Splitting signal through two gain stages, then merging. Best method?

[chumbox]

Hey All

I'm trying to establish the best way to split my input signal through two separate gain opamps and then merge it. I'm not worried about blending them with a pot they can just both come together.

Being they are both gain stages do I need a pre and or post buffer or will the gain stages be enough and can I just merge the signals as below?

[chumbox]

Deleted. Double post.

[PBE6]

Method A is probably fine, but strictly speaking "it depends."

You want high input impedances on the front end of each stage, which you can achieve with a JFET-based opamp like a TL072, a non-inverting gain stage configuration and a suitably high biasing resistor. Other choices may complicate things.

You also want low output impedances on the back end of each stage, which you automatically get by virtue of standard opamp operation. Just make sure the total load is at least 2k or so to prevent issues with the TL072.


Sent from my iPhone using Tapatalk

[chumbox]

Thanks PBE6. I was planning on running TL072s so what you are saying sounds like a good plan.

[robthequiet]

Ditto. I vote for B with a mixer/summer output amp at the output.

[Keppy]

C.

Opamp inputs are high impedance, so they don't load down the input source, so the effect of the buffer is small. That means B is not much different from A for most uses.

Opamps set their output through feedback to the input. Tying the outputs together will cause the two opamps to try to overpower one another, as they can't both have the correct feedback at the same time if the two drive controls are even a little bit different. That means both A and B are bad from a design standpoint, although that's no guarantee you won't like it.

For predictable operation, you need to mix the outputs together through resistors. 1k-10k is probably fine. If you're tacking a volume control onto the end of the circuit, then the added resistance should be negligible. You'll lose half the volume of each stage mixing them passively, so if you want the full volume of each stage you need a summing amp stage.

[chumbox]

Hey Keppy. Are you saying I need a resistor on the output of each opamps stage before they come together? From all the replies I will definitely add a summing amp before hitting a volume pot. I'll breadboard it anyhow and see what works.  Just wasn't exactly sure where you meant to put these resistors?

[PRR]

> Are you saying I need a resistor on the output of each opamp

Yes.

Summing amp is a frill you may not need.

Not at all clear what you expect from two parallel amp paths. If they are linear, the output is the same. If they distort different, yes, they will distort different and you will get some function of the input. But the experiment is cheap so go ahead and try it.

[chumbox]

Thanks all. The concept is coming from Sunn amp style inputs whet you can choose the both input and the signal is split. Was going to try something similar on a pedal input and see what I can get out of it.

[Hatredman]

Ditto. I vote for B with a mixer/summer output amp at the output.

If this is a poll, B + mixer gets my vote.

[chumbox]

Based on having a look at a first gen Model T schematic and everyone's comments here is my latest loose scribble. A lot of values yet to be determined from the breadboard.  Still not sure if I even need a buffer on the end, depends on my volume drop from the tonestack.  Turns out the bright channel and the normal channel on the preamp really only vary by the one 470pF cap, so if there is a better way to combine these than splitting a signal, that would save a lot of potential headaches.

It'll hopefully give me a bit of a Sunn preamp vibe with a bit of tinkering, or maybe the split signal will be a massive pain in the butt, who knows.  Feel free to add comment on anything that I should be tinkering with.

I'll keep in touch as I work on it. Really appreciate the advice so far 

[merlinb]

Turns out the bright channel and the normal channel on the preamp really only vary by the one 470pF cap, so if there is a better way to combine these than splitting a signal, that would save a lot of potential headaches.

As drawn your first 'splitting' stage doesn't do anything. Firstly the component values basically just give you a flat frequency reponse -the model T uses 470k resistors, not 10k. Secondly the result is exactly the same as if you put the lower 22n+10k path in parallel with the upper 22n+10k||470p path. There is no point in path splitting if you're only doing EQ; it's only useful if there's some sort of nonlinearity difference (e.g. different clipping diodes) in the separate paths.

[antonis]

Mark's FLEXIDRIVE may give you some ideas....
(maybe replacing pots with mixing resistors..)

http://www.diystompboxes.com/smfforum/index.php?topic=51810.msg389960;topicseen#msg389960

[chumbox]

Thanks merlinb and antonis for the extra advice.  This has really given me something new thoughts about the process.  Mark's flexidrive sheds a bit more light on some ways to either tweak a split signal to make it more interesting, or like merlinb said, maybe splitting it ain't worth the effort to achieve the goal.

Really appreciate everyone's thoughts.
Back to the board. 

[robthequiet]

Merlinb's is good point -- also, in the summing amp instead of merging them through the pot and then the summer, what I would have done is take the output from the series resistors straight into the op amp and putting your gain control at the input buffer, thus driving your middle gain stage/clipping circuit, if you were adding one. Then a final pot after the tone control. As drawn, not sure if you would need to recover gain after the tone control if you had enough signal coming after your mixing opamp. I am working on a multi-band drive/distortion idea and ran into pretty much the same thing by trying to split the signal into different paths, as in are they sufficiently different that they change the sound more than just keeping one path and using filter caps and gain stages for the same effect.

Definitely play with it a bit and let us know what you find.