Tweaking cathodyne PI in Magnatone 260 for smoother drive

[jazbo8]

Looking at the OP's comments, it seems that the title of the thread might be mis-leading - yes it could be improved somewhat, but the PI isn't the place to look for "smoother overdrive"...

[dinkotom]

Yeah, it seems so.

[jazbo8]

PI:
Plate:209V
Grid (with 1M stopper):99V
Cathode:110V

There is a big difference in the grid voltage with and without the large grid stopper, neither offers center bias as mentioned. Try center bias the PI and let us know if you can hear a difference in the overdrive.

[dinkotom]

Well I would if I could Before I start reading Merlin's load lines from the beginning for the n-th time, I can do by-the-numbers job pretty well. What should I do to make it center biased? I guess I should finally get an O-scope and learn how to use it. Without the 1M stopper the grid voltage was 109V.

[tubeswell]

Well I would if I could Before I start reading Merlin's load lines from the beginning for the n-th time, I can do by-the-numbers job pretty well. What should I do to make it center biased? I guess I should finally get an O-scope and learn how to use it. Without the 1M stopper the grid voltage was 109V.

Purely as a suggestion:

Get the HT voltage as high as you can for the driving stage and the cathodyne.

Set the driving stage plate voltage at about 2/3 HT for that stage (100k plate resistor for a 12AX7 or 5751, or 22k-47k plate resistor for a 12AU7). Bias the driving stage at centre bias (about -2V for a 12AX7 or -4V to -6V for a 12AU7). Use an unbypassed cathode for higher fidelity.

Use AC coupling to the cathodyne, and cathode-bias the cathodyne with bootstrapping on the grid.

Set up the load resistors to give a cathode voltage of 1/3 HT and a plate voltage of 2/3 HT, (or 1/4 and 3/4 respectively).

[dinkotom]

Thanks Pete, all understood, wíll try that.

[martin manning]

The PI could be centered a little better, but it is not that far off, really. Notice it is running with its cathode and plate at ~1/3 and ~2/3 of the HT respectively now.

I think the stage driving it is compromised by its very high anode load and the need to get the anode voltage low enough for the DC coupled PI, and that may be the source of the unsatisfactory distortion tone you are getting

Magnatone went to an AC coupled PI on the 260a, which allowed them to change the bias point on the driving stage, but they kept the high anode load and the fixed bias arrangement. That would seem to be aimed at getting as large a signal as possible going into the PI.

I would change the PI to an AC coupled design, bias it at about 70V on the cathode (assuming 320V supply), and then play with the driving stage. I think a more conventional cathode biased stage with a moderately high anode load and appropriate Rk (say 220k and 3k3), and Rk bypassed with a 5-10uF cap, would be better.

[jazbo8]

+1, I think re-doing the PI's bias would not make that much of an improvement without re-configuring the previous stage and the coupling, which basically makes it into a 260A. May be that's a good thing...

[dinkotom]

Thanks for suggestions guys, much aqpreciated. I will try that.

[jazbo8]

The PI could be centered a little better, but it is not that far off, really. Notice it is running with its cathode and plate at ~1/3 and ~2/3 of the HT respectively now.

Taking a closer look at the 260A schematic, the grid voltage of the PI is 4.8V, is that a mis-print? Doesn't that cut off the 12AX7 completely? From your uTracer grid-leak bias chart, shouldn't the voltage be closer to 0? Or perhaps I am mis-reading the chart...

[martin manning]

It's either a typo, or perhaps more likely it's the result of a measurement where the meter's input impedance significantly altered the bias point.

From the grid leak bias chart, the grid voltage should be a bit below the cathode voltage, which is listed on the schematic as 105V. Connecting the grid to ground through a meter biases the triode into cut off (and the listed anode and cathode voltages are no longer valid).

[jazbo8]

It's either a typo, or perhaps more likely it's the result of a measurement where the meter's input impedance significantly altered the bias point.

From the grid leak bias chart, the grid voltage should be just a bit below the cathode voltage, which is listed on the schematic as 105V. Connecting the grid to ground through a meter biases the triode into cut off (and the listed anode and cathode voltages are no longer valid).

That makes more sense, and thanks for correcting my "close to 0" comment, by which I meant the grid voltage referenced to the cathode not to ground... Anyway, showing 4.8V is quite mis-leading. If one really want to measure the grid voltage, would a VTVM or DVM with very high input impedance work (connected between the grid and the cathode)?

[martin manning]

If one really want to measure the grid voltage, would a VTVM or DVM with very high input impedance work (connected between the grid and the cathode)?

The grid leak bias resistance would be reduced, from 10M to 5M say, assuming a 10M input impedance for the meter. That would still change the bias point slightly, but it would give a much more accurate result.

[teemuk]

First and foremost, with negative feedback and cathodyne PI you are not going to achieve that smooth overdrive in the first place:

Cathodyne inherently overdrives very harshly, and the global negative feedback will further on accentuate that characteristic. Grid stoppers on the PI will help some but they are more a bandaid for weak overdrive characteristics to begin with. Your best bet in "smoothing" out overdrive is actually getting rid of that global feedback loop.

Yeah, no grid stoppers on power tubes either. Whenever you overdrive the grids hard enough they begin to conduct and the PI will see almost a short circuit. It will clip harder to lower impedance loads. Additionally DC offset point will shift due to capacitive coupling and the bias will drift to a different level resulting to both crossover and blocking distortion. Lack of screen grid resistors to limit current accentuates all these characteristics.

Cathode biasing of the power tubes provides tad smoother clipping characteristics than fixed bias does. At least that's good. If you want to enhance that softness you also get rid of that power tube cathode bypass cap so that degenerative feedback via cathodes can compress the signal during peaks. This will naturally rob your headroom considerably. Be advised that not so many people are actually too fond of too "smooth" overdrive characteristics that such mod creates.

As is, there are no individual power tube screen resistors in this design that would allow generating a dynamically alterating AC signal at screens and therefore aid gain compression via power tube screen voltage. Not to mention, even protect the output tubes adequately from faults. I would probably experiment with adding these too.

And then we have 5881 power tubes, largely equivalent to 6L6. These are beam power tetrodes that - unlike true pentode tubes like EL34 - have much harder knees in their transfer characteristics that.... surprise, surprise.. also result to equivalently harder clipping.

Looks to me like pretty much everything in this power amp design aims towards opposite of soft clipping overdrive.

Ultimately you may wish to ponder if you want to convert a perfect Magnatone amp with its unique and classic characteristics to something entirely different and not that "classic" in the first place. When things come to that I would likely sell the Magnatone to someone who appreciates it -as is-, and then simply search for another amp more closer to my initial preferences. Basically, if I were shopping for an amp with great overdrive tones a vintage Magnatone wouldn't be my first, second or even hundreth option to try.

[Littlewyan]

Well.......he has a point.

Make a decision, keep the amp and mod it or sell it and buy another. Bear in mind if you do mod it you don't necessarily have to carry out every mod suggested. Although personally I would add screen grid resistors, control grid stoppers and tweak the PI as Merlin suggests in his book.

My £0.02