Understanding "grid drive" to drive, 2, 4, 6 power tubes

[Helmholtz]

But, when the PI pushes into AB2 operation, all the power tube grids become LOW-impedance loads, requiring CURRENT from the PI (usually provided by Cathode-follower circuits).

That's why I mentioned AB2 above. I think the SVT with cathode follower drive is designed for AB2. This operation mode requires low impedance drive to keep distortion at the non-linear grid input impedance down.

[Helmholtz]

This circuit won't work without coupling caps.

[R.G.]

There are several things I didn't dive deeply into, largely because I didn't want to keep on typing. For every setup one speculates about, one ought to use the implied technique from the post: pick a circuit node, evaluate all the loads that tie to that point, and then all the driving signal sources, including their internal impedances. Then start summing things up.

The actual grid capacitance isn't as bad as the Miller capacitance, caused by the hundreds of volts of swing on the plates coupling through whatever the real capacitance is to the grid. It works out that the current shoved into/out of the grid by the plate-grid feedback capacitance is multiplied by the voltage gain from grid to plate. That adds up FAST.

The issues with Miller capacitance was one of the motivators in the development of pentode. The screen grid screens the control grid from much of the high voltage swing on the plate, decreasing the Miller capacitance and letting the tube have much better RF response. Triodes had suppressor grids inserted for this purpose. That got us tetrodes. Tetrodes had better HF response, but had a "kink" in the low voltage/high current part of their plate curves because of electrons literally bouncing off the plate. Suppressor grids near the plate let electrons sail through but suppressed the "bounced" electrons back into the plate, eliminating the "kink". Beam forming plates did something similar, but with fields instead of actual grid wires. Hence the name "kinkless tetrode" for the KT series. RF frequency response was a major issue with high frequency tubes. Special tube construction and special operating modes (common grid, anyone) were used to make this a bit better.

A MOSFET has a higher intrinsic gate capacitance, all right. Gate capacitance and how to deal with i are part of nearly every design effort with MOSFETs. One way to get around this is to use the MOSFET in common drain mode or common gate mode. Common gate is much like common grid, the basic capacitance is still there, but the amplification of Miller capacitance can't happen. Using MOSFETs in common drain/source follower sidesteps some of the issues by using the very high transconductance of the MOSFET itself to drive the gate-source capacitance. A crude handwaving of this would say that using a MOSFET in common-drain divides the gate-source capacitance by the transconductance. That's simplistic, but a place to start looking for what's really happening.

In all this, it keeps looking more and more like the Devil is indeed in the details.

[Helmholtz]

The actual grid capacitance isn't as bad as the Miller capacitance, caused by the hundreds of volts of swing on the plates coupling through whatever the real capacitance is to the grid. It works out that the current shoved into/out of the grid by the plate-grid feedback capacitance is multiplied by the voltage gain from grid to plate. That adds up FAST.

The grid to plate capacitance of a 6L6 is 0.6pF. With a voltage gain of 10 or lower, the total effective grid input capacitance stays well below 20pF. When driven from a 50k source impedance, an upper corner frequency of around 160kHz results.

[J M Fahey]

I think what you MEANT to say was, "more efficient speakers!"

Just don't sell me out for digital, okay?

Jusrin

No, not digital.

But I stand by my point ,which is: Tubes are unexcelled at clipping, which means you have to ... um ... clip them, meaning full blast for long times.

Even when "clean", if you are playing at any significant volume , say a Club or even rehearsal near a drummer, you are clipping peaks all the time, simply because of Guitar dynamic range, just scope speaker out.

I envision no realistic situation where you canīt have more than enough Guitar volume with 30/50/60/100W , the latter being unusably loud except on very few situations.

Do you need more than 100W?
Sure, but for Bass, Keyboards, PA or Pedal Steel ...... all of which are happy with SS.

[nsubulysses]

This is a lot to learn and think about so I am still checking it out. All the answers are definitely helping me to understand though so thank you

In the meantime I will say that I am not using 6L6. I like to use 6CA7 power tube.

What about sound city 120 or hiwatt DR201?

[Old Tele man]

6CA7s can handle more B+ than 6L6GCs.

[Helmholtz]

In the meantime I will say that I am not using 6L6. I like to use 6CA7 power tube.

Not much difference regarding drive requirements between 6L6 and 6CA7/EL34.

[nsubulysses]

My apologies the coupling caps were missing in my original drawing as helmholtz pointed out

here is more info as per my case
B+ 585V
Screen 560V

I have no choke, I have 1K 20W in its place

With 4 power tube amp I can do 4x6CA7 and get 144W at clipping and it does not punish the power tubes.

With 6 power tube version I was hoping for 200W+ and was disappointed in 180W or so, so I am looking at the phase inverter as my error?


My 4 power tube output transformer is 150W OT at 1.9K primary. 6 power tube transformer is 200W at 1.6K primary.

[g1]

In the meantime I will say that I am not using 6L6. I like to use 6CA7 power tube.

