Splitter Blend Modification (blend pot pre loops possible?)

[iandy4]

Hi Guys,

I built the splitter blend circuit a bit ago, http://www.runoffgroove.com/splitter-blend.html and I really like it. I built it to use with a foot pedal to mix in my delay pedal with the clean signal.  It does that but there are two things that have started to bug me.  

1) When putting my toe down on the pedal, I mix in the delay but you can hear the delay was already seeing the signal which is a problem for how I intended to use it.  As I mix it in you can hear my previous chords (played while with my foot in the clean position).

2) When putting my heel down to go back to the clean sound, it instantly cuts all the ambient noises from the delay.  

Correct me if I am wrong but I think this is because in the circuit the fx loops happen before the blend pot.  If the blend pot was before the fx loops then only the loop that was selected would see the signal, solving the first problem I mentioned.  I'm not sure but I think it would solve the second as well because the loops lasting noises aren't really cut off if the blend pot is before them (I hope!).

Sorry for the long post but I am determined to get this working.  I made an educated guess based on the original schematic (refer to link above) as to how to put the pot before the loops and tried it out on a breadboard.  The blend pot didn't do any blending.  So I made a diagram of what I did and was hoping someone with experience could point me in the right direction.  Here is the schematic (that didn't work):



Any help would be grrreeeatly appreciated!!

[iandy4]

bump

[caress]

you really want to blend at the point where the 2 10uf caps meet the output.  try that instead!

[iandy4]

you really want to blend at the point where the 2 10uf caps meet the output.  try that instead!

Thanks Caress, but then it ends up being pretty much the original schematic in the link I posted.  My goal is to get the blend pot before the loops in the signal chain in order to solve the two problems that I mentioned in the first post.  I have a bad feeling that it's not nearly as easy as the schematic I posted...

[ashcat_lt]

I can't see why that wouldn't work.  Of course, I can only see as far as the output caps.  Can we assume you've got mixing resistors before the output. 

Perhaps your blend pot is just too small and therefore not providing enough isolation?

[caress]

you really want to blend at the point where the 2 10uf caps meet the output.  try that instead!

Thanks Caress, but then it ends up being pretty much the original schematic in the link I posted.  My goal is to get the blend pot before the loops in the signal chain in order to solve the two problems that I mentioned in the first post.  I have a bad feeling that it's not nearly as easy as the schematic I posted...

whoops yep brain fart.
what you posted should work though - maybe do what ashcat suggested and try a larger pot.  100k?

[Projectile]

No, a bigger pot wont help. This circuit will not work. The noninverting inputs of those opamps are Megaohms, and the bias resistors are 1Meg. All that mixing pot is doing is changing the series resistance in front of them. The original mixing circuit works because the outputs of the opamps are very low impedance. Each opamp output basically looks like ground to the other opamp output, so the mixing pot creates a voltage divider. In your circuit, the input side of the opamps are high impedance, so that circuit is not going to do jack unless you use some absurdly huge value for the pot, which creates other problems.

Also, your opamp inputs are all referenced to ground. Unless you are using a dual supply, they need a 4.5v reference.

I'll try to cook up something that should work. Give me a few minutes.

EDIT: This is a more challenging problem than I expected. Maybe someone more clever than me can come up with a solution. All of my attempts require either a dual-gang pot (or separate mix controls), or they don't linearly mix the signals and you end up getting a humped (or attenuated) response in the middle pot position. I can't seem to come up with an elegant solution that only uses a single pot. If you can install a dual gang pot in your foot pedal, I can show you a solution that will work, but that's all I've got for now. Good luck.

EDIT2: Alright, I had a stroke of genius and came up with a circuit that should work in theory (with a single mix pot), but it's a little dicey and slightly complicated. I'll have to try building it first to make sure it works as intended. In the mean time, I'll see if someone else can come up with a simpler design. I might be overcomplicating this.

