Idea for a clean blend circuit - feedback wanted

[aion]

I had an idea for an effect and really quickly threw together a schematic this afternoon. I was hoping for some feedback. It's pretty similar to the ROG Split/Blend in concept (though not the actual circuit topology) but the idea is that it's kind of a utility effect that lets you blend a clean signal in with whatever other effect is running in parallel. This unit would have a master bypass that would also bypass the parallel effect when this unit is off.

My major goal here was to give each channel its own mix so you can dial in an exact perceived unity gain. This was as opposed to most other split/blenders or parallel mixers that use a ratio-type blend control (e.g. pan between the two effects) where you don't really have control over the final volume.

The clean side is a full-featured boost with Baxandall treble and bass controls, for instances where you want to EQ the clean side either to remove bass content, add treble, or whatever, more than just a change in volume. There is also a phase inverter (thanks R.G.!).

The major application I had in mind was for bass guitar, where they love to blend the clean signal in with an overdrive/fuzz or even a compressor, if the parallel effect is lacking its own clean-blend control. But it works equally well for guitar, and the flexibility could result in some interesting combinations.

Here it is:



My questions:

- Is it fine to leave the return path hanging without a return buffer?

- Is there any practical reason I would want to use an active mixer stage at the end, or is it okay to just run the signals together like that?

It wouldn't necessarily play perfectly with any effect in the world in the parallel path, and there's no counteracting for a poorly designed effect. But the initial thought was that I'd be pretty safe in delegating to the parallel effect the task of providing a volume control and a low-impedance output.

[ElectricDruid]

Since you've got no mixer resistors at all (not even a simple passive mix) the two outputs are going to interact a lot, and you might well find you can't get a reasonable balance of one with the other.

Given that you've got a spare op-amp in your clean path (that second one isn't doing anything that couldn't be done elsewhere), why *not* include the active mixer at the output? It'd give you a more reliable result.

In fact, if it was me, I'd also drop the phase inverter stage, and I'd buffer the parallel effect return before I sent it to the mixer (since a mixer with a 1M input impedance is a bit much to ask for, but a buffer would make that easy). If you wanted phase inversion, you could switch the clean channel between being sent to the mixer's -ve input (traditional mixer style) or the +ve input (differential amp style).

HTH,
Tom

[GibsonGM]

Why 1M bias resistors?  I may be mistaken and maybe it's trivial, but I thought that was sort of 'starving' the opamp? "The smaller the resistor values, the smaller the errors" and all that.   More like 100k may be more appropriate for stability, and higher resistances don't result in saving appreciable current and they probably introduce thermal noise.  You're still in uA with 100k's...neat project!

[PRR]

> Why 1M bias resistors?

Why not? The TL072 won't mind 10Meg or more. These bias resistors are bypassed for audio by the signal source, so the *higher* they are the lower the hiss (though by a Meg or three diminishing returns are set-in).

[antonis]

What's puzzling me is subsonic (0.016Hz) cut-off frequency of C11/R15.. 

[aion]

In fact, if it was me, I'd also drop the phase inverter stage, and I'd buffer the parallel effect return before I sent it to the mixer (since a mixer with a 1M input impedance is a bit much to ask for, but a buffer would make that easy). If you wanted phase inversion, you could switch the clean channel between being sent to the mixer's -ve input (traditional mixer style) or the +ve input (differential amp style).

Hmm... how about this?



Is there a simpler way to do the switch, e.g. with only a SPDT, or would a DPDT be required here as shown?

[duck_arse]

I'm not best w/ opamps, but, I think if you take the volume pot cold to VB, you can drop C10. you've also then got bias for the (-) input, so R18 would move to SW1B. and you would need a DC isolation cap between "RETURN" and R14, even if all I've typed is wrong.

[marcelomd]

Love this. I'm a (clean) blend junkie and have been messing with it for a few months now, but never as an independent unit.

In an integrated blend control, the hardest part is deciding how much control you want over either side and over the sum. Deciding where to put tone controls too. Only on effect side? Effect+clean side? After the mixer? Independent level controls or mix+volume? It is easy to end with 10+ knobs (a la AMT preamps). The hardest part for me was matching volume levels. If any side comes in too hot, you end up clipping the mixer. Not what you usually want.

An independent unit makes those choices for you, gives a lot of control but the _impression_ that you have fewer knobs, because they are spread over two units.

I really like it =)

On my blends I use a lower noise opamp for the mixer+tone control, like the NJM2068. Cheap, silent, and comes in SIP package. I wanted to use TLC22{6|7}x all around but cannot find them here.

[aion]

All right, one more attempt at this. The thing I adjusted here was to simplify the inverter/mixer stage at the end (IC2B) based on reading R.G.'s polarity inverter article a fifth and sixth time.



Knowing that all of the inputs are independent of each other, I realized I could just copy that schematic exactly and then add the other signal going into the inverting pin which should be unaffected by the phase switch. So the phase switch would only impact the effect return, not the clean boost.

Can anyone tell me whether this concept would work and whether the inverter can in fact serve as a mixer as shown?

[ElectricDruid]

Can anyone tell me whether this concept would work and whether the inverter can in fact serve as a mixer as shown?

