Blender / mixer problem

[DUCKFACE]

Hello
im building an soft-hard clipping stompbox with mixer/blender between low and hiugh clipping.
the problem: realy low level of signal. Soft clipping is actually no clip at all. when i remove R15 - 1kbetween mixer pot and baxandall EQ the levvel is just fine - soft and hard clipping are working perfect except that you can use only soft or hard clip.
The pot for blending is working only on min or max level - you have to chose between soft and hard. I want to mix the signals.
Whne the pot is on middle position its absolute silence.
anybody sees a problem in the mixing stage ?

[GibsonGM]

I'm not sure I'm correct, Duckface, but first, I question how you're splitting the signal.  Maybe the 2 circuits are too interactive?  Maybe "IC1A" needs a 1k on its input... Seems some sort of JFET or opamp splitter might work better there, tho, instead of the BJT, feeding both (1 JFET per opamp).  AMZ has a simple JFET splitter schematic that I use sometimes for things like this, driving multiple circuits. 

I'd check out some of the split/blend stuff that's out there, runoffgroove dot com has a good one using opamps.

Re. mix pot issue, seems maybe there's something about too low a load on those opamp outputs?

[ElectricDruid]

Working in from the input left to right:

Why is R5 such a low value? If you're using it as a highpass filter and that's why it's so low, you'd be better off reducing the value of the cap instead and using 470K like on the other half. Otherwise the input impedance of this upper stage is very low.

Do you have solid signals at C10 and C8? (E.g. are both clipping stages working correctly at once?) If not, then whatever the blender stage is doing isn't going to make any difference!

Do you really need another 26dB of gain from the blender stages? (100K/4K7 = x21 = 26dB). You're following the carefully shaped clipping of the previous two stages with basic op-amp clipping. Is that the intention?
You could increase R11 and R12 to back the gain off a bit, and that would then allow me to use a non-polarised cap for C8 and C10 - say 470n instead.

Personally, I'd probably recommend replacing this part of the circuit with RG's panning/blender circuit. Does the same job equally well, and only uses one op-amp, so you save a chip.

   http://geofex.com/Article_Folders/panner.pdf

(You want the bit in the fine black box on the left)

Sorry if this is a lot of questions, but I'm hoping that thinking about the answers might help you fix the circuit!

[Processaurus]

Your problem is R15 (input to inverting opamp stage after blend pot) is way too low at 1K. The inverting opamp configuration makes a virtual ground at the - input, so that resistor sets the input impedance for that stage. Good voltage transfer requires 10x input impedance to the last stages output impedance (your 10k blend pot) so I would make R15 at least 50k or 100k, and the R in the feedback loop at least 100k or 220K.

Side note, one could eliminate all of the pots or trimmers for all of the gain stages except the clipping stages, you could just socket those resistors if you want to play with gain between stages- pretty much all you're looking for is to keep the gain up and out of the noise floor through the circuit without clipping, you have a wide window. Just the last stage before the volume should get tweaked with to set the max output gain.

EDIT: Actually IC3B (opamp stage after blend pot) could be changed to a non-inverting buffer/follower. Less parts, high input impedance, low output impedance.

[Processaurus]

Re: RG’s blender/pan circuit, it does work well, but with correlated signals like this, I would expect a volume hump in the middle. It does a half rms power thing (like .7x input voltage) in the middle of the pot rather than half voltage. With correlated audio, if both sides are at .7x gain in the middle, the two sides sum to ~1.4x gain, vs 1x on the ends of the blend pot.  It does work well for two unrelated sounds, like dry/wet on a reverb or delay, or line mixing two instruments.

I think your simple blend pot is more user friendly for correlated sound blending. The pair of op amps feeding it have a low output impedance, and an identical output impedance. And even if the opposite side bleeds a tiny bit at the extreme end of the pot rotation it may not be audible in this application. 

[iainpunk]

there is a 10k ''GAIN'' pot around IC3B, and it determines the level coming out of the opamp, what is that pot set to? does turning that pot change anything?

cheers

[ElectricDruid]

Re: RG's blender/pan circuit, it does work well, but with correlated signals like this, I would expect a volume hump in the middle. It does a half rms power thing (like .7x input voltage) in the middle of the pot rather than half voltage. With correlated audio, if both sides are at .7x gain in the middle, the two sides sum to ~1.4x gain, vs 1x on the ends of the blend pot.  It does work well for two unrelated sounds, like dry/wet on a reverb or delay, or line mixing two instruments.

Fair point. It's true that it's set up for constant power panning, so when the two inputs are similar, you get a +3dB hump in the middle. That's not a huge amount though. You can reduce this effect by making the pot smaller w.r.t. the R value. By the time you're to Pot = R/5, the hump is only +1.36dB, and at Pot = R/10 (So 10K pot with 100K resistors) you get only +0.74dB. So my advice is "Tweak it!"

[DUCKFACE]

Thanks for the fast responce:)
The reason for so many pots in the gain stages - im experimenting with different type of diodes in clipping stages. some of them like 1n34a are kind of quite. so i need a little bit of gain.
So far i'v changed the R5 to 510k - tubesreamer still wont to scream. Drive pot just increases the volume but not the clipping.
When there is no resistor R15 tubesreamer works just fine. If i put any resistor there soft clipping stage is not clipping. 
After following hte advices of Processaurus to put 100k on R15 and 100k in the feedback mixer is working BUT pot for clipping in the hard part is working like volume pot and the soft clipping is not clipping at all.

edit: R5 on tubescreamer is 10k ? do you think that 510k is kind of much ?

edint on the edit: if i take the signal direct from ICb2 the soft clipping part is not clipping too. so the problem of non clipping is not R15

[antonis]

After following hte advices of Processaurus to put 100k on R15 and 100k in the feedback mixer is working BUT pot for clipping in the hard part is working like volume pot and the soft clipping is not clipping at all.