[iandy4]

No, a bigger pot wont help. The noninverting inputs of those opamps are Megaohms, and the bias resistors are 1Meg. All that mixing pot is doing is changing the series resistance in front of them. The original mixing circuit works because the outputs of the opamps are very low impedance. Each opamp output basically looks like ground to the other opamp output, so your mixing pot creates a voltage divider. The input side of the opamps are high impedance, so that circuit is not going to do jack unless you use some absurdly huge value for the pot, which creates other problems.

Thanks so much for the explanation! I'm very glad to know why it doesn't work.  At least now I have a few ideas of what to google.

Also, your opamps all referenced to ground. Unless you are using a dual supply, they need a 4.5v reference.

Oops, I made that mistake in the diagram but not when I built it.  The connections to ground from the resistors at the buffers inputs should all go to Vref.

I'll try to cook up something that should work. Give me a few minutes.

EDIT: This is a more challenging problem than I expected. Maybe someone more clever than me can come up with a solution. All of my attempts require either a dual-gang pot (or separate mix controls), or they don't linearly mix the signals and you end up getting a humped (or attenuated) response in the middle pot position. I can't seem to come up with an elegant solution that only uses a single pot. If you can install a dual gang pot in your foot pedal, I can show you a solution that will work, but that's all I've got for now. Good luck.

I really appreciate that you've tried! I'd definitely like to see the dual pot solution.  I think I can install one but the price difference is huuge.  I'll keep looking for a one pot solution but thanks so much for explaining the cause.  I find that almost more helpful than finding a solution with no explanation!

[Projectile]

I added another edit to my post above while you were replying. I may have come up with a working solution (no dual gang pot), but it's messy. I'll see if I can find some time to try it out. If you really want, I could post the circuit, but I will give you no guarantee that it works, since I haven't actually built it. It would probably be better to wait and see if someone else comes up with a less complicated solution

[iandy4]

Hey Projectile,
This might be a stupid question but if the problem is that the buffers have a super high input impedance, if I were to use inverting buffers instead of non-inverting, couldn't I set the input impedance of the buffers following the blend pot to be lower?  Do you know what the ideal impedance would have to be for the blend pot to work correctly?

I was looking at this page that explains how with an inverting buffer, the input impedance is set by R2. (the section called "Inverting and Non-Inverting Buffers" half way down the page): http://sound.westhost.com/dwopa.htm

[iandy4]

I added another edit to my post above while you were replying. I may have come up with a working solution (no dual gang pot), but it's messy. I'll see if I can find some time to try it out. If you really want, I could post the circuit, but I will give you no guarantee that it works, since I haven't actually built it. It would probably be better to wait and see if someone else comes up with a less complicated solution

I'd definitely like to see it anyways! In the meantime I'll wait to see if anyone has another solution.  Again thanks for your help.  You are the man!

[iandy4]

By the way it's incredibly thoughtful of you to even consider building it for me.  I love this online community and hope to contribute when I know more.

[Projectile]

Hey Projectile,
This might be a stupid question but if the problem is that the buffers have a super high input impedance, if I were to use inverting buffers instead of non-inverting, couldn't I set the input impedance of the buffers following the blend pot to be lower?  Do you know what the ideal impedance would have to be for the blend pot to work correctly?

I was looking at this page that explains how with an inverting buffer, the input impedance is set by R2. (the section called "Inverting and Non-Inverting Buffers" half way down the page): http://sound.westhost.com/dwopa.htm

Yes you could lower the input impedance that way, or you could lower it more easily it by just putting a smaller resistor to ground in front of each input (but not too small or you will distort the output of the buffer feeding it). If you use the inverting configuration, remember that the relationship between the input impedance (pot + R2) and R1 also effects the gain!

The main problem I see with these two approaches is that since you can't tie the inputs of the opamps directly to ground, you will never be able to completely get either signal to drop to 0v, and they will always be slight mixing at either end of the pot. You can fix this problem by making the value of the pot very large, so that the small input impedance becomes insignificant, but this creates another problem: as you increase the value of the pot in relation to the input impedance, you end up getting signal attenuation toward the middle position too, so the pot will never have a linear response.


Also, don't forget to add the mixing resistors before the output as ashcat already mentioned.

By the way it's incredibly thoughtful of you to even consider building it for me.  I love this online community and hope to contribute when I know more.

