Pentode PI

[tubeswell]

Trying to get my head around this: (The PI design appears in RDHB4 fig 12.29C, altho' I couldn't find any explanation of the circuit there)



This is surely a case where the screen is unbypassed, so the Vg2 will be bouncing up and down in phase with the Va. BUT is it not the case (as with any pentode), that prevailing the screen current is in opposing phase to the prevailing plate current (albeit you get the 'neutralising' effect of screen compression from the lack of g2 bypassing)?

Surely the screen is still at some +ve voltage that acts as an electron particle accelerator, even when the plate current is lowered

How do the internal screen resistance (rg2) and the screen resistor (Rg2) have an effect on the phasing of the screen current?

Only I am querying this elsewhere and there seems to be diverging opinions on how it all works.

Looking forward to some debate on this.

Pete

[jazbo8]

It looks to be a very un-balanced "splitter", as you pointed out, the plate & screen voltages are in phase but with large difference in their outputs, i.e., Ap ~ gm * rp and As ~ gm12 * rs (ignoring Rp, Rs, Rk and RL for the moment). The difference is then amplified by the output section, which results in pretty high distortion.

[tubeswell]

Here's a link to a 6AC7 datasheet BTW http://tubedata.milbert.com/sheets/093/6/6AC7.pdf

[overtone]

RDH4 references Peter Sulzer 1948. I found a very bad quality pdf of his '48 article which to me is illegible except for the sub-headline explanation - see screen shot below.
Here is the link to the article, scroll down to page 10 - 11:
http://www.americanradiohistory.com/...8%20August.pdf

best,
tony

[tubeswell]

Thanks for the article info Tony. Using the author's name etc I found this link to a clearer version (on a side note, check out the ad for JBL D130 on page 37 (65 years ago this month in 1948, the year they were introduced!) http://www.americanradiohistory.com/...8%20August.pdf

[jazbo8]

Thanks for the link, read it and understood the logic, but unfortunately I am not able to verify the circuit with simulation, perhaps it's my model (I used EF86 instead of the the 6AC7), if anyone has a working simulation, please let me know.

[Lauri]

Thanks for the link, read it and understood the logic, but unfortunately I am not able to verify the circuit with simulation, perhaps it's my model (I used EF86 instead of the the 6AC7), if anyone has a working simulation, please let me know.

To simulate that circuit you would need a pentode spice model with a supressor grid. I'm not sure if such a thing exists at least not anything that simulates the supressor grid correctly since spice models don't even simulate screen grid properly.

[jazbo8]

I have done some simulations (with 6SJ7 which worked a lot better than the EF86), you can see the results at Hoffman´s forum, but have no interest in actually building one - can´t see much benefit over the conventional triode version.

[Lauri]

Here's my explanation how this circuit works.


When you put a negative voltage to the suppressor grid then control grid to plate transconductance is is lowered and control grid to screen transconductance is increased. The more negative the suppressor grid is the lower the control grid to plate transconductance is. In this circuit the suppressor grid is so negative that the control grid to plate transconductance is close to 0, which means the control grid isn't controlling the plate at all. Instead it's controlling the screen grid which now acts as a plate for the control grid. Also now the screen grid is controlling the plate and that's why the output from the plate is in phase with control grid.

Even though this circuit works as a phase splitter it's not a very good one. Output impedance from the plate is much higher than from the screen. Also the output signal from the plate is most likely going to be very distorted.

[tubeswell]

Even though this circuit works as a phase splitter it's not a very good one.

but can be made to work - see post number 67 in this thread unusual phase invertor? - Hoffman Ampifiers Guitar Tube amp forum

These scope shots are @5V/division using a 6SJ7

Plate (top) and grid


Screen (top) and grid


I take it you read the Sulzer article?

[Lauri]

Yes I read the article but it didn't explain very well why output from the plate is in phase with the control grid. Or at least it took me several times to read it before I understood it.

From those scope pictures it looks like there's not much gain in that circuit. In the article it said the gain is about 30 and distortion is low but maybe 6AC7 is much better as a phase splitter than 6SJ7. It would be interesting to see what the maximum output voltage is before it starts to clip and which output clips first plate or screen. My guess is the output from plate will clip first if Rg2 is half the value of Ra like in the article but if Rg2 and Ra are equal like in the post in Hoffman amps forum then the output from screen will clip first. Also would be interesting to know output impedances from plate and screen. I originally thought output impedance from plate would be much higher than from screen but negative voltage on suppressor grid might actually lower plate resistance.

[jazbo8]

FYI, I could not get the simulation to work based on the Sulzer article, while SPICE simulations seldom match the real circuits exactly due to variations in tube tolerances, SPICE model accuracy, wiring/layout, etc. but it does give a very good indication whether the circuit works or not. In the case of the Sulzer circuit with 6AC7, I am not convinced that it works until someone actually builds one and show me otherwise.

[Lauri]

FYI, I could not get the simulation to work based on the Sulzer article, while SPICE simulations seldom match the real circuits exactly due to variations in tube tolerances, SPICE model accuracy, wiring/layout, etc. It does give a very good indication whether the circuit works or not. In the case of the Sulzer circuit with 6AC7, I am not sure it works until someone actually builds one and prove it otherwise.

Do you have spice model with supressor grid for 6SJ7 or 6AC7? Where can I get them? I've never been able to find any pentode models with separate pin for supressor grid.

[jazbo8]

PM to you, did you get them? BTW, this is the result using the 6SJ7 model:

[Merlinb]

Trying to get my head around this:

How do the internal screen resistance (rg2) and the screen resistor (Rg2) have an effect on the phasing of the screen current?

Only I am querying this elsewhere and there seems to be diverging opinions on how it all works.

This circuit would make more sense if the cathode was unbypassed.

I built a funny ring mod / sitar effect type thing using this same principle:
AX84.com - The Cooperative Tube Guitar Amp Project

Basically, the suppressor determines the proportion of cathode current that splits off to go to the screen. If you were to apply opposing signals to the control grid and suppressor grid then, with a bit of fiddling, you could get the proportion of screen current to increase just when the anode current is decreasing. hence out-of-phase output signals.

In practice it is easier to fix the suppressor voltage at ground and allow the cathode voltage to move instead, since an increasing cathode voltage is the same thing as a decreasing supressor voltage, and vice versa.