Wet and dry signal mixer with level control...would this work?

[rockhorst]

I just finished my first build (Ruby, pictures soon) and was contemplating on doing an octaver (say a Green Ringer). I'd like to mix the wet and dry signals to give a fuller guitarsound when playing live with only one guitarist. I've seen some implementations off this (I think it was on GEO), but also read posts about people wondering if splitting the signal would result in a volume loss. To split the signal and prevent any effective signal loss I thought up the schematic below. Would that work? Bare in mind that I'm still a big noob, so posting this is scary as heel 

*edit: fixed a first mistake: the slider of the input pot should be on the guitar side, not on the wet/dry side*



Oh, read 'level' instead of 'gain' to avoid confusion of what is meant. Should there be an output cap after the opamp, or is that not really necessary?

[cd]

No, it won't work.

The guitar signal should go to a non-inverting op-amp stage (a buffer - see Adjusticator on GEOFEX).

From the input stage, the signal gets split - one path to your stomp box, the other to the mixer.

Now the mixer should be an INVERTING variable gain stage.  That thick vertical line should actually be two 10k or higher resistors that come together to the - input of the inverting gain stage.  You can make the gain variable by adjusting the feedback resistor. 

For headroom compliance, stick a couple of back to back diodes to ground on the output, with a 100k pot wired as a variable resistor between the output and the diodes.

That should be enough to get your started.

[rockhorst]

Thanks cd! I used a non-inverting op-amp cuz I wasn't sure if that phase shift of pi radians you get from an inverting opamp would matter. Could you draw me that diode part? I'm not too sure on what you mean exactly. Good to know that the idea itself isn't completely bogus.

[R.G.]

Take a look at the Adjusticator, including the new adaptation for a mixer.

This is at GEO, http://www.geofex.com under "Circuit Sweepings".

The input circuit, especially the high impedance version, makes fine guitar buffer with variable gain. This can then drive both an effect and a secondary volume control for the dry signal. The mixer will then mix them back together, provide a master gain, and allow a nice, low output impedance.

[Gladmarr]

why would you necessarily need an inverting opamp stage?  It seems to me that inverting or non-inverting could be used in any situation.  

[cd]

why would you necessarily need an inverting opamp stage?  It seems to me that inverting or non-inverting could be used in any situation.   

It's just easier to do it with an inverting stage, less parts.

Diode info: imagine a diode clipper to ground, like in the MXR Distortion+.  The diodes tie together to the audio at one point, and to ground at another.  We lift the point that connects to the audio, and insert a resistor in between, connecting the former audio point to the point where the two diodes tie together.  So instead of having two diodes to ground, you have a resistor then two diodes to ground.  Turn the resistor into a variable one (rheostat) and you have a variable clipping threshold.

[rockhorst]

Ah, thanks RG! that looks a lot like what I had in mind. Still don't fully get the diode part, but I'll think about it. Drawing would still be helpfull. Think I'm gonna play with this mixer idea over the weekend.

And about that output cap+resistor: should there only be one of those at the end of the mixer, or also still in the stompbox circuit? If it even matters, that is...

[R.G.]

why would you necessarily need an inverting opamp stage?  It seems to me that inverting or non-inverting could be used in any situation.

Almost any situation. There are slight differences, and these are biggest in a virtual-ground mixer.

In a mixer, you want the input channels to not interact - changing the volume on one stage does not change the volume on the others, even slightly. The modern audio world has settled onto the virtual ground mixer to do this.

That is an inverting opamp stage where the inputs to the (-) input are currents, not voltages. The (-) input is held by feedback to the same potential as the (+) input, and in most cases that is ground, or a virtual ground like a bias voltage. This has the advantage that the input point is always 0V, so the currents going into the (-) node show up as voltage at the output , but not as voltages at the (-) point, so they cannot interact there. One input does not affect the others.

The way this is almost always done is with a volume control pot at the input, and then a fixed value resistor to the (-) mixing node.

There are other ways to build a mixer, but this has proven universally simple and effective.

Given that the mixer is inverting, you either need to provide an inverting stage after or before the mixer. I chose to do it before, to keep the number of parts down.