Yes, the inverter should be fine, and the great thing about op-amps is that you can use them in all their various configurations *at once*!
The normal presentation is "here's the non-inverting amp" and "here's the inverting mixer" and "here's the differential amp" etc etc. but in fact the circuit is just the same for all of those really - you're just leaving a few bits you don't need out. You can have signals in one input with one gain, and signals in the other with another, and differential signals, and mixed signals and so on.
My only query would be leaving IC1B doing a bit of gain rather than buffering the effects loop return. For me, that's a more important job. If you need a bit of gain for that dry path, you can easily tweak R13. So using a whole op-amp for it is redundant. Given that it is, what would it be better used for? To me, the thing that leaps out is buffering the effects return, especially when R14 and R12 are in parallel (when SW1 is closed both inputs are at VB) and the input impedance comes down to 100K or so.
But what you've got works (I would expect - I haven't breadboarded it of course) and I'm getting picky now.

T.

[aion]

My only query would be leaving IC1B doing a bit of gain rather than buffering the effects loop return. For me, that's a more important job. If you need a bit of gain for that dry path, you can easily tweak R13. So using a whole op-amp for it is redundant. Given that it is, what would it be better used for? To me, the thing that leaps out is buffering the effects return, especially when R14 and R12 are in parallel (when SW1 is closed both inputs are at VB) and the input impedance comes down to 100K or so.
But what you've got works (I would expect - I haven't breadboarded it of course) and I'm getting picky now.

T.

Would there be any negatives to having a gain increase after the volume control? I've always tried to follow a rule of no amplification after the volume control just for noise purposes - but maybe this is just me being picky myself!

Also, with the two inputs being in parallel, should I increase R12, R13 and R15 to 470k to get a better input impedance? (also possibly dropping R13 down for a gain increase as well, to free up IC1B) Or would it be good enough just to throw a return buffer in there?

[ElectricDruid]

If you leave it as it is, then yeah, perhaps boosting those resistors to 470k to improve the input impedance on the return is a good idea.
If you free up IC2B and use it as a buffer, it won't matter, since the mixer will be driven by an op-amp.

[duck_arse]

R13 is putting DC on that pot, will need a cap.

[Groovenut]

R13 is putting DC on that pot, will need a cap.

Or you could connect leg 1 of the pot to the bias supply and leave the cap out.

[duck_arse]

R13 is putting DC on that pot, will need a cap.

Or you could connect leg 1 of the pot to the bias supply and leave the cap out.

[which one is leg 1 again?]

[aion]

R13 is putting DC on that pot, will need a cap.

Or you could connect leg 1 of the pot to the bias supply and leave the cap out.

[which one is leg 1 again?]

The lug that is currently tied to ground. I think that's the same thing you were talking about a few posts up.

Would this mess with the bias of IC2B though, especially as the volume pot is turned, since it involves a variable resistance to 1/2V? Would it be better to move C10 in between the volume control and R13, and let the volume control ride on the bias of IC2A?

Seems overkill to put a cap both before and after the volume control so the volume can go to ground, but that would also be an option.

[duck_arse]

the tone section oppie out sits at VB, because you told it to. if the pot cold goes to VB as well, there will be no DC across it, (4V5 - 4V5 = 0V), so you wont need the cap to the pot, or from the wiper, because the following stage is also tied to VB. VB - R18 - R12 - R13 - pot - gnd is why you need a cap on the wiper.

^ as I understand them, will stand for correction. [also, I was being cross referencing arse with pin 1. I only ever call it CCW or cold.]

[aion]

the tone section oppie out sits at VB, because you told it to. if the pot cold goes to VB as well, there will be no DC across it, (4V5 - 4V5 = 0V), so you wont need the cap to the pot, or from the wiper, because the following stage is also tied to VB. VB - R18 - R12 - R13 - pot - gnd is why you need a cap on the wiper.

^ as I understand them, will stand for correction. [also, I was being cross referencing arse with pin 1. I only ever call it CCW or cold.]

By adding a variable resistance to VB from the volume control, would that affect the input impedance of IC2B (being connected to bias resistor R18 with no coupling capacitor, by way of R12 and R13)? Or is it too little to make a difference?

[ElectricDruid]

the tone section oppie out sits at VB, because you told it to. if the pot cold goes to VB as well, there will be no DC across it, (4V5 - 4V5 = 0V), so you wont need the cap to the pot, or from the wiper, because the following stage is also tied to VB. VB - R18 - R12 - R13 - pot - gnd is why you need a cap on the wiper.

^ as I understand them, will stand for correction. [also, I was being cross referencing arse with pin 1. I only ever call it CCW or cold.]

+1 agree. No caps are needed if you take the bottom of the pot to the Vbias.

Currently, I'm not seeing why it should affect the impedance. R12 and R13 can't pass signals from one to the other easily because that -ve input is a virtual ground. That's the whole point of the op-amp mixer circuit - unlike the passive mixer, the inputs *don't* interact. So I don't see that tweaking the pot affects the Return input impedance.

Tom

[Groovenut]

As far as AC signal is concerned, Vbias is the same as ground, so there should be no change in I/O impedance (or at least no more than was there before you attached pot pin 1 to Vbias). With pin 1 to Vbias you will have the same voltage on all pins of the pot (as well as all I/O pins on the opamp) and no current will flow through the pot (regardless of the post wiper position). My 2 cents anyway