splitter blend, why the JFET?

[m_charles]

hello.
Was looking at this blender circuit. my question is, why is that JFET needed at all?
Wouldn't it be fine to send "green return" right into the 220n then into the opamp (U2A)?
Since it was designed by people way smarter than me, I gotta assume I'm missing something here.... 

[PRR]

In U2A inverting mode, the RETURN input impedance drops to 99K, which may be low for some sources.

I've instead wondered if U1A and U1B are optimal. Seems like one opamp should drive two of any input we find on stage. (Then the "saved" opamp can buffer the inverter.)

[Rob Strand]

I'm pretty sure the motivation for the buffer was ensure a high input impedance.
With the opamp along you either have a low input impedance (perhaps not the best
for old pedals with 500k output pots) or high noise (as you need to raise the 220k's),
which are already high.

Some bugs:
- The gate resistor is 1M but the input impedance is actually 2M or so due to the tapped source resistors.
- The JFET isn't biased very well.  It seems there is some loss of swing.  Also there is a small amount
   of signal loss in the buffer, which is normal, but you can strip it back.
- Even with the buffer the opamps resistor are 220k which is relatively high when considering noise.

[Dormammu]

hello.
Was looking at this blender circuit. my question is, why is that JFET needed at all?
Wouldn't it be fine to send "green return" right into the 220n then into the opamp (U2A)?
Since it was designed by people way smarter than me, I gotta assume I'm missing something here.... 

Don't overestimate the people who design these things.
Guitar electronics is a wide field for a lot of wilde experiments, myths and delusions.
I will say that you are right in thinking that this buffer is not really needed.

[ElectricDruid]

I've instead wondered if U1A and U1B are optimal. Seems like one opamp should drive two of any input we find on stage. (Then the "saved" opamp can buffer the inverter.)

Yes, this seems like a good idea. Do the split directly after *one* op-amp at the front, and then use the spare one to replace the FET buffer with another op-amp buffer. If you did that, you could knock a factor of ten out of the resistor values in the polarity-flipping stage too, which would help keep noise down a bit.

[m_charles]

I'm pretty sure the motivation for the buffer was endure a high input impedance.
With the opamp along you either have a low input impedance (perhaps not the best
for old pedals with 500k output pot) or high noise (as you need to raise the 220k's),
which are already high.

Some bugs:
- The gate resistor is 1M but the input impedance is actually 2M or so due to the tapped source resistors.
- The JFET isn't biased very well.  It seems there is some loss of swing.  Also there is a small amount
   of signal loss in the buffer, which is normal, but you can strip it back.
- Even with the buffer the opamps resistor are 220k which is relatively high when considering noise.

Thanks Rob. So, sounds like your kind saying with a few tweaks, the JFET may not be needed for most modern effects? like an average Boss pedal or similar? OF course JFETs are great for some things, but, its always nice to not have to use them, haha.

[m_charles]

hello.
Was looking at this blender circuit. my question is, why is that JFET needed at all?
Wouldn't it be fine to send "green return" right into the 220n then into the opamp (U2A)?
Since it was designed by people way smarter than me, I gotta assume I'm missing something here.... 

Don't overestimate the people who design these things.
Guitar electronics is a wide field for a lot of wilde experiments, myths and delusions.
I will say that you are right in thinking that this buffer is not really needed.

It's also from 2004, so... haha

[m_charles]

I've instead wondered if U1A and U1B are optimal. Seems like one opamp should drive two of any input we find on stage. (Then the "saved" opamp can buffer the inverter.)

Yes, this seems like a good idea. Do the split directly after *one* op-amp at the front, and then use the spare one to replace the FET buffer with another op-amp buffer. If you did that, you could knock a factor of ten out of the resistor values in the polarity-flipping stage too, which would help keep noise down a bit.

Ahh, this seems like a much tidier solution! nice

[m_charles]

I'm pretty sure the motivation for the buffer was endure a high input impedance.
With the opamp along you either have a low input impedance (perhaps not the best
for old pedals with 500k output pot) or high noise (as you need to raise the 220k's),
which are already high.

