Sustain - Grid Blockin or Anode Clipping

[wizard333]

To which part of the bassman schematic are you referring?

As far as, is 6 stages necessary for gain, no. You can get complete melt down levels of gain from 3 stages.

[poorboybrad]

I was looking at an AB165 schematic, on the bass channel after the first two gain stages there is a 7025 (12ax7) stage where a 470k resister from the plate to before the coupling capacitor. I had read that this resister was there to control gain and compress the signal. Am I correct in calling this "local" feedback?

[defaced]

I'll let someone else field your specific question. But I do have an amp that uses local negative feedback around the last gain stage and CF, see attached image.

[R.G.]

Are there 2 ways to clip the grid?
1 being a an input signal whose positive cycle exceeds Vgk.
2 being a colder biased stage, or, larger Rk.
(is grid current limiting the same as my number 1 above?)

Grid voltage affects the stream of electrons going from the cathode to the plate. If you set up a situation where the grid voltage can't move around any more, that's grid clipping - at least the way I think of it. There isn't a set vocabulary for it that I've found.

The grid on a tube is free to move as far negative with respect to the cathode as you can pull it, subject to getting arcing if you get into kilovolts. The impedance of the grid itself doesn't change over this range (significantly at least, there are always edge effects.) When the grid gets to a voltage near or at the same voltage as the cathode, it starts intercepting instead of repelling the electrons streaming by it.

The intercepted electrons will make the grid voltage more negative again, unless they're bled away by the external signal source siphoning them away - that is, pulling the electrons away as a current. This is the basic operation of the start of grid current, and why grid blocking happens. In grid blocking, the grid is effectively AC coupled by having a very large grid leak resistor - hey, grid leak! - that's why that resistor is called that! - and so any electrons it picks up when the signal pulls the grid up to Vgk=0 go primarily into charging the coupling cap more negative, causing a colder bias shift.

However, if the signal source can pull the grid-conduction electrons away fast enough to keep it from building up a negative charge, then the grid voltage can climb above Vgk=0. In a 12AX7, the grid goes from being quite a high resistance for Vgk being very negative to about 4k-7K for Vgk>0. This number is a reflection of the area of the grid wires available to intercept electrons from the stream headed for the plate. Since the grid being positive pulls more electrons from the cathode's emission toward the plate, the tube's current still increases as you make the grid modestly positive. Hence the effective use of positive grid voltages in class AB-2 power amps.

It is the sudden change of the grid impedance from near-infinite to low-K's that causes grid clipping by exceeding the ability of the signal source to drive the grid. In some tube logic circuits and very few amplifier circuits from the Golden Age, the grid was pulled to B+ with a quite large resistor for bias. The resistor was so big that the grid ran effectively at Vgk=0, and the tube was said to be "running in clamp" as the grid was clamped to the cathode voltage.

Driving a grid with a high impedance source can clip the driving signal HARD. Driving a grid with a low impedance source can produce very soft clipping. This is one reason putting a buffer right in front of a tube amp can soften up the clipping for signals which cause the front end to clip.

That's the only way for a grid to clip a signal that I can think of - the transition from nonconducting to conducting. The impedance of the driving source determines how sharp the clipping is.

However, the grid does still affect the amount of current flowing in the cathode-plate electron stream. If the tube is running with a low current (high Rk, high Rp, etc.) then it doesn't take much for the grid to run the electron stream from fully cut off - and this happens with a 12AX7 at only half to a few volts, depending on the Rk - to as much current as the cathode can supply. There is a limit to how much current the cathode can supply, but generally the grid conditions and rp limit things first. Triodes don't run a "saturation" like semiconductors and pentodes do, in general, just a mushy ending with the high rp and other impedances limiting things.

Note that it's Vgk that matters, and the bigger the unbypassed Rk is, the more that the cathode follows the grid up as Vgs rises. The logical conclusion to this is a cathode follower, where the cathode is alway only a volt or two higher than the grid, no matter where the grid goes. This happens when the plate resistor is so low that the plate and cathode don't "meet in the middle" by pseudo-saturating the tube current to B+ divided by the some of Rk, Rp, and rp.

So no, there's only one way to get grid clipping, at least with my own personal terminology - exceeding Vgk=0 with a source that can't supply the current. But the action of the grid on the electron stream can cause clipping on the output.

Is there more than 1 way to clip the anode?
Other than exceeding the supply rail voltage?

Yes. You can either turn the tube current off by making the grid way negative (that's "exceeding the supply rail voltage"; actually, it's turning off the internal current so the external Rp pulls the output up to the B+), you can turn the tube on as hard as you can and have the sum of Rp, Rk and rp limit the amount of current that can flow (this is what passes for saturation in triodes) or you can pull the grid positive and get into mushy, nonlinear, on the edge squashing as Vgk goes positive with a low impedance driver.

The anode will clip on both positive and negative sides, by different processes.

[R.G.]

I was looking at an AB165 schematic, on the bass channel after the first two gain stages there is a 7025 (12ax7) stage where a 470k resister from the plate to before the coupling capacitor. I had read that this resister was there to control gain and compress the signal. Am I correct in calling this "local" feedback?

It's local feedback, but it's not to "compress the signal".

That tube has a gain of about 30. It's fed by two 220K resistors, one each to each of the input channels, through the 0.01uF input cap. It's output (plate signal) is fed back to the sum of the two 220K resistors. You could replace the tube with an opamp, fixing up the bias and power supplