single-gang potentiometer for Dry / Wet balance control?

[waspclothes]

i seem to be having a brain fart today.. it seems like this would be easy, but i don't know how i can use a single-gang potentiometer to control the balance between 100% dry and 100% wet. i'd like the perceived volume to remain constant at any setting ideally. i know i can do it with a dual-gang potentiometer where one pot is audio taper and the other is reverse audio taper.. but i can't find pots like that anywhere! so is there any other way? i have no problem adding extra circuitry to the schematic, more op-amps, etc.

thanks guys!

[R.G.]

i have no problem adding extra circuitry to the schematic, more op-amps, etc.

See: Panning For Fun at Geofex.

A constant level pan-pot is better done this way than with actual pots. It's from a National Semiconductor audio applications book. The circuit expects a low source impedance driving it on both inputs, and inverts phase, so you may need buffers before the two inputs and an inverting opamp stage after it. Two dual opamps make this work even for high source impedances, producing an in-phase result and a constant relative volume pan.

[waspclothes]

See: Panning For Fun at Geofex.

A constant level pan-pot is better done this way than with actual pots. It's from a National Semiconductor audio applications book. The circuit expects a low source impedance driving it on both inputs, and inverts phase, so you may need buffers before the two inputs and an inverting opamp stage after it. Two dual opamps make this work even for high source impedances, producing an in-phase result and a constant relative volume pan.

Thank you R.G., very nice document!

My only problem is that the amplitude doesn't seem to remain constant with the solution and resistor values in the PDF going into the inverting amplifier.

Assuming A and B are equal, wouldn't that mean the signal is .4 at the middle point and only .29 at the end points.. so an overall higher amplitude (like 140%) of the combined signals at the mid point. Simulating it quickly shows the same thing.

Would something like this be appropriate:



The first op-amp sections on the two inputs are already present in my circuit for scaling. This is just a 100K pot with the output of each op-amp at an end and the wiper going into a buffer follower. You can ignore the section after (volume control and final output buffer). It seems to work in simulation and the actual output amplitude for two equal signals seems to remain pretty constant, although at the midpoint the amplitude dips but only to 90% or so.

[R.G.]

As I mentioned, that circuit was lifted from a National Semi application note, and back when I messed with it, it seemed to be a constant volume setup. Fortunately, I still have that book in paper. I just opened it up...

And I think it's OK. I removed some very specific advice about the feedback resistors on the two opamps, as I thought that an overall gain of more than unity at the extremes would be useful.

The National app says you want unity at the extremes of the pot and -3db in the middle. I'll go see if I have my original sims on it.

I've used that before, and it seems to work fine to my ear. I'll be back.

[R.G.]

I ran the sim again. It acts as expected. It's always possible I mucked up the typing of the math.

Nation wanted you to use 15K for the two series resistors, a 10K pot in the middle, and 51K feedback resistors. That give unity gain (inverted, of course) at either end of the pot travel, and 0.7071 for each signal when the pot is in the middle, which is what they were trying to do. (Hmmm. I notice I did put the NatSemi values in there.)

I did a single input, then mixed the two outputs in a unity gain inverting mixer. I get unity gain out of the mixer at either pot extreme, and 2x at 50% of the pot travel, which is +3db power and +6db voltage, as intended. This is correct for a 100% correlated input, I believe. You're panning between two signals, and get 1X at each end, and the sum of the two in the middle.

I think that the circuit as shown is a constant-power (and that ought to mean loudness) pan.

Of course, computers allow me to make mistakes faster and larger, too.

[sarv]

Bringing this thread back from the dead

So I'm using this design, just slightly modified since I've got a single-supply configuration. So I've put the virtual ground on the non-inverting op amp input and AC coupled the pot wiper to GND.



Both the 'wet' and 'dry' inputs are coming from other op amps so are low impedance.

The circuit works exactly as it should other than a couple issues:

a) when I have a very loud input, I get a kind of 'scratching' at the extremity of the pot, as it's turned from its max up to about the first 5% of the turn or so
b) the other signal isn't completely killed at either extreme, it comes through albeit very faintly

Both of these issues seem to diminish proportionally as the pot value increases (scaling up the resistors along with it to keep all the ratios the same). So you're probably wondering why I'm using a 1K pot in the schematic then... well this is an 11th hour fix for a project that's meant to be finished already, and unfortunately the pot and mixer circuit are located on the PCB next to a very noisy part of the circuit as far as EMI goes. If I go any higher than a 1K pot I'm getting a clock signal bleeding into my audio output. Barring a complete rerouting of the board, which at this point is basically out of the question, it seems I'm stuck using the 1K pot.

