blackface PI cap question

[uneumann]

Ok...
I assume if the value is too small that LF would suffer due to the non inverting triode grid creating local NFB. Suffer being relative since I also notice that some designs that use both LTP grids as inputs omit the bypass cap altogether. Ok, so what of too large a value? Is there a time constant consequence? And how is the bypass cap value interacting with any global NFB? It looks to me that the bypass cap value would be somewhat relative to the global NFB circuit impedance, like the presence control cap value. Yet the BF type designs (relatively low impedance NFB loop) and the Bassman/Marshall type designs (relatively high impedance NFB loop) both use the .1uf LTP bypass cap value.?. Why is .1uf "simply the value that IS used" by the entire industry like a mindless congregation genuflecting?

Chuck - in writing about the PI stage, I used 10nf caps on both grids - inspired by some Marshall 18w drawings.

For example... http://mhuss.com/18watt/schematics/18wattLite.gif

The behavior at low signal levels (up to ~2v peak) just depends on the AC coupling response. As you point out, LF suffers as either cap decreases. tubeswell's comment is on target - LF response is cleaner and more "liquid" sounding as the 2nd grid cap increases. Try swapping the 100n for a 10n in a Bassman - it becomes a completely different amp.

At bigger signals (~4v peak) you get grid current and grid shifts. I measured about 7v shifts on both grids. The time constants for the attack/decay will vary with different cap values. With 10n on both grids, the shifts track well so they are common mode and have little impact on the output. As OD gets more severe, (10-20v) the shifts become more dissimilar (no longer common mode) so the output starts to show blocking - although relatively minor with two 10nf caps.

A similar experiment with the "standard" 100n cap on grid 2 is illustrated in the Bassman chapter - though grid 2 now has NFB. It does show that dissimilar caps produce far more shift differences, and some "ringing" blocking effects due to the dynamics of the shift decays. I didn't try bigger caps as tubeswell suggests, but I imagine the shift/blocking effects will only worsen so there's a limit there.

The source impedance feeding the second grid cap - be it a hard wire ground or a lo-z or hi-z NFB source have minor impacts during OD. Think of these impedances as similar to adding a grid stopper to grid 2. My experiments showed a grid stopper on either grid has very minor impacts.

I'm left with no sense of right/wrong values for these caps, but .1u simply seems a good compromise for full LF clean tone and acceptable blocking behavior with OD. Even when mindlessly copied, .1u in the ballpark for what people consider good - so few designs bother with variations.

[Chuck H]

Chuck - in writing about the PI stage, I used 10nf caps on both grids - inspired by some Marshall 18w drawings.

For example... http://mhuss.com/18watt/schematics/18wattLite.gif

The behavior at low signal levels (up to ~2v peak) just depends on the AC coupling response. As you point out, LF suffers as either cap decreases. tubeswell's comment is on target - LF response is cleaner and more "liquid" sounding as the 2nd grid cap increases. Try swapping the 100n for a 10n in a Bassman - it becomes a completely different amp.

At bigger signals (~4v peak) you get grid current and grid shifts. I measured about 7v shifts on both grids. The time constants for the attack/decay will vary with different cap values. With 10n on both grids, the shifts track well so they are common mode and have little impact on the output. As OD gets more severe, (10-20v) the shifts become more dissimilar (no longer common mode) so the output starts to show blocking - although relatively minor with two 10nf caps.

A similar experiment with the "standard" 100n cap on grid 2 is illustrated in the Bassman chapter - though grid 2 now has NFB. It does show that dissimilar caps produce far more shift differences, and some "ringing" blocking effects due to the dynamics of the shift decays. I didn't try bigger caps as tubeswell suggests, but I imagine the shift/blocking effects will only worsen so there's a limit there.

The source impedance feeding the second grid cap - be it a hard wire ground or a lo-z or hi-z NFB source have minor impacts during OD. Think of these impedances as similar to adding a grid stopper to grid 2. My experiments showed a grid stopper on either grid has very minor impacts.

I'm left with no sense of right/wrong values for these caps, but .1u simply seems a good compromise for full LF clean tone and acceptable blocking behavior with OD. Even when mindlessly copied, .1u in the ballpark for what people consider good - so few designs bother with variations.

That's pretty much as I figured. Thank you for the trouble and confirmation because that's just what I was hoping for

I'm actually in the middle of designing right now and bench testing parameters that make the PI cattywhompus under OD conditions relative to the preamp wave form, grid clamping and the NFB circuit Obviously the cap discussed above can also be part of the effect too. I'm actually aiming for the right amount of imbalance because I like the way it sounds in the design I'm working on. Too asymmetrical gets mushy, but too symmetrical can sound sort of bland. It's the little things