Traditional Grounded Plate/Grounded Grid stage and Sulzer's "Cath-code" input stage

[SoulFetish]

I was hoping to get some thoughts on designing an input consisting of a more conventional cathode follower driving a grounded grid stage compared to Mike Sulzer's cathode coupled cascode, seen here:

http://music-electronics-forum.com/t41822/
(c'mon, admit it, "cathcode" is kind of a badass name for that circuit).

Here is a simple schematic drawing showing a conventional set up and Mike's input stage (ignore the lack of values in the first drawing, it is more to show the basic configuration).


Here are a couple of thoughts on Mike's design (the second circuit in the drawing above), and some of the things I like about it (please let me know if I'm incorrect in any of the following)...
Starting from the top down, a quick look at this design and one could almost mistake it for a Long Tail Pair with two 100K load resistors on each plate. However, the first triode is operating as a grounded plate stage, so the 100k resistor can actually be thought of as a B+ "dropping" resistor with the 10uF cap (which AC grounds the plate) as a filter capacitor. The 100K resistor on the second triode is acting as a traditional load resistor. But, here is the interesting thing–Because the both triodes have a grid bias of 0V ref to ground, and cathodes are operating at the same potential (and share the "long tail" resistor), the idle current is the same for both tubes. This means that, both triodes are operating with the same plate voltage at quiescence.
Another interesting thing happens under signal conditions. It looks to me (correct me if I'm wrong) like this is a completely non-inverting gain stage. As the grid voltage swings positive in the first triode, the current increases across the tail resistor and the voltage at the cathode "follows" the grid. Because the cathode of the second triode is directly coupled to first, the cathode voltage of the second triode also increases (and decreases) with the cathode of first triode. The second triode also operates as a grounded grid stage with the grid is held at a fixed voltage. So, in the second triode, as the cathode voltage swings positive, current decreases across the second triode's 100K load resistor (because the grid becomes more negative in reference to the cathode voltage). Likewise, when the cathode voltage swings negative, current increases across the 100K load resistor (because the grid voltage becomes more positive in reference to the cathode voltage). So, by gounding the grid of the second triode (keeping the input held at a fixed voltage), this circuit becomes fully differential, non-inverting gain stage.
Please tell me if I understand this correctly. If I do, I really like this configuration because it allows for DC coupling and solves some biasing problems one might encounter in a traditional grounded grid stage. But, what are some of the benefits and drawbacks of the first circuit vs the second in the drawing I posted. Also, what how do I calculate the input impedance in a traditional grounded grid stage?
I appreciate any thoughts. Thanks.

[Mike Sulzer]

Hi,

I like your explanation. One thing I wanted to do in this high gain amp was to stop as much feedback to the input as possible. From later stages, you have to use good layout and shielding. From the output of first stage itself, you select a circuit geometry that minimizes it. (Now maybe I do not really need to do that, but considering that it is the total feedback from all stages that counts, it might. I will try to find out with the next amp.) Why this particular little used cascode, rather than the usual one? Well, I was already going to have a negative supply in order to eliminate blocking distortion, so why not find ways to use it to make circuits simpler, or more elegant, or at least more useful?

The differential pair with two supplies is an elegant circuit requiring no fussy biassing; so why not alter it to do this?

When you look into the cathode of a grounded grid circuit, the current-voltage relationship is described by the transconductance. Take the inverse of that and you have an impedance describing the ratio of input voltage to input current; that is the input impedance.

I was hoping to get some thoughts on designing an input consisting of a more conventional cathode follower driving a grounded grid stage compared to Mike Sulzer's cathode coupled cascode, seen here:

http://music-electronics-forum.com/t41822/
(c'mon, admit it, "cathcode" is kind of a badass name for that circuit).

Here is a simple schematic drawing showing a conventional set up and Mike's input stage (ignore the lack of values in the first drawing, it is more to show the basic configuration).
[ATTACH=CONFIG]40849[/ATTACH]

Here are a couple of thoughts on Mike's design (the second circuit in the drawing above), and some of the things I like about it (please let me know if I'm incorrect in any of the following)...
Starting from the top down, a quick look at this design and one could almost mistake it for a Long Tail Pair with two 100K load resistors on each plate. However, the first triode is operating as a grounded plate stage, so the 100k resistor can actually be thought of as a B+ "dropping" resistor with the 10uF cap (which AC grounds the plate) as a filter capacitor. The 100K resistor on the second triode is acting as a traditional load resistor. But, here is the interesting thing–Because the both triodes have a grid bias of 0V ref to ground, and cathodes are operating at the same potential (and share the "long tail" resistor), the idle current is the same for both tubes. This means that, both triodes are operating with the same plate voltage at quiescence.
Another interesting thing happens under signal conditions. It looks to me (correct me if I'm wrong) like this is a completely non-inverting gain stage. As the grid voltage swings positive in the first triode, the current increases across the tail resistor and the voltage at the cathode "follows" the grid. Because the cathode of the second triode is directly coupled to first, the cathode voltage of the second triode also increases (and decreases) with the cathode of first triode. The second triode also operates as a grounded grid stage with the grid is held at a fixed voltage. So, in the second triode, as the cathode voltage swings positive, current decreases across the second triode's 100K load resistor (because the grid becomes more negative in reference to the cathode voltage). Likewise, when the cathode voltage swings negative, current increases across the 100K load resistor (because the grid voltage becomes more positive in reference to the cathode voltage). So, by gounding the grid of the second triode (keeping the input held at a fixed voltage), this circuit becomes fully differential, non-inverting gain stage.
Please tell me if I understand this correctly. If I do, I really like this configuration because it allows for DC coupling and solves some biasing problems one might encounter in a traditional grounded grid stage. But, what are some of the benefits and drawbacks of the first circuit vs the second in the drawing I posted. Also, what how do I calculate the input impedance in a traditional grounded grid stage?
I appreciate any thoughts. Thanks.

