Blender

[shreditup]

Hey,
So i have built this here blender to mix in some delays(full wet on the Lexicon MPX110) to the dry signal of my Carvin Legacy clone. While it was better than running through my Lexicon, i still experienced some tone loss because of having to attenuate some dry in favor of the wet. So i decided to modify the design so that the dry signal remains un-attenuated, and the only variable is the delay level. And because i could set any delay level i wanted on the Lexicon's output and input knob, i decided that i wouldn't need a dry/wet pot on the blender itself.
So with my limited theory knowledge i came up with this. Not surprisingly, it doesn't work, and i guess the problem is in the design. Can anyone help me with this?
Thanks!

[J M Fahey]

Starting from the end:
2) you can *not* just link together Op Amp outputs.
You need to add separation/mixing resistors.
1) that blender pedal is absolutely flat and can not modify your sound.
And it does have the proper mixing resistors, made out of the pot halves, which have the added bonus of being variable to taste.
3) bonus tip. : I *guess* your delay *may* be inverting the signal phase, so at certain settings you *may* have weird cancellations.
But that would be a Lexicon problem; the blender is quite innocent.

[shreditup]

Ok, i get your point. While the problem might be in the lexicon, i would still like the dry signal not to be attenuated in favor of the wet delays. Judging from my experience the tone loss happens because the fx loop return gets a signal that differs in level from that in the fx loop send. So in other words, how can i modify the design so that the dry signal remains the same in level(while still remaining relatively unaffected), and only the delays can be varied in level? Just like in any delay pedal, increasing the mix knob doesn't get the dry signal attenuated( at least up to a point, 50% usually), i would like to achieve the same.

[J M Fahey]

You are achieving that now (with the original schematic), up to 50% rotation.
From 50% to 100% yes, the wet signal may surpass the dry one, which starts getting attenuated.
To gain more independence, you need to copy the channel mixer used in PA mixers, where each channel has its own, fully independent volume.
Consider the dry signal one channel and the wet one another.

[shreditup]

Well that's precisely the problem, the more i turn the knob from dry to wet, the more tone degradation i get(with the Lexicon's output level set to 0). But then again, i looked at the output section in the schematic of the Lexicon, and it seems to have no output cap and having a 7k resistor to ground(by the sum of resistances) so it essentially forms a voltage divider whenever i turn the knob from dry to wet. So i guess the blender was innocent after all. And i guess by adding an output cap in the Lexicon(well in the return of the blender actually) i can fix this?

[shreditup]

Ok, so the problem is indeed in the Lexicon. Whenever i plug the output of the Lexicon to the return of the blender(the original one) i get tone degradation, and by turning from dry to wet the signal level decreases. And when turning to full-wet the dry signal disappears almost completely(with the Lexicon unplugged i essentially only increase the series resistance up to 10k, so obviously i don't get any signal attenuation). This does seem like the pot forming a voltage divider with the mentioned 7k, but adding a cap(.047uf and 10uf) after blender return didn't solve the issue(although did alter it a bit). What can i do about this? My guess is that a buffer in place of the cap will help.

Btw, here's the Lexicon's output stage. Sorry for the crappy quality, but that's the way the whole schematic is.

[R.G.]

While that's the bog standard go-to blend ciruit that's most often used, it has some issues. You're finding some of them.

That circuit relies on the output impedances of the driving points (the opamp for dry and the return from whatever is on the send/return line) to be quite low compared to 10K, and for the impedance of whatever is connected to the pot wiper to be both high compared to 10K/4 and DC isolated. The unseen impedances on the outputs of whatever drives the pot and whatever loads the wiper are always with you, and cause issues unless you are lucky enough for them to be negligible. The attenuation curves for this simple blender are not ideal either, once you get past the impedance issues.

See Panning for Fun at geofex.com, where there's set of design information for how you might do this with less dependence on the other circuits involved.

[J M Fahey]

Agree and add a bad news: that output series resistor is 75 ohms; not 7 K.
That said, I must repeat that that blender schematic does not alter the tone *at all*, so what you hear may be some weird phase shift or cancellation in the Lexicon.
What does this mean?:
that even if you can build the circuit you want (and which *is* possible) , you will still get the sound you get now.

[shreditup]

Well actually, i had the chance to play at a normal volume today, and that 10uf cap i added in series after the blender return really improved the tone of both the dry and wet. I'm really pleased by it so far, and i'll probably stick with that, but in case i'll need even more channel independence, i'll probably build the one on the geofex, so thanks R.G. for the link!
And J M, what i meant by the 7k is the resistance to ground by the sum of that 75ohm, that 4.99k R56 and the 2.15k R57 and the fact that it formed a voltage divider to ground, which seemed to be the root of the problem. Although you may be right, maybe that cap only passes half of the waveform since it's an electrolytic cap, and that half is the one that's not cancelling the dry. Though that's just a pure guess and i don't really know if electrolytic caps pass only half of the waveform, and if not, then if the cancellation is the problem, how do i fix it?

[J M Fahey]

Glad you solved it.
No, by cancellation I refer to odd phase shifts within the Lexicon.
Electrolytic caps used properly pass the full wave.
And the resistance that matters is the series one.
Any resistance value from the last Op Amp to ground is irrelevant because said Op Amp behaves as a generatop with zero internal impedance, that's why outputs can't be paralleled, hence the need for mixing resistors.