SE 5881 power section, normal variation or is something wrong?

[CascoSieg]

I'm breadboarding a SE stereo power section for a possible build. Have been planning around and testing with 5881s, each driven by a dedicated 12AX7 driver. I believe my current circuit puts the 5881s at nearly full plate dissipation, and a fairly neutral bias point.  I've been testing using a guitar boosted to roughly line level input, but they - that is both channels - don't sound right: highly compressed attacks and unnatural sounding note decay at all volumes, and fizzy break up tone when pushed, but it never really gets loud, or what I think should be the right volume for these tubes. Values for both channels are similar, within <1% of each other:
SS rectified B+ = 325v
Cathode R = 220
V(plate) = 312v
V(screen) = 260 (using 22k , <edit>SCREEN resistors to lower the voltage)
V(cathode) = 14.2

Lowering the level of the input signal doesn't really change things tonally. I have read here that big screen resistors induce compression, but only when grid is near clipping or cutoff, which I don't think should be case here at half or less volume, which is pretty quiet. When I swap in a pair of KT66s, keeping everything else the same, I get a nice clean loud output as expected, good dynamic range and sensitivity, and just slightly detectable compression. 

First question: With these values, am I wrong to expect the 5881s to provide a good dynamic range with clean tones, and no significant distortion until it is given a big signal?

The KT66's show slightly lower plate voltages (302v and 305v) and higher cathode voltages (19v and 17v) but differ greatly in their screen voltages (287 and 262).

Second question: Does this variation between measured screen voltages of the KT66 tubes fall within what might be considered normal healthy variation, or is one of them defective/substandard? If the former is true, and I want to use both tubes, I suppose all I can do is just fiddle with the individual screen resistors to get both tubes in same general operating range, yes?

I'm on a steep learning curve, so any and all help is appreciated!

[shooter]

using 22k grid resistors

try 2.2k, then 200 ohms
It's a guitar amp, not a fine wine music box


Didn't look to see if you can get a bigger B+ on the PA tubes, that might help some.
just tweak the cathode R to keep the tubes somewhere close to, sweaty (I like between 85-105% dissipation)

[DummyLoad]

V(screen) = 260 (using 22k grid resistors to lower the voltage)

why? 500R to 1K are adequate there. screen voltage to 310V not an issue.

why the shared cathode resistor for the drivers (V3) if it's a stereo amp?

why a 2K pot for bias adj. if all you need is 200Ω +/-? wouldn't a 200Ω-250Ω be a better choice? e.g. about 90% of the wiper rotation of a 2K pot is totally useless.

respectfully,   

--pete

[CascoSieg]

Shooter: Thanks for the suggestion - I'll plan to experiment in that direction, and see how things look and sound. 
I'd like to understand it a little better though.  Maybe I'm too hung up on the load line plot, but it looks to me like when the plate v and screen v get closer together, say 312v/300v, and I adjust the cathode R to keep dissipation near 90-100% (around 300 ohms), my load line intersects way below the knee of the 0v grid contour on the graph. Maybe that's not a problem if I keep my grid voltage from ever reaching 0.  ... ?
I've got a 300v PT secondary tap that should give me  ~380v B+, so I can always move into that testing scenario, but again, I'd like to better understand how the calcs + what I'm hearing would point me in that direction.

Dummyload: The screen R choice came from the load line plot. Maybe that was just the wrong approach, but I can't see how to get a good line with the screen v up at 300.
Shared V3 cathode just for efficiency - it doesn't seem to cause any problems, like signal bleed through - but easy to separate I suppose. Do you suppose that is causing me problems I can't hear?
A lower bias adjust pot would be better , but I couldn't find a multi-turn one at 2W rating, and I can always make a 2k pot a 200 pot.

Thank-you both!!!

[pdf64]

There will be a significant degree of signal level dependent negative feedback across those massive screen grids.
Any results / tonal impression you got with those are invalid.
Plus it seems unrealistic to expect valves to have matching g2 current draw.
I suspect that with 470 - 1k fitted, it will knock your socks off

Why the weird vol control arrangement?

[CascoSieg]

Ok great! Screen grids will be changed and I'll go from there.
Re: volume controls: I didn't know they were weird    but they work OK. The placement promotes isolation of R/L signals. I could easily/equally put them after V3 if it was advantageous... am I to infer this would be less weird?
:)
Thanks!

[edit]  I should add that this FX return circuit will support the pass-through of a single un-effected signal to both return triodes, OR if you have something in one return jacks, it will pass unaffected signal one way and fx signal the other, OR if both returns are filled, will send those two ways, keeping things separated each case one to each triode.
[edit] Also, I don't know what my return signal voltage will be - line level or stomp, could be either, so it seems some control over signal into the V3 would be good, and this arrangement keeps R/L separate (on either side of the 220ks).

[SILVERGUN]

You are attenuating the input signal before the first stage pretty significantly.
A simple fix would be to move the 100K pot tap to the other side of the 220K resistor and see how that sounds.
A single 12AX7 trying to drive a 5881 doesn't seem like a recipe for a great sounding power amp, dependent on the input signal of course, but it seems odd to start your power amp design off with a 3.2:1 voltage divider at full blast. 

[PRR]

> promotes isolation of R/L signals.

The common cathode resistor and small capacitor for L and R of V3 is sure to cause bass crosstalk and queer stereo.

[CascoSieg]

Well that's a good point, Silvergun.  My assumption is that the output from V3 shouldn't put more than 0 volts onto the grid - that max peak signal voltage should be about the same as voltage drop across cathode. In previous configuration that was around 14v. A 1 volt (fx return) peak signal going in to V3 ends up as about a 14v signal on the grid (unless I'm doing the math wrong).
Just now, after putting in the 2.2k screen Rs suggested by shooter, with the 5881s I have:

SS rectified B+ = 334v
Cathode R = 330
V(plate) = 322v
V(at screen) = 322v (using 2.2k screen R)
V(cathode) = 20.5
V(across screen R) - 2.3v

So now my bias is much different that before and thing sounds great - clean, bright, loud.
Am I wrong to be designing things based on the idea of not exceeding 0V on the power tube grids?
And, that further begs the earlier question about whether it matters that my load line doesn't cross anywhere the knee of the 0 volt contour on the load line plot.

And I'll separate the cathode RC too.

So grateful for your help!
~Sieg

[shooter]

EDIT:
0V, yes if it's 0Vdc, otherwise;

Am I wrong to be designing things based on the idea of not exceeding 0V on the power tube grids

if music, no, if guitar, yep!


here's my guessing start point;
take the cathode VDC (20.5)
I try and have 1.5X that value in VACrms available on the grid
that puts you into the cutoff/clip zone, your volume determines how much

[CascoSieg]

Eureka!  That's huge information, changes a lot of things. Thanks!   
So, about the 0V contour line on load line plot, and crossing above the knee: relevant only when overdriving the tube? Or  is it even a design concern for mostly clean, possibly slightly crunchy territory?
Thanks again, all!
~Sieg

[shooter]

0V

use either 0Vdc or 0Vac, i get confused easy


SE typically "self regulates" so the louder you play the cooler the tube runs
I like enough signal (Vac) to slam the grid, hence your Cathode Vdc = ~~~ max signal Vacrms to hit clip/cutoff.


The smart kids can take you in the deep weeds, if you want