio signals with a linear control voltage using the above chip and three-quarters of a quad analogue switch. You'll also need a bit of filtering on the op-amp summing amplifier (a capacitor across R1). The switching can be at 1MHz and this is 50 times greater than audio so the filtering is probably just a capacitor.

To finish the design you need to take the potentiometer and connect it across a stable voltage so that the pot wiper output is (say) 0V to 5V and, between 0V and 2.5V you mix X and Y with Y being dominant at 2.5V. Between 2.5V and 5V, Y loses dominance to Z so that at 5V on the wiper Z rules the roost. You'll probably need a comparator here to differentiate between upper and lower parts of the pot control.

The LTC6992 PWM device you need to use is in the following table: -

enter image description here

I'd choose the 0% to 100% device because at extremes it fully switches on or fully switches off the analogue signal it seeks to control.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you Andy! This is a really interesting idea to chop the audio and scan between the signals. If I cannot find a center-tapped pot that works, I will definitely try this! \$\endgroup\$
    – kraft
    Dec 28, 2015 at 2:11
1
\$\begingroup\$

It's possible but probably most easily with some digital magic. The basic element of your system will be a voltage controlled amplifier (VCA). You would need six of these to control your three stereo channels (or three for mono).

schematic

simulate this circuit – Schematic created using CircuitLab

VCA based on AN-6603 application note.

Next you'll have to monitor the pot with a micro and use the signal to give three analog outputs to each of the AGC pairs.

schematic

simulate this circuit

The code for the micro would have to generate the pulse-width modulation (PWM) output signals as follows. (Pseudo code assumes analog in goes from 0 to 100%, 100% being fully clockwise.

//                     Output at     CCW    Midway    CW
PWM1 = (50 - AIN) * 2;           //  100      0      -100
if (PWM1 < 0) PWM1 = 0;          //  100      0         0
PWM2 = 100 - ABS(AIN - 50) * 2;  //    0    100         0
PWM3 = (AIN - 50) * 2;           // -100      0       100
if (PWM3 < 0) PWM3 = 0;          //    0      0       100

The PWM outputs are passed through low-pass filters to give a smooth DC control voltage to the VCA cells.

I have absolutely no idea how much distortion or noise the VCAs would introduce and I haven't worked out weather the VCA would give the correct logarithmic response to simulate a log-law potentiometer. Any problems with the log-law could be sorted out in the software with a little work.

All analog solution

It's probably possible to create the VCA control signals from the potentiometer using analog electronics only.

schematic

simulate this circuit

The potentiometer wiper goes from -5 to +5 V and is (optionally) buffered by OA1.

OA2 and D1 form a basic precision rectifier which will pass any pot signal from mid to +5 end.

0A3 inverts the pot signal and feeds it to the second precision rectifier built around OA4 which will pass any signal from mid to -5 end.

Finally OA5 is an inverting summing amplifer. With the pot at mid position the voltage on the top end of R6 will drive OA5 output to +5 V. If either of the rectifier outputs start to rise OA5 output will fall. This satisfies the requirements for the midway channel.

Now tell us what on earth you are doing!

\$\endgroup\$
2
  • \$\begingroup\$ We certainly have similar ideas, don't we? Note that the second diode in my precision rectifiers will improve frequency response by keeping the op amps out of saturation when reverse biased. Not important in this application, but a good general practice. \$\endgroup\$ Dec 28, 2015 at 4:05
  • \$\begingroup\$ The saturation problem is explained in the basic precision rectifier article. As you've said, it doesn't matter in this application and the 'basic' reduced clutter on the schematic. Yes, I've seen you approach several problems the way I would on several other posts. :^) \$\endgroup\$
    – Transistor
    Dec 28, 2015 at 11:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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