"Virtual earth" mixing amp with tubes

Just an idea I sketched during a coffee break. Comments invited.

would you be using a cap on the INV output as well? What would be your different cap value cutoffs for high passing on the outputs? I only ask because I assume since the Zs are different that would affect your coupling cap values right? I would also assume that since you absolutely have to use a cap on the NI side, you would want to use one on the INV side to balance the cap induced distortion too right?

You'd have to cap-couple either output; it's just not shown on the diagram.

The intention in showing two take-off points for the output wasn't so much to provide a balanced output (although you could use it for that, just keep the load impedances high and equal) as it was to provide a choice of signal polarity.

If the cathode output only is used, it wouldn't be a bad idea to bypass the plate to ground with a big cap to make it a true "grounded plate" (cathode follower) stage. Or the plate resistor could be eliminated, but all the resistances in the circuit would have to be adjusted in order to keep the direct coupling between stages.

The usual reason for driving a feedback network from a low impedance like a cathode follower is to avoid loading the output of the amplifying stage. However, your feedback net work is relatively high Z (100K) so this is not really necessary. You could just as well connect the feedback direct to the anode of the first stage.

Ian

Hmmmm... I have a question.

-Why does Joan Jett look like Freddie Mercury doing a Joan Jett impersonation?

Keef

[quote author="SSLtech"]...Why does Joan Jett look like Freddie Mercury...[/quote]
I was thinking the same thing, lol!

Going in the mixer meta...

You could use a transformer instead of a coupling cap on the output. And what about volume control for each input?

Lol!

Using a voltage amplifier and cathode follower in this manner was suggested by Charles Boegli in his article "The Anode Follower", which I believe appeared in Audio c. 1962 but I could be wrong. In essence, the circuit behaves like an inverting opamp circuit.

Ampex used something similar for the summing circuits in the MX-10 mixer; the summing amplifier itself was an EF86 pentode, while the cathode follower was a 12AU7. The feedback network was a bit complex in order to provide variable gain; one novel feature was that the pentode's screen grid was driven, not by a resistor from B+, but by a resistor from the cathode of the cathode follower. It's a very clean circuit.

Peace,
Paul

[quote author="pstamler"]
Ampex used something similar for the summing circuits in the MX-10 mixer; the summing amplifier itself was an EF86 pentode, while the cathode follower was a 12AU7. The feedback network was a bit complex in order to provide variable gain; one novel feature was that the pentode's screen grid was driven, not by a resistor from B+, but by a resistor from the cathode of the cathode follower. It's a very clean circuit.
[/quote]

That novelty long time ago was called "Ultralinear Operation".

Summing with an open loop gain of 10 seems a little weak. (If i read your notes correctly).

If you have to use tubes, its better to go passive into a non inverting high Z amp.

Just my opinion.

Why does Joan Jett look like Freddie Mercury doing a Joan Jett impersonation?

Click to expand...

HA!

Were Joan Jett and F. Mercury ever seen in the same place at the same time?

"I dunno man. I was loaded and nodded out" is not an acceptable response.

What about a transformer-coupled input, set up as "zero-field", i.e. virtual-ground-transformer? That way you'll easier interface to tube's optimum impedances - by using e.g. a 5+5:1 transformer; the "1" from inputs, the "5" to tube gate, and the other "5" from tube output.

I'm using that sorta setup in my G16 - http://www.gyraf.dk/tmp/G16.jpg

Jakob E.

[quote author="Kit"]"Summing with an open loop gain of 10 seems a little weak. (If i read your notes correctly)." [/quote]

No, you read it wrong. The open loop gain of a small signal pentode with a very high impedance load would be more on the order of 150x. The maximum number of inputs should be no more than 1/10th of this ("open loop gain / 10", as it says) to ensure that there is at least 20dB of feedback.

"If you have to use tubes, its better to go passive into a non inverting high Z amp."

Click to expand...

That's a sweeping statement. Would you like to give more detail?

Really, it depends on the number of inputs, how much isolation you need between them, and how much gain is readily available--among other things. There is no one way that's "best" in all circumstances.

[quote author="NewYorkDave"]The open loop gain of a small signal pentode with a very high impedance load would be more on the order of 150x. The maximum number of inputs should be no more than 1/10th of this ("open loop gain / 10", as it says) to ensure that there is at least 20dB of feedback.[/quote]
I don't understand this. The feedback does not depend on the number of inputs. However, the virtual earth impedance is approximately the feedback resistor over the open loop gain so to prevent interaction between inputs (and crosstalk) you would want this to be at least one tenth of the input arm which in your circuit amounts to the same thing as number of inputs no more than 1/10th the open loop gain.

Ian

Remember, the actual gain ("noise gain") of the circuit is the feedback resistor divided by the parallel value of all the input resistors. So, with 10 sources, the input resistance is effectively 10K and the noise gain is 10x (20dB). If the open-loop gain of the circuit is 150, the maximum number of sources that could be used, while maintaining a minimum of 20dB of feedback, is 15.

You made a point earlier about driving the feedback resistor directly from the plate. That could be done, but since the plate resistance of the pentode is very high, any additional shunting of it it would reduce the open-loop gain--which we want to be as high as possible.

[quote author="NewYorkDave"]Remember, the actual gain ("noise gain") of the circuit is the feedback resistor divided by the parallel value of all the input resistors..[/quote]
Of course, I forgot that :!:

You made a point earlier about driving the feedback resistor directly from the plate. That could be done, but since the plate resistance of the pentode is very high, any additional shunting of it it would reduce the open-loop gain--which we want to be as high as possible.

Click to expand...

Agreed.

Ian

[quote author="Wavebourn"][quote author="pstamler"]
Ampex used something similar for the summing circuits in the MX-10 mixer; the summing amplifier itself was an EF86 pentode, while the cathode follower was a 12AU7. The feedback network was a bit complex in order to provide variable gain; one novel feature was that the pentode's screen grid was driven, not by a resistor from B+, but by a resistor from the cathode of the cathode follower. It's a very clean circuit.
[/quote]

That novelty long time ago was called "Ultralinear Operation".[/quote]

Well, yes; it's novel to see it in a transformerless circuit, though.

Peace,
Paul

It has nothing to do with U.L. The screen grid is decoupled at AC, so the feedback is DC-only. It is clever, though, since it would tend to compensate for tube aging to some degree. Taking the follower's cathode return through the pentode shield connection is pretty slick, too--it prevents damage to the CF (or overheating of its cathode resistor) in the event the pentode dies or is removed with the mixer powered up. And then there's the method of coupling mixers together via mix amp cathodes... Yes, it's a clever circuit.

Schematic

Paul, that Boegli article appeared in Audio in December, 1960. It describes an active mixing circuit, but there's no illustration. It does include an illustration of an "amplifier of extremely low output resistance" that features a common-cathode direct-coupled to a CF with feedback to the grid of the first stage.