TUTORIAL Why Your Clean Blends Suck ... or "How to (Clean) Blend"

[phi1]

I don’t think the originally posted circuit would work. The 470R needs to be bigger to isolate the two signals so that muting the clean doesn’t nearly mute the dirty. Bringing that up to 10k might work well enough without changing anything else.

Another way to do additive blend is to exactly copy the op amp panning (#2 in the original post). But, don’t connect the wet signal to the blend pot. I’ve done this quite a bit with good results, for example using fv-1 delay or reverb. It can help to add a separate 10k resistor from the dirty side to ground (otherwise the dirty/wet signal can be really loud). But, this is case by case.

 

[phi1]

you can also change the buffer on the clean side into a boost so that the max clean is louder.

Add resistor between the op-amp “-“ & “out” pins. More resistance is more volume here. Include a 100p or so capacitor too for good measure. And a resistor and cap from the “-“ to vref. Maybe like this.

 

[thunderaxe]

i definitely want the clean signal to be boosted at some point, so that it can compete with the dirty signal, especially since it will be pre-volume control. that would be the reason for having a smaller resistor for the clean side of the mixer, but you're saying that would cause the dirty signal to take that path to ground?

so if that isn't a solution then it definitely seems like using the clean buffer as a boost is the best solution to that problem. the reason i haven't done it yet is that i was still hoping to find a way to use the unity-gain buffered clean signal for an optional buffered bypass, so that you can switch between true bypass or buffered. but that's increasingly looking more complicated than i suspected so i might have to abandon it.

that also might open up another solution i was considering, which was just putting a jazz bass style volume control on the clean signal. so instead of the wiper of the clean pot being the output, and panning between the clean signal and ground; instead the wiper is tied to the clean signal, and pans between the output (to the input of the op-amp summing amplifier) and ground, like so:

 

[phi1]

Yeah the 470R was just dumping the dirty mostly Vref with the blend knob at min.

Think of the typical volume knob as a voltage divider. This would be like having a volume pot at almost minimum (10k at the top, 470R at the bottom). According to simple voltage divider calc (ratio of bottom R / total R), only 4.5% of the signal would get through.

The last idea may solve that since there’s always resistance from the dirty to VRef. If you use GND like your sketch, you’ll need another capacitor after since you wouldn’t be operating of the VRef bias anymore. So VRef would be better.

I’m still skeptical that having only a very small R isolating the signals (at max blend) may not be the best idea.

If you look up the simplest way to mix two input signals, it’s to give them each their own resistor before joining. So that resistor isolation is always the first requirement. I can’t say for sure what resistance is adequate (470R ok?), so I suggested following the pedalpcb example as a guide.

 

[thunderaxe]

okay, up to this point i haven't really understood why to use VREF as the ground rather than 0V, and to be honest i'm still not entirely clear on it.

in my circuit the clean signal gets a non-inverting op amp to maintain the same phase as the dirty signal, which means it gets a DC bias injected directly into it across a 470K resistor before the op-amp, and which is cancelled out by a 10uF capacitor (trying to retain as much low-end in the clean signal as possible in case it's being used as a bass pedal) before the volume/mix control. so at that point it should be an un-biased bipolar signal. is it because of the bias of the subsequent summing amplifier stage that the ground for the clean volume/mix control should be VREF?

 

[thunderaxe]

also, if i'm giving up on my idea of using the buffered clean signal path for a buffered bypass option, i can use the clean buffer op amp as a boost and have equal value resistors on both the clean and dirty sides for the summing amplifier. if i'm using the jazz bass style volume control for the clean signal, any idea what value that pot should be?

 

[phi1]

okay, up to this point i haven't really understood why to use VREF as the ground rather than 0V, and to be honest i'm still not entirely clear on it.

in my circuit the clean signal gets a non-inverting op amp to maintain the same phase as the dirty signal, which means it gets a DC bias injected directly into it across a 470K resistor before the op-amp, and which is cancelled out by a 10uF capacitor (trying to retain as much low-end in the clean signal as possible in case it's being used as a bass pedal) before the volume/mix control. so at that point it should be an un-biased bipolar signal. is it because of the bias of the subsequent summing amplifier stage that the ground for the clean volume/mix control should be VREF?

