YADTMI - yet another dual triode mixing idea

[Helmholtz]

I don't have a pentode circuit drawn up. Here's the frequency response of a 12AX7 unity gain stage with 500k feedback and load resistors and a 500p cap. It's -3dB at 20Hz.

[ATTACH=CONFIG]56358[/ATTACH]

Thanks, this confirms my estimation. I assumed a somewhat lower total load resistance as the circuit will be connected to something.

[SoulFetish]

I don't have a pentode circuit drawn up. Here's the frequency response of a 12AX7 unity gain stage with 500k feedback and load resistors and a 500p cap. It's -3dB at 20Hz.

[ATTACH=CONFIG]56358[/ATTACH]

thanks for posting that sim. That's what I appreciate about you, man.
So, I gotta run and go help out a buddy, but here are a couple of links I used as a resources in calculating values. I'll hit you guys up later about this. To be cont.....
Designing Single-Stage Inverting Feedback Amplifiers
Triodes with Local Negative Feedback

[Helmholtz]

I think a good case can be made that it's much simpler and more predictable to set the Low frequency rolloff at the input once the input impedance is calculated. Because it is in the feedback loop, it should not be used to set the corner frequency.

I agree that it is probably no good idea to make the coupling cap inside the feedback loop too small. My concerns would by frequency dependent phase shift and stability.
But I have no idea how to calculate the impedance at the "virtual zero" point - other that it's supposed to be "low". Maybe something like feedback resistor divided by open loop gain (would be consistent with the Miller theorem) and shunted by total source impedance? Need to look this up in my opamp books.

OTOH, the input impedance as seen from the sources (means before the mixing resistors) varies with your mixing pot setting.

[Helmholtz]

Designing Single-Stage Inverting Feedback Amplifiers
Triodes with Local Negative Feedback

Having seen these sources earlier would have saved me time.

[SoulFetish]

Having seen these sources earlier would have saved me time.

sorry about that

[Helmholtz]

As the 2 different formulae for the input impedance used by Merlin in his book and his web article both seem to contain errors, I derived my own formula using Kuehnel's methods.

In a convenient form I get: Zin = R1 + (Rf + ra)/(1 + Go), with

R1: input mixing resistor
Rf: feedback resistor
ra: open loop output impedance
Go: open loop gain (absolute value)

The formula is consistent with the results of Aiken. He uses -A instead of Go.

Setting the mixing resistor (R1) to zero gives the impedance of the virtual ground point as Z = (Rf + ra)/(1 + Go).

[SoulFetish]

As the 2 different formulae for the input impedance used by Merlin in his book and his web article both seem to contain errors, I derived my own formula using Kuehnel's methods.

In a convenient form I get: Zin = R1 + (Rf + ra)/(1 + Go), with

R1: input mixing resistor
Rf: feedback resistor
ra: open loop output impedance
Go: open loop gain (absolute value)

The formula is consistent with the results of Aiken. He uses -A instead of Go.

Setting the mixing resistor (R1) to zero gives the impedance of the virtual ground point as Z = (Rf + ra)/(1 + Go).

Thanks for looking into that
Where did you find Kuehnel's methods for calculating anode follower circuits? I like Aiken's white papers, because his analysis is pretty thorough. He provides enough derivations of the equations to make my head start to swim, but I think that the discipline of going through the exercises and trying to understanding the math is important for me.

The miniature blue accutronics spring reverb arrived, and I have all the parts to start building (or testing anyway). I'm trying to iron out a couple of things in the circuit, maybe you could offer some advice.
I added another small gain stage to the recovery amp, which was also a convenient place to add a reverb level control. I think you made a good point, in that Ri can be adjusted easy enough to factor in the driving stage output impedance.
I was thinking about incorporating a "dry" signal volume control, but the dry signal is tapped off at the input of the reverb driving stage. I didn't want to change the loading of the reverb driver input, so I was thinking about this (see drawing below).



My question is:, how will different level settings affect the virtual ground input circuit (if much at all)? Is this a viable way of doing this? Or would a overall volume control at the output of final mixing stage be the way to go?

[Helmholtz]

Where did you find Kuehnel's methods for calculating anode follower circuits?

No, it's not in my books.

What I meant was, I used his methods to analyze the circuit for its input impedance.

