Two Channels, One Volume, One TMB Tone Stack - Mixing Resistors.

[pdf64]

Shared anodes or mixing resistors, the gain / signal level will be at least halved.
However, shared anodes will have a somewhat lower output impedance (and so drive a tone stack ‘better’), than mixing resistors.

[Helmholtz]

I would like to know:
- What are the signal sources?
- Will they be used simultaneously?.
- Do the input stages need to have different properties (gain, frequency response..)?
- How many tubes are available?

[uneumann]

The OP question about signal mixing has varied answers based on varied assumptions. Mixing can be a complex design problem. The best approach depends on many factors.
This thread seems like a “teachable moment” so I’ll offer a brief survey of approaches and their pros and cons.

Mixing resistors – This is the most common approach. The valve wizard page covers this pretty well so I’ll just say it works well when the load impedance is higher than the mixing resistors and the input stages have lower impedance than the mixing resistors. A typical case is where two preamp stages (plate signals) are mixed with two resistors and the load is another gain stage grid. Using ballpark numbers, the plate signals have an output impedance of 50k. The mixing resistors will range from 100k to 1M, and the resistors feed a grid with impedance of 1M or greater. There is always signal loss (as others stated) due to the mixing resistors. Each input gain loss may be different and controllable by using varied R values (e.g., 200k and 510k). As pdf64 points out – it’s not ideal to use this approach for tone stacks designed to be driven by a single tube since the effective source impedance is raised by the mixing Rs.

Connecting plates – This also has a gain loss, as described in the thread, but the advantage of this approach is it has a low output impedance which is often desirable to drive a tone stack. A disadvantage is pointed out by Helmholtz – if the plates have very different load resistors, voltages, etc, their variation is lost by shorting the plates. Both plates will see the same load and voltage. This may be OK or not – it depends on the situation.

The above methods both decrease an input’s signal by about 50% and possibly more. You can think of them as passive mixers.
There are also active mixers that are good to be aware of and may be useful in more demanding cases.

Current summing – good examples of these are the mixing circuits used with opamps. They depend on a “virtual ground” into which a signal flows through an input resistor. That signal current produces an output that cancels the input current. You can find more about this at varied sites (tubemonster posted some links). There are also tube versions of this approach. Typically, these have moderate Input and low output impedances so they are very flexible, but they come at a cost (a gain stage).

Voltage summing – this is another active mixer that requires additional stages. It’s a less common type of mixer, but a form is used in the fender 5G15 reverb. In this circuit, a low impedance source is mixed with a high impedance source through a resistor. The output is taken from the high impedance side of the resistor. The gain is unity (for the low z input) and the gain is variable for the high-z input. The mix proportions are therefore set by varying the high-z gain. This is ideal for reverb circuits when the dry signal is the low-z input, and the reverb pan signal is the high-z input. Since the dry signal has unity gain, this circuit is ideal for inserting into existing amps without impacting their gain structure. I really like this circuit (personal preference) and it’s used in my 5e3 reverb card (among others).

https://sites.google.com/site/ampaid16/home/5e3verb The schematic is below… The dry signal goes to Q8 which is a follower to create the low-z signal. The high-z signal comes from the pan amplifier through Q9. The two are mixed in R26, whose upper output feeds the output buffer Q10. In the 5G15, the low-z signal comes from a cathode follower and the high-z signal comes from a plate and the mix resistor is a pot which controls the relative gains.

[tubemonster]

I would like to know:
- What are the signal sources?
- Will they be used simultaneously?.
- Do the input stages need to have different properties (gain, frequency response..)?
- How many tubes are available?

For me it's a curious trying to learn/understand better thing, a coffee table chat - beyond the normal usage in fenders, etc.
Essentially instead of two channels, two tone stacks, and two volume controls - have two channels, one tone stack, and one volume.
So, I'd say either or perhaps even both channels at once.

[tubemonster]

The OP question about signal mixing has varied answers based on varied assumptions. Mixing can be a complex design problem. The best approach depends on many factors.
This thread seems like a “teachable moment” so I’ll offer a brief survey of approaches and their pros and cons.