Yeah I think that’s a good idea at of saying it. Your signal can be bias around 0V or 4.5V, and it can jump back and forth within your pedal. Each time you want to jump bias, you need a capacitor to block the DC offset.

Any op amp stage will be operating at 4.5v bias. You might as well keep it in between if there’s no reason not to jump. So putting pot reference to GND would bias the signal to 0v in that area, so you’d need a capacitor to isolate from the next op amp stage (the summing).

 

[thunderaxe]

that's how i understand it. correct me if i'm wrong but it seems like there are some instances in which i need the coupling capacitor to cancel the existing bias, only to re-introduce it in the next stage. for instance, in between inverting and non-inverting op amp stages. in my circuit (check the extremely long falstad link in a previous post if you want to take a look at it) the first two stages of the dirty path (boost and sallen-key filter) are non-inverting so they don't need a capacitor between them. but the next stage (clipping) is inverting, which introduces the bias via a different method (VREF into the non-inverting input, rather than directly into the signal path), so that requires a cancelling of the existing bias by a coupling capacitor... right?

and then the next stage (active tone control) is also inverting so there's no need for a coupling capacitor between those two. but then the last stage, the summing mixer, is mixing together the outputs of both inverting (dirty) and non-inverting (clean) outputs, so coupling capacitors needed for both paths?

 

[phi1]

also, if i'm giving up on my idea of using the buffered clean signal path for a buffered bypass option, i can use the clean buffer op amp as a boost and have equal value resistors on both the clean and dirty sides for the summing amplifier. if i'm using the jazz bass style volume control for the clean signal, any idea what value that pot should be?

10k is probably still reasonable starting point.

I would put a series resistor before lug 2 so that you aren’t overloading the op amp output with a short to GND

 

[thunderaxe]

10k is probably still reasonable starting point.

I would put a series resistor before lug 2 so that you aren’t overloading the op amp output with a short to GND

so 10k fixed series and a 10k pot?
if i put the 10k fixed before the pot like you suggested, could that serve double duty protecting the op amp and setting the gain for the summing amp?

 

[phi1]

so 10k fixed series and a 10k pot?
if i put the 10k fixed before the pot like you suggested, could that serve double duty protecting the op amp and setting the gain for the summing amp?

Yeah that seems possible, time to test it out I think.

 

[cdwillis]

Maybe a better idea is to have the blend set so the clean and dirty signals are evenly mixed at the summing op amp and make the clean side a clean boost that you can increase the gain/volume on before it hits the summing op amp. That would probably make a fine additive clean blend since you're not wanting to mess with lowering the dirty side. You'd just have to make sure both signals are in phase before they hit the summing op amp.

 

[thunderaxe]

it's not a bad idea, unfortunately in order to keep it in phase with the dirty side (which is 2x non-inverting and 2x inverting = not inverted) i'd have to use a non-inverting op amp as the clean buffer, which means i can't go below a gain of 1, and thus can't dial out the clean signal entirely using this method. it would be possible in theory to change the first stage of the dirty side to be inverting, and i could then do the same for the clean buffer/boost so it's worth a try but i have a couple of other options i'd like to try first.

 

[jesuscrisp]

You could try the blend from the Scenario/Gladio SC

 

[thunderaxe]

You could try the blend from the Scenario/Gladio SC

right, that would be what cdwillis is describing, but in order to keep it in-phase with the dirty section and not waste an extra op amp just to invert the phase again, i would need to invert the phase of the first stage of the dirty section.

what i was trying to do at first was use the clean blend from your modified scenario/gladio, the way you had it in your first post.