Some bugs:
- The gate resistor is 1M but the input impedance is actually 2M or so due to the tapped source resistors.
- The JFET isn't biased very well.  It seems there is some loss of swing.  Also there is a small amount
   of signal loss in the buffer, which is normal, but you can strip it back.
- Even with the buffer the opamps resistor are 220k which is relatively high when considering noise.

OMG, sorry Rob, didn't see you had marked up the schem already, haha. question answered!

[m_charles]

How does this look?
Replaced the 10uF and .22uf coming out of U2A with a single 1uF cap. 2 caps seemed redundant, also seemed like the cap polarity should be switched?, not sure if that's correct?

[mark2]

Sorry I hope this isn't hijacking, but on a very related note: I've always liked using the mixer from RG's article on the return like this.

But why use inverting on the inputs here? Any downside to switching those to non-inverting if there's no other high impedance stage before it?

[ElectricDruid]

But why use inverting on the inputs here?

Because the Blend circuit is inverting, so if the inputs invert, the whole lot is non-inverting overall.

Otherwise, non-inverting buffers at the front would definitely be preferred. There's only 100K impedance as it is (Ok-ish for the 70's, but really low by today's standards), and that could easily be higher with non-inverting.

Again, like above, you could probably use a single non-inverting buffer for the split at the front, and then use the saved op-amp to add an extra inverting buffer (with 10Ks not 100Ks!) at the end, so that the mix/Blend part is non-inverting too.

HTH

[antonis]

Any downside to switching those to non-inverting if there's no other high impedance stage before it?

Actually, in case of preceding stage high impedance existence, we DO have to switch to non-inverting configuration..
(with a large value VB bias resistor, of course..)

[m_charles]

But why use inverting on the inputs here?

Because the Blend circuit is inverting, so if the inputs invert, the whole lot is non-inverting overall.

Otherwise, non-inverting buffers at the front would definitely be preferred. There's only 100K impedance as it is (Ok-ish for the 70's, but really low by today's standards), and that could easily be higher with non-inverting.

Again, like above, you could probably use a single non-inverting buffer for the split at the front, and then use the saved op-amp to add an extra inverting buffer (with 10Ks not 100Ks!) at the end, so that the mix/Blend part is non-inverting too.

HTH

mind looking at the schem I made with the tweaks and telling me what you think? It's in my other reply

[ElectricDruid]

mind looking at the schem I made with the tweaks and telling me what you think? It's in my other reply

I will do, but Postimg is having a wobbly over here at the moment, so I'll have to get back to you on that...

[Rob Strand]

Thanks Rob. So, sounds like your kind saying with a few tweaks, the JFET may not be needed for most modern effects? like an average Boss pedal or similar? OF course JFETs are great for some things, but, its always nice to not have to use them, haha.

Yes it will work with many pedals without the buffer.

[Rob Strand]

Yes, this seems like a good idea. Do the split directly after *one* op-amp at the front, and then use the spare one to replace the FET buffer with another op-amp buffer. If you did that, you could knock a factor of ten out of the resistor values in the polarity-flipping stage too, which would help keep noise down a bit.

Also instead of using the switchable phase inverter opamp just have this structure for the Green Return circuit:
non inverting stage as is already present (output = green non inverting) -->  inverting stage with low value resistors like 10k and no resistors on the non-inverting opamp input (output = green inverting)

Then switch between inverting and non-inverting green outputs.

Like this:
https://www.parasitstudio.se/stripboard-layouts/simple-buffer-and-phase-inverter

[m_charles]

mind looking at the schem I made with the tweaks and telling me what you think? It's in my other reply

I will do, but Postimg is having a wobbly over here at the moment, so I'll have to get back to you on that...

Any luck viewing it?

[ElectricDruid]

Any luck viewing it?

Yep, it's working this morning!
I adjusted it a bit for you:


You could adjust it further by replacing the phase invertor op-amp with just a plain inverting op-amp, like Rob suggested, and then the switch just chooses the output from either the buffer or the inverted version to go to the blend pot. I don't know why exactly, but it does feel simpler than this way. There's nothing in it really though, so it doesn't *matter*.

[MrStab]

A dual op-amp would allow for this high-impedance input with better PSRR and buffered Vref on the other half. The discrete FET always felt a bit inelegant to me.