So I drew up the equivalent AC circuit replacing the tube with a (grid voltage controlled) voltage source (Va = - Go*Vg) in series with its open loop output impedance ra.
As the equivalent circuit only has one signal current path, there is only one current: the input current Iin.
Applying Kirchhoff's Voltage Law (KVL) and Ohm's Law yields Iin = (Vin - Va)/Rtot with Rtot = R1 + Rf + ra.
The rest (eliminating Va and Vg by using Ohm's Law and calculating Zin = Vin/Iin) is straightforward.

Yes, Aiken is very thorough. But to derive his formulas he uses general signal and system theory, which I am not familiar with.


Regarding your modified circuit:

I wouldn't like to have 3 controls that all change the mix ratio (making the "mix" control redundant). But maybe a "drive" pot controlling the 6943 reverb driver grid voltage would make sense, especially if there is any chance for grid conduction.

[SoulFetish]

Regarding your modified circuit:

I wouldn't like to have 3 controls that all change the mix ratio (making the "mix" control redundant). But maybe a "drive" pot controlling the 6943 reverb driver grid voltage would make sense, especially if there is any chance for grid conduction.

Hmmm.. that's fair criticism. You're right, each of those controls essentially do the same thing that the mix pot did. I liked the idea of the mix control at the virtual earth point earlier, but I think that having a traditional reverb level control which allows you to dial the reverb off is an important feature.
But there is a case to be made that controlling the drive signal could be important so as not to overload the drive amp. Placing the reverb control in the send is annoying and brings it's own problems.
So here's my solution: I streamlined the interface and kept the reverb control in the recovery amplifier, but I put in a "threshold" control at the drive input with a LED/diode clipping indicator (I should have included a resistor to set the minimum level, so that both controls can't attenuate the signal completely.)
In the next iteration, a more elegant solution might be a soft limiter circuit or jfet AGC control in it's place.
Here's the idea:




for those who may dig 3 controls, it may look something like this:

[Helmholtz]

As soon as the limiting diodes (or the 6943 grid) conduct you will have distortion across the threshold pot and thus in your dry signal.
The dry signal needs to be taken directly from the rel. low impedance output of the first stage.


Is this going to be a stand-alone reverb? How/where is it going to be used? What input/output levels?

[Helmholtz]

I don't think the limiter diodes make much sense, as there is no need to protect the tube.
Rather they will lower the threshold of distortion at the grid.

[SoulFetish]

I don't think the limiter diodes make much sense, as there is no need to protect the tube.
Rather they will lower the threshold of distortion at the grid.

Well, The aren’t really there as limiting diodes, they would be there as led clipping indicators to let the user adjust the driving input signal below the threshold

[Helmholtz]

Well, The aren’t really there as limiting diodes, they would be there as led clipping indicators to let the user adjust the driving input signal below the threshold

This LED arrangement will distort the grid signal long before any visible light emission. Don't expect to see much with LED currents of 10ľA or lower.

[SoulFetish]

This LED arrangement will distort the grid signal long before any visible light emission. Don't expect to see much with LED currents of 10ľA or lower.

Ahhhh. Damn, you're right. Hadn't thought about that. This is not a good peak signal indicator.

Just thinking, though... Perhaps there is a better way by leveraging something inherently unique to pentodes and tetrodes.
According to the datasheet (at V_g2 100V, close enough for this conversation), screen current increases to about 6mA, as V_g1 reaches 0V for a 20k plate load.
What if I have a resistor/LED parallel circuit, where the voltage drop across the resistor reaches the ≈ 1.5V turn on voltage for the LED when the I_g2 current reaches 6mA? like this:
(the red LEDs I have are extremely linear over their operating range and are rated at a continuous current upwards of 100mA)


(edit: corrected mistake in location of diode/resistor pair.)

This way, no diode clipping and the grid circuit can operate without any other influence. What do you think?

ps. easy enough to test; scope the signal at the grid and the voltage across a sense resistor at the screen to confirm a more precise voltage/current relationship. Then accordingly, choose the appropriate resistor value. But, it's so simple it just might work!

[Helmholtz]

the red LEDs I have are extremely linear over their operating range

Linear in respect of what? Light emission vs current or current vs voltage?

scope the voltage across a sense resistor at the screen

Will require scope with differential mode for "floating" measurement.

What if I have a resistor/LED parallel circuit, where the voltage drop across the resistor reaches the ≈ 1.5V turn on voltage for the LED when the Ig2 current reaches 6mA?

Might work. Results will depend on how the g2 current is shared between resistor and LED.