 

[thunderaxe]

you know what? i take everything back.
i found the problem with my breadboarded circuit -- i forgot that i wasn't pulling my VREF from the base pedal (which has a filtering cap for its VREF), but creating my own on the breadboard, without a filtering cap. i had initially figured, so what if there's a little ripple in the supply, this is just for testing. what i neglected to consider was that if i'm dumping any amount of signal into VREF as a way to ground it out, and i don't have the filter cap in there to immediately dump any AC signal straight to hard ground while allowing the DC bias to remain, then of course i'm going to get weird feedback loops and other interactions between that signal and everything else that VREF is connected to.

as soon as i added the filtering cap, every suggested circuit on here worked perfectly, including @jesuscrisp 's original one. which also is great because i found that when i was boosting the clean signal with the first preamp to match the dirty signal, only to eventually have it attenuated by the output volume control back down to essentially unity, not only is that (theoretically) wasteful and (theoretically) introduces noise unnecessarily, but i found i was clipping the op amp at points with what was supposed to be a clean signal, especially when playing a bass.

the only weirdness i found that still confuses me a little, was that when i was boosting the signal with the first op amp, because it was in non-inverting configuration the way to set the gain includes a resistor to ground, and when that ground was VREF, no problems, but when it was across a 4.7uF electrolytic capacitor to 0V ground, the same as the first stage of the rotten apple/sour grape dirty path other than the value and type of capacitor (it uses a 470n metal film), everything went weird and the signal would just kind of pulse in and out. any ideas why?

 

[neiltheseal]

Is there such a thing as a transistor panning mixer? As in something that works the same as the op amp mixer in .2 of this list, but using a transistor instead.

Would it be as simple as replacing Ic1.2 with some simple transistor gain circuit like a LPB-1?

 

[Chuck D. Bones]

Good write-up, BtR. To expand on what Feral said about partial phase inversion...
ANY CIRCUIT with EQ of any kind will alter the phase at some frequencies. The phase shift might be a little, it might be a lot. These freq-dependent phase shifts will cause peaks and notches in the freq response when two circuits are blended. Changing the Blend or EQ will make those peaks and/or notches move in freq, or come and go. Without an accurate circuit simulation, it is difficult to predict the results. If you don't believe me, try bridging the inputs on a Fender amp and then play with the tone and volume controls.

 

[Chuck D. Bones]

To go a bit further, blending non-linear circuits (fuzz, distortion, OD, octave) creates a whole different set of "surprises."

Let's look at a simple one: Octave. The point of an octave pedal is to generate 2nd harmonic while suppressing the fundamental. If you put a clean blend around an octave circuit, then you're adding the fundamental back in. The octave signal is then less pronounced. This may be desirable and some pedals do it.

Let's look at an extreme case (I actually tried this one): Put a clean blend around a BMP. Suppose we set the Blend so that with very loud notes, the clean and dirty signals have the same volume. We'll hear a medium dirty signal that quickly decays into a very dirty signal. Not amp-like in the least. Let's set the Blend so that with loud notes, clean dominates. We'll hear a mostly clean signal that quickly decays into a very dirty signal. Even more unnatural. We could set the Blend so that the dirty signal always dominates, but that will sound the same as a purely dirty signal with no clean blend.

Putting two dirt pedals circuits in parallel has the same problem because the gain, compression and EQ of the two circuits will be different, otherwise what's the point?

The ONLY pedal I've heard that does the clean-blend correctly is the Klon Centaur. Bill Finnegan worked very hard to balance the tones across the range of the GAIN control. He had to use a dual pot to accomplish this. The effective tapers* of both halves of the GAIN pot had to work together to maintain a proper balance.

My point here is that unless you are going for a very weird tone, putting a clean blend around a dirt pedal takes a lot of knowledge and effort. Or maybe you'll just get lucky.

I'll suggest that a better way to put two pedals in parallel is to split your guitar signal into two separate paths, run those paths thru whatever pedals you like and then run the outputs from those two paths into separate amps.

* Effective taper refers to how the pot rotation interacts with the circuitry around it. A linear taper does not always have a linear response.

 

[Chuck D. Bones]

In my mind, the circuits that benefit most from a clean blend are modulation and delay circuits. Most of them already have a clean blend built-in. Phasor, Vibe, Flanger, Chorus, Echo. In the circuits where the blend is fixed, it is easily made variable. I bought a Moen AM-CH BBD chorus 5 years ago specifically because it had a BLEND knob. Works great and still not a bad deal.