I couldn't agree more

[tubemonster]

@uneumann

I'll be checking the site out today and this weekend, thanks for sharing - appreciate it.

https://sites.google.com/site/stringsandfrets/

[tubemonster]

One last picture. This is just using a switch rather than mixing.
It's not mixing two channels, but it is an alternative to mixing so I guess it's part of the mixing topic

[Chuck H]

Well right off I'll say that you need a 0V reverence (resistance load to ground) at the switch side of the coupling caps to avoid switching noise.

After that I have to say...

I'm with Helmholtz at this point. What is the goal, arrangement needs, problem or other circumstances affecting the topic at hand? Channel mix circuits DO already exist. There are not many and that's for a reason. The few already in use after eighty years are the best way to do it. Is "different" a requirement of it's own?

[The Dude]

...... and what if you want to use both channels simultaneously?

[tubemonster]

Well right off I'll say that you need a 0V reverence (resistance load to ground) at the switch side of the coupling caps to avoid switching noise.

Ya, switching noise was a secondary thought, this pic kind of came into play from "- Will they be used simultaneously?" earlier in this thread.
And, this picture, would be a nope - not using both at the same type instance...

After that I have to say...
I'm with Helmholtz at this point. What is the goal, arrangement needs, problem or other circumstances affecting the topic at hand?

There really is no goal per say other than a coffee table chat on two channels with a shared volume control and tone stack, rather than the 'norm' two channels, two volumes, two tone stacks, or similar, type thing.

Just something I started wondering about is all, for the most part.

[tubemonster]

...... and what if you want to use both channels simultaneously?

Ya, that was the mixing part, but earlier in the thread I read "Will they be used simultaneously?".
I was like hmm, what if, and cranked out the newest picture to pop into this thread as a "not being used simultaneously" thing

Honestly, this whole thread I started was just to see what others thought, if someones done it, someone might know of amps that I'm not aware of that might be doing it, etc

[Chuck H]

Well of the circuits you've proposed there have been some with considerable signal losses due to voltage division and this last one with the switch avoiding any voltage division and additional losses. And you've reported that signal losses may not be a problem for you (or a given design at any rate). If all you want to do is understand channel mix circuits then I'd suggest perusing https://schematicheaven.net/ and look at all the two channel amps you can. You'll see an awful lot are done in pretty much the same way. And those that aren't should be interesting to you from a learning perspective.

One circuit not covered here yet and that you won't find an any guitar amps (apart from my own) is the 'virtual earth' mixer. It requires a triode and there is a phase inversion but it offers very low losses, is transparent and can have a fairly low impedance output. I used this circuit in one design to simultaneously mix two channels and balance the wet/dry for reverb recovery. Since an additional triode is generally necessary for reverb mixing anyway this solution was a win/win. It even occurred to me later that with it's lower output impedance the virtual earth mixer could be used in the way I just described AND as an effects loop send buffer simultaneously. That's a lot of bang for one triode. I'll post a diagram here in an edit in a bit. Hungry now so I have to make some pancakes and eggs.

[Helmholtz]

The "virtual earth" mixer is shown at the end of the page here: http://www.valvewizard.co.uk/resistive.html

It offers perfect source isolation but typically has a gain of only 1.

For circuit parameter see here:http://www.valvewizard.co.uk/localfeedback.htm

(These links have already been posted by tubemonster in post #14)

[Chuck H]

Ah. Ok. I hadn't opened every link posted. So it HAS been brought up. Good. Saves me some trouble. I didn't see it proposed in any of the schematics posted.

FWIW in the circuit I described above I'm using the VE to sum two channels with 470k input and shunt resistors (keeps them at unity gain) and implementing a 500k voltage divider after the reverb recovery. This is a dual gang pot so I then have a variable input resistance to the VE grid for the reverb only. So the 500k resistance to ground for the reverb at the VE grid remains constant but the input resistance can be zero allowing that triode to further amplify the reverb while not amplifying the dry signals. Worked very well.

EDIT: I just did some cad work and it looks like mixing the two channels, doing the reverb/dry balancing and adjusting the circuit to be an effects loop send buffer works just fine. Not sure I'll ever get a chance to implement this idea, but it's cool that works.

[Helmholtz]

I'm using the VE to sum two channels with 470k input and shunt resistors (keeps them at unity gain).

Not sure what you mean with shunt resistor.
As the VE grid has low AC impedance to ground (that why it's called a virtual earth node), a large value resistor from grid to ground should not significantly change the gain.
Gain is given by feedback resistor divided by input series (mixing) resistor + source resistance.