Wet/Dry or Blend Implementation - How does it work?

[JFace]

I have a good grasp on electronics, but everytime I see this blend implementation I get puzzled 

The blend pot in this schematic: http://www.runoffgroove.com/splitter-blend.html

Wouldn't the two op amps work harder as the resistance is increased? Does 25K really make that much of a difference?

[Mark Hammer]

I have not built it, but I suspect that a 25k blend pot will effectively alter the balance, without necessarily eliminating either signal.

Look at the circuit, not from the perspective of the op-amps, but at the output.  Think of the portion of the blend pot between the wiper and each outside lug as one half of a voltage divider formed by that resistance, and the 1M fixed resistor to ground.

So, if the blend pot is rotated fully clockwise, there is 25k on one side of the wiper, and 0k on the other.  If I take the signal from the junction of a 25k resistance, and a 1M resistance, I am essentially looking at the outut of a 1025k "pot" turned down a bit.  In that respect, a 100k pot might provide greater contrast, although I have to say that the soundclips indicate more than enough contrast for most tastes.

[R.G.]

I have a good grasp on electronics, but everytime I see this blend implementation I get puzzled 
Wouldn't the two op amps work harder as the resistance is increased? Does 25K really make that much of a difference?

Principle of superposition.

Each opamp output has a very, very small output impedance, in the 10-20 ohm range while linear. If you have signal into one opamp, and no signal into the other, then the non-signal opamp may be considered as a very small resistor to ground, as the no-signal opamp is holding it at "zero volts" for AC purposes.

So for the signal-conducting opamp, the 25K is acting like a volume pot. When the wiper is closest to the signal conducting opamp, the wiper has nearly all the signal. When the wiper is closest to the non-signal opamp, the wiper is at "ground" and little or no signal goes out the wiper.

And now for the superposition. In a linear network, you may compute each excitation separately, then add them. So the opamps driving the blend pot from equal sides both act like low-impedance sources, and "ground" the opposite side, but drive their own signal. The signal at the wiper is then the sum of the two signals from each opamp, but modified by the wiper position.

[JFace]

Brilliant! That makes sense. I've seen this implementation enough to think that it should be explained somewhere. And now it is 

[Bill Mountain]

Here's an add on question.  Since these signals are buffered prior to the blend pot is the blending action comparable to mixing inputs at an inverting opamp stage?

[JFace]

They are similar but the blend implementation is more limited. As you increase the level of one, it consequently decreases the level of the other. With the active mixer inverting op amp, you can get any combination of levels from your inputs. For example, if you wanted 100% clean tone mixed with 50% overdrive, you would want to use the inverting op amp method.

[R.G.]

Here's an add on question.  Since these signals are buffered prior to the blend pot is the blending action comparable to mixing inputs at an inverting opamp stage?

Depends on what you mean by "comparable". This setup is a passive mixer with a buffer on each input, and a fixed mixing ratio dependent on the wiper position. It's also non-buffered on the output, so the following loading changes both the output level and the degree of mixing, and also the effective "taper" of each side.

Think about this: If the following load is big, maybe 1M, then the voltage out is just the voltage ratios you expect from the position of the wiper on the 25K pot. What if it's 1K? Now you get full volume from either opamp when the wiper is at its end, and this drops by half in the first 1/25th of pot rotation, and is at 1/12th at half-rotation, the other side doing the same from the opposite side. This becomes a kind of dead zone in the middle and sudden "on" at either end.

Simple blenders like this have loading issues and effective-taper issues as well as not always summing to unity gain through the thing depending on what it drives.

If you're doing anything critical in the least, you can add one more opamp and a few resistors and have a pretty decently-performing pan-pot. See "Panning for Fun" at http://www.geofex.com/Article_Folders/panner.pdf. The schemo in the rectangle at upper left as an addon/mod to the two opamps already used fix many of the issues with the passive "blend" control setup and make it load immune.