Turn GAIN pot into 1M..
(to maintain X10 ratio..)

R5 on tubescreamer is 10k ? do you think that 510k is kind of much ?

Its value depends on previous stage output impedance and op-amp input bias current..
The former counts for signal attenuation where the later for Vref offset..

510k should result into about 400mV voltage drop (800nA max input bias current) so better proceed to 100k (and make C3 100nF)
(it will also be more than fine with input buffer output impedance..)

[Processaurus]

Re: RG's blender/pan circuit, it does work well, but with correlated signals like this, I would expect a volume hump in the middle. It does a half rms power thing (like .7x input voltage) in the middle of the pot rather than half voltage. With correlated audio, if both sides are at .7x gain in the middle, the two sides sum to ~1.4x gain, vs 1x on the ends of the blend pot.  It does work well for two unrelated sounds, like dry/wet on a reverb or delay, or line mixing two instruments.

Fair point. It's true that it's set up for constant power panning, so when the two inputs are similar, you get a +3dB hump in the middle. That's not a huge amount though. You can reduce this effect by making the pot smaller w.r.t. the R value. By the time you're to Pot = R/5, the hump is only +1.36dB, and at Pot = R/10 (So 10K pot with 100K resistors) you get only +0.74dB. So my advice is "Tweak it!"

Still not sold it would be a better blend for correlated sounds, but that is a really interesting way to tailor that circuit.  Worthy to add to the Geofex article. One possible penalty of reducing the volume hump in the center by going 10x on the resistors is you're effectively reducing the gain to 1/10 at the pot, then (presumably) increasing it 10x with makeup gain on the next stage.  Could add noise.

[ElectricDruid]

Still not sold it would be a better blend for correlated sounds, but that is a really interesting way to tailor that circuit.  Worthy to add to the Geofex article. One possible penalty of reducing the volume hump in the center by going 10x on the resistors is you're effectively reducing the gain to 1/10 at the pot, then (presumably) increasing it 10x with makeup gain on the next stage.  Could add noise.

Yes, this is also true. The required make-up gain does increase. RG's constant power version needs x3.41, +10dB. Flattening the curve out pushes the required gain up to x12.5, +22dB. More gain never helps your noise figures, so it's worse from that point of view.

TBH, I'm still looking for the perfect "single pot" way to do this.

[DUCKFACE]

Mixer stage done. 100k before IC3b and 220k for gain pot and the job is done.
the problem with soft clipping stage dat wont clipping stays.

[antonis]

Are we talking about this section..??

[DUCKFACE]

thats right. it was working just fine till i start to play with the clipping pot and R15 and R5.
In the beginin R5 was 1k, after that i put 4,7k and from final advices i put 100k.

[antonis]

R5 is effectively set in parallel with R4 (and R6) thus dominating input buffer Emitter equivalent resistance..
(so, the higher its value, the better the AC load line..)

Anyway, I don't think your issue lies around R5..
Have checked R8, R9, R11, R14 and respective pots for correct values..??

[DUCKFACE]

all of the elements are correct.
im wondering about R9. is it suppose to be connected to the ground or to +4.5v
on a different schematic is different
for example :
1. from electrosmash:
https://www.electrosmash.com/images/tech/tube-screamer/tube-screamer-block-diagram.png

2. from originalfuz:
https://cdn.shopify.com/s/files/1/0097/1902/files/static1.squarespace.gif?8068671476027599051

for me it looks reasonable to be like all of non inverting opams are on the ground but.... ?

[antonis]

As far as there is a big electro cap on Vref, it doesn't matter..

In your schematic, there are 2 of them, so R9 goes both into GND & +9V..
220μF has a capacitive reactance (impedance) of less than 10 Ohms at lowest frequency of interest..
+9V power supply internal resistace is considered negligible..
So, R9 "sees" a series resistance of less than 5R to GND..
(Frankly, R9/C7 HPF doesn't give a damn about that tiny weeny "obstacle"..)

Connecting items to AC ground (Vref here) might serve PCB design..
(some guys also claim for AC ground clarity 'cause actual GND might contain DC "garbage" but this is another story..)

[DUCKFACE]

double checked:
R8 - 1k
R9 - 4,7k
R11 - 4,7k
R14 - 51k
still no clipping. i think i just made absolutely no clipping circuit 
the gain pot is increasing just the volume but not the clip.
do i need R8 ? in the tubescreamer its valued by zero.

edit: iv even change the diodes to blue leds ... still no clipping
changed 4558 to tl072 and ne5532 - still no clipping

[antonis]

Make C7 47nF (or better, 100nF)..!!
(as it is, R9/C7 HPF has cut-off frequency at 7.2 kHz..)

[DUCKFACE]

i made it 68n. Everything is just fine now.
sounds kind of .. ruff. how can i make more smooth, creamy sound ? diodes? any suggestion for tweaks ?
btw thanks Antonis