What about sound city 120 or hiwatt DR201?

Good practical examples of higher impedance PI plate circuits (rather than cathode follower drivers) being used to drive 6 x EL34. I would note that for the Hiwatt STA400, with 6 x KT88, they went with cathode follower driver tubes style.

[Helmholtz]

My apologies the coupling caps were missing in my original drawing as helmholtz pointed out

here is more info as per my case
B+ 585V
Screen 560V

I have no choke, I have 1K 20W in its place

With 4 power tube amp I can do 4x6CA7 and get 144W at clipping and it does not punish the power tubes.

With 6 power tube version I was hoping for 200W+ and was disappointed in 180W or so, so I am looking at the phase inverter as my error?
[ATTACH=CONFIG]51720[/ATTACH]

My 4 power tube output transformer is 150W OT at 1.9K primary. 6 power tube transformer is 200W at 1.6K primary.

Easy to find out if you have a PI drive issue. Just compare max. undistorted grid signal voltages between the 4 and 6 tube versions.
What is plate supply voltage at full output?
It is not uncommon to have around 15% power losses in the OT.
A load impedance (OT primary) of 1.1K to 1.2k for the 6 tube version may produce somewhat higher output power.

[Helmholtz]

Good practical examples of higher impedance PI plate circuits being used to drive 6 x EL34. I would note that for the Hiwatt STA400, with 6 x KT88, they went with cathode follower driver tubes style.

What do you mean with "higher impedance PI plate circuits"? To supply more drive current, PI plate resistors need to be smaller and plate currents be higher.

[Mike Sulzer]

That's why I mentioned AB2 above. I think the SVT with cathode follower drive is designed for AB2. This operation mode requires low impedance drive to keep distortion at the non-linear grid input impedance down.

I do not understand the point of this discussion. If you want to make a guitar amp with multiple tubes in parallel, you want it to sound as an amp with one tube per side does. So you keep the same ac coupled drive, but lower the impedance so that the drive per tube is the same.

[g1]

What do you mean with "higher impedance PI plate circuits"? To supply more drive current, PI plate resistors need to be smaller and plate currents be higher.

Sorry, I meant 'as opposed to lower impedance cathode followers' and have edited my comment. I was trying to say that these examples show that extra current drive is not a necessity, at least when using 6 x EL34.

[J M Fahey]

Just thinking aloud and "thinking like an SS amp designer" which I am, I left the Tube boat *long* ago:

1) if we respect the datasheet constraint (and why not respect it?) of "100k grid leak resistor per tube" then 6 tube amp should have 33k per side ... this was already mentioned above.

2) even if tubes themselves behave as "infinite input impedance devices" , the very real 33k per grid net means "infinite impedance my *ss "

The PI driver does not know or care whether power tube grids take little or no current, it still has to drive those 33k, period ... which are a tough load.

Considering that EL34 will need about 40v peak signal and 6L6 above 50V peak (both must be brought from negative bias to 0V to meet maximum AB1 peak current) then letīs play it safe and design for 60V peak (at the very least) and 33k load, which means 2 mA peak drive current.

Now if we were getting such current from a 1:1 transformer driver, which is "almost" 100% efficient, then 2 mA idle per PI plate (which are plain Class A stages) would be fine ... but we are using horribly inefficient resistor loaded Class A stages (both PI halves) capacitively driving a demanding 33k load ... which means they will need, at least twice as much idle current per half, so about 4 mA idle.
And that, if you are lucky and can meet best case constraint: 50% of current through its own plate resistor, 50% into the load (33k) so idle current should be 4 mA at least.
Personally Iīd add some safety margin and use 5 or 6 mA idle per PI half.

What do we have in the classic Fender type 82k/100k pair?

Whatīs plate to rail voltage drop?
Too sleepy (6AM here, just woke up for a leak and will continue sleeping) to search for schematics with voltages on them but supposing 100V across them, we are talking about 1 mA per half: woefully inadequate to drive 3 pairs "by the book" .

Last 2 neurons still awake suggest 2 basic solutions:

a) drastically reduce PI plate resistors to increase possible idle current.
Say go from 82k/100k or 100k/100k to at least 47k/47k (what Silverface designers did, no doubt because they hired real Engineers) and in this case I would think 33k/33k
Also replace weak 12AX7 by at least 12AT7 (Fender did that) and even better 12AU7 , which are respectively just capable or happy to pass some 5mA idle , OR:

b) grab the bull by the horns and add a buffer stage.
Mind you , I am not even considering Class AB2 or any grid current, just a couple of tough to drive 33k resistors.

Last side note: why did I call myself "thinking like an SS designer"?

Simply because *in principle" Tube designers assume infinite Z loads, with a very easy to drive 1M resistor to ground ... easy peasy even if you drive them from a 100k to 220k plate resistor... while SS designers are 24/7/365 aware and worried about drive *current* ... which in this case is very real.