[SoulFetish]

Also, I was wondering if you could explain what is meant by " instability of Miller capacitance"? Often, we design around miller capacitance to include it in setting the rolloff frequency of a LPF.
Thanks.

[KevinOConnor]

Hi Guys

Soulfetish: Both circuits in post-1 are noninverting. In the separated stages, the coupling cap might have to be quite high value to assure good bass response, but it does give you freedom of voicing in one more circuit position. There are more components overall and this might effect your decision as to which circuit to go for. The separated stages also gives the option to bias the sections differently, whether there is benefit is another story.

The notion that the split rails offer an advantage because the grid is ground-referenced I believe makes no difference compared to the standard CC stage. In the CC, the grid is tied to ground through a resistor just as it is for the split-rail-powered tube. I don't see a difference. In both cases, the grid ends up at ground potential and the cathode sits positive with respect to the grid.

The split rail has an advantage inasmuch as its operating conditions can be set quite freely using two Rs. Rk sets the current and Ra sets the signal swing centre for the output. There is an interdependence just as in the CC case but it "feels" much more independent. With very high negative rails, high Rk values are needed and thee behave much more like a constant current source than a low Rk with a low or absent V- (ground instead as in CC).

Have fun

[Merlinb]

I really like this configuration because it allows for DC coupling and solves some biasing problems one might encounter in a traditional grounded grid stage. But, what are some of the benefits and drawbacks of the first circuit vs the second in the drawing I posted.

Both are types of cathode-coupled amp. The input valve is a cathode follower and hence has the input inpedance of a cathode follower, i.e. very high input resistance and very low Miller capacitance. But you pay for this with higher equivalent input noise, higher distortion from driving the low-Z grounded-grid stage (not that distortion is a problem in a guitar amp), and an extra tube of course.
Non-inverting is a non-issue in a guitar amp; you could just as well have a cathode follower driving a regular gain stage. In other words, the overarching feature of these circuits is the cathode follower input stage; the grounded-grid stage is just a novelty. If you like novelty, knock yourself out.

Also, I was wondering if you could explain what is meant by " instability of Miller capacitance"?

Probably it refers to the way Miller capacitance varies with tube sample and with ageing. But a cathode follower has near unity gain throughout its life, so its input capacitance varies very little, and is very small in any case.

[Mike Sulzer]

Both are types of cathode-coupled amp. The input valve is a cathode follower and hence has the input inpedance of a cathode follower, i.e. very high input resistance and very low Miller capacitance. But you pay for this with higher equivalent input noise, higher distortion from driving the low-Z grounded-grid stage (not that distortion is a problem in a guitar amp), and an extra tube of course.
Non-inverting is a non-issue in a guitar amp; you could just as well have a cathode follower driving a regular gain stage. In other words, the overarching feature of these circuits is the cathode follower input stage; the grounded-grid stage is just a novelty. If you like novelty, knock yourself out.


Probably it refers to the way Miller capacitance varies with tube sample and with ageing. But a cathode follower has near unity gain throughout its life, so its input capacitance varies very little, and is very small in any case.

I see no reason why you cannot analyze my circuit as a cathode follower into a grounded grid, but you will have to work a bit to get the right answers. it is easier to analyze the circuit directly, maybe something like this if I am not making silly mistakes: You can see by inspection that the sum of the two cathode currents stays the same, when input is applied, to the level of accuracy allowed by the cathodes resistor to the negative supply. Thus a signal on the left hand grid wrt ground must alter the cathode voltage by almost exactly half of its value. The input impedance is thus a C of .5Cgk in parallel with Cgp with no Miller enhancement. Gain to the right hand plate is of course half what you would expect from a grounded grid stage driven by a low source impedance. Since one cathode increases in current while the other decreases, non linearity should drop , but this decrease in distortion over a grounded grid stage driven from a low source impedance is probably of no significance in this application.

Noise certainly goes up, but this has been fully analyzed in the main discussion with measurements showing that you get what is expected..

[SoulFetish]

If you like novelty, knock yourself out.

.

I do like novelty.
Although, I suspect at frequencies higher than I'm concerned with, there is more utility for a grounded grid.
I really just want to experiment with different input stages.
Another thought i had was using a bootstrapped Mu Follower input stage (originally, i was going to use this as a buffer and gain stage for acoustics) But as an input stage i think offers some interesting things. I suppose you have to deal with a gain less than mu, but I can live with that. Plus you still get the benefit of a very high input impedance and very low Output impedance in order to drive a tone stack or other heavy load.