And actually, even more esoterically, it is always better from the op-amp's point of view to use an inverting opamp stage. In fact, the op-amp design world has a rule of thumb "Always invert - unless you just can't." The feedback stability is better, the chances to have problems are less. We violate that rule of thumb a lot, but we are getting away with it, not invalidating the rule.

In effects, unless you know why it does not matter, the overall phase through the effect should be non-inverting. That is to prevent bass loss by cancellation if the dry and effect signals ever get mixed. You can do that by only using non-inverting stages, or an even number of inverting stages. It takes some discipline to get it right, but it's a good practice to get into.

[rockhorst]

So if I understand you correctly, RG, I'd have to put an inverting opamp (say with unity gain) before splitting the input signal, so that on the output of the end mixer I get an uninverted opamp? It's something I had considered, but then thought 'why not use an non inverting sum-configuration'...but that just isn't good practise as you just said...and those caps? 

[zpyder]

R.G.-

looking at geo's Adjusticator and reading your last post, What is Vbias?  I see on the High-Impedance Adjusticator a lead coming from +9v, through a 10k res and then to "Vbias"... what is that?

zpyder

[rockhorst]

I'll take the risk of getting totally flamed over this, but I took the parts of advice and documentation that I understand and drew a second schematic. It should at least be better than the first proposal. I'm trying to understand everything that's going on in RG's Adjusticator, but I just don't understand most of it, so I stuck to opamp basics. Would it be a good idea to convert the stompbox input to a low input impedance instead of the usual high input, so that with the pot half way you roughly mix wet and dry 50/50? Would a LM 386 opamp do?



Let me apologize for any stupidity up front 

[B Tremblay]

Take a look at the Splitter-Blend.  It may meet your needs.

[rockhorst]

Thanks Tremblay, that's an interesting link I hadn't stumbled across yet. Can somebody explain to me why the green and red sections are different? (one has a FET and some extra resistors, the other one doesn't. Is that purely because of the phase switch?).

[B Tremblay]

Can somebody explain to me why the green and red sections are different? (one has a FET and some extra resistors, the other one doesn't. Is that purely because of the phase switch?).

Yes, the FET stage was added to provide a high-impedance input for the phase polarity inverter stage.

[R.G.]

... or you could use the last section of "Panning for Fun" at GEO.

I just updated it with a high input impedance stage that invalidates the need for an extra high impedance buffer.

[B Tremblay]

R.G., could you please e-mail me?  The address is at bottom of all 'Groove pages.  Thanks.

[zpyder]

I know it's retardedly simple... but could someone please clear up something for me...?

When see leads (as in the Adjusticator) that go to nothing but a label: Vbias, or Vref .... what is that?

Is that just a junction point that is labeled rather than traced all the way around/across to the other point(s)??

muchos gracias,
zpyder

[R.G.]

When see leads (as in the Adjusticator) that go to nothing but a label: Vbias, or Vref .... what is that?

It's a form of shorthand.

To keep schematics from getting all covered up with lines that go everywhere, it is common to name certain nets. For instance, in stompboxes, there is usually a net named "+9V". All of the places where you would ordinarily draw a line to the battery clip, you can just label "+9V" and know that when you make the circuit in real copper, all of those points are connected with a wire. So the things that would otherwise go everywhere, like power and ground, can be left off by labeling. This leaves the schematic uncluttered for the signal lines which are less boring and more important to understanding the circuit. We have a special symbol for the reference voltage - we use the triangular ground symbol, meaning "connect all of the places with this symbol attached together and that is your ground reference."

Commonly, the power supply (or supplies) are named nets, ground(s) are named nets. So are things like the reference and bias voltages which may be supplied to several places and would clutter up the schematic. So "Vref", "Vr", "Vbias" and the like are stuck in there to show these nets.

In formal design automation every net has a name, even if only a machine-generated one.

A "net" is the set of circuit points which are connected by one unbroken copper trace or wire. The more nodes (i.e. circuit points, like the end of a resistor or the collector of a transistor) that connect in a net, the more likely it is that nothing happens on that net - it's likely to be a power supply or ground net. The fewer nodes in a net, the more likely that you need to know something special about that net, so the more it needs to be shown on a schematic as a wire. Two-node nets are IMPORTANT. Fifty node nets may be important, but they're usually not interesting.