I found that by increasing the value of the capacitor on the pot wiper I can at least diminish issue (b), but it doesn't seem to make much difference for the scratching. Also it seems to only diminish issue (b) up to a certain point, i.e. 1000uF isn't much different than 470uF.

I suspected that the scratching may have been related to some small amount of DC on the pot and so tried using a 'plastic conductive' pot instead of the carbon types that I would normally use (I've read that they can be more forgiving if you have some DC present), but I found it made no difference.

I also tried AC coupling pretty much everywhere possible, a cap in series with each of the four 1.5k resistors. No difference.

I'm thinking about maybe trying a different op-amp (currently using a TL072), but not sure which would be the crucial parameter here if this solution is worth investigating?

Or of course open to any other suggestions anyone might have.

Thanks in advance for your help!

[R.G.]

I wish I had a magic bullet for you, but I can only offer some "where to look" advice.

First, attenuation to full off is heavily dependent on the equivalent series resistance of the pot wiper. We have this idea that a pot wiper has zero resistance to the resistor surface it touches, but it's not really that way. Pots behave as through there is a small series resistance in series with the wiper, inside where you can't get at it. This is heavily dependent on initial construction and material quality, as well as contamination or lack thereof. The best this circuit can do is to attenuate to the pot resistance to "ground" divided by the input resistor. Obviously, the lower the pot value, the worse this gets, as you've found out.

In your circuit, you also have the ESR of the wiper capacitor to deal with, again exacerbated by the low values of resistance. If it were me, I'd get rid of that cap and tie the pot wiper to the bias voltage for the non-inverting input and bypass that point to "real" ground like crazy. There's still an issue with current through the path to real ground, but now both the wiper contact and the noninverting input experience it the same.

The "distortion" is suspicious. It may well be that it's your opamp doing something funny when it's pushed to the edges of its power supply, and when it has to source a lot of current back into the inverting input node to cancel the incoming input current. You're down in the few-k region, and the TL0-72 is not all that happy down there. It may be a burst of oscillatory grack near the power supply limits too. The TL072 can only pull within about 1.5-2V of either power supply, so you may be pushing it into momentary limiting or even a burst of RF. Oscilloscopes are good for looking at things like this. If there is distortion on the output of the opamp but not at the summing point, the opamp is immediately suspect.

If you have any options or time, try something like the LM833 or NE5532. Both are well regarded, and can work at lower impedance levels than the TL072. Both will, alas, have their own quirks. Can you run this thing from a higher power supply voltage? All of these opamps can run happily from a total of 30-36V, and that might move the limits up to where they're not interfering with signal - if it's some form of running out of power supply that hits you. Again, get a scope on it and see what's really happening.

[sarv]

OK so first of all thanks for the recommendation of connecting the wiper to the bias voltage, that got it a little quieter. I actually kept the 470uF on the bias voltage as it helped to keep my clock bleed into the audio signal at a minimum (compared to a 10uF, for example, it gave much better results).

As for the 'scratching', it actually turned out that when I got everything soldered onto the PCB I no longer had this issue. So I'm guessing my wires going to the breadboard (I had the mixer circuit only on the breadboard, the rest of the circuit on a PCB) must have been picking up some noise?

I still get a noticeable amount of wet signal coming through with the pot fully turned dry, but I think this issue actually runs deeper than I had initially thought. So this is a guitar effects pedal and I've got a true bypass wired 3PDT switch (with effect input grounded when bypassed). This is an analog delay, however, so when the feedback is turned all the way up is when it produces the loudest output, and it's not necessary to have any input to get it. So even when I've bypassed the effect circuit, when I'm at max feedback I still get some of the effect output leaking through. I'd say it's between 1/5th and 1/4th of the amplitude of what's leaking through with the effect on but turned fully dry. I can only imagine it's coming through GND? Any ideas on how my GND might be coupling so much noise and/or how to get rid of it?