I'm only really doing it to satisfy my own curiosity. Seems like it would be a useful circuit to have in the toolbox.

[iandy4]

Haha well it's still thoughtful to post the results for me   I wasn't sure if you saw my post but if you have the schematic i'd definitely like to see it and try it out myself as well.

[Projectile]

I'd have to draw it up. Not going to happen tonight, but maybe tomorrow.

[iandy4]

I'd have to draw it up. Not going to happen tonight, but maybe tomorrow.

No problem! Thanks a lot!

[iandy4]

Anybody else know of a solution?

[Projectile]

Try this:



I haven't had a chance to build this yet, so I can't guarantee it works, but theoretically it should.

As stated in the schematic, the trimmer U4 is optional. The idea is that you want the gain of U3 and U4 to be exactly the same, and some kind of slight gain adjustment would help. If you use a trimmer, be sure to lower the value of the 220k resistor that's in series with it so that they add to the proper value when the trimmer is in the middle of its range. Another way to approach this problem is to replace the two 25k mix resistors in the send1 stage with a 50k (or 100k, the value isn't critical) trim pot (set in middle position with wiper to output). this would allow you to adjust the mix rather than the gain of each opamp. You can try building the circuit without any trimmers and see how much bleed you get. It might not even be noticeable, especially if you use that channel (send1) for the dry signal. I'm not sure until I try it. You might get by fine without using any trimmers at all.

The circuit works by adding together two copies of the exact same signals 180 degrees out of phase in order to cancel each other out and produce zero output. That's why matching the gain is important. If one of the signals is stronger than the other, then you will get bleed when there is supposed to be no signal.

Parts values are not critical in this circuit. A wide range of values should work. Just make sure that the 2 resistors on U4 are around at least 10 times the size of the mix pot in your foot pedal, or it will start to mess with the gain. What's important in this circuit is that the matching resistor pairs used for U4 (220k) and the two mix resistors (25k) in the first stage, are as close to each other as possible. If you have metal film, use them here. Otherwise try to find close matching values for each pair, and make sure you use the same opamp for U3 and U4!

Also, I omitted some inter-stage coupling caps for sake of simplicity. I've heard of cases where DC voltage across a pot can cause pot noise, but I've never personally had a problem with this and stopped worrying about it in my builds ages ago. If you do have problems with any noise on your blend pot, then add the appropriate coupling caps and reference voltage resistors.   

I'm curious to see if this works for you. I have a busy weekend, so I probably wont have time to build this before you do. Good luck.

[iandy4]

It works!!! Yes!!

It appears to attenuate the signal and adds some hum but I had to make several parts subs. and its built on a bread board and also didn't use a trim pot so I'm pretty sure those problems are coming from the way I had to build it.  I'll go to the store and get the right values soon and report how it sounds when I can properly build it.  Either way it already works quite well! Thanks so much projectile!

One last question.  I posted my non working diagram without the phase switching part from the original that its based off of (http://www.runoffgroove.com/splitter-blend.html) but that is definitely a feature it should have.  Is it as easy as adding it to return 1 like the original schematic?

[Projectile]

It works!!! Yes!!

It appears to attenuate the signal and adds some hum but I had to make several parts subs. and its built on a bread board and also didn't use a trim pot so I'm pretty sure those problems are coming from the way I had to build it.  I'll go to the store and get the right values soon and report how it sounds when I can properly build it.  Either way it already works quite well! Thanks so much projectile!

One last question.  I posted my non working diagram without the phase switching part from the original that its based off of (http://www.runoffgroove.com/splitter-blend.html) but that is definitely a feature it should have.  Is it as easy as adding it to return 1 like the original schematic?

Cool! Glad it works. I'm not sure about the hum problem. I wasn't expecting that. I was more concerned about signal bleed on send1 being an issue. Adding in inter-stage coupling caps might solve the hum issue. Maybe one of the experts here could take at the schematic and point out some design issues I may have overlooked. I'm no engineer. 

You should be able to substitute in the phase switching from the runoff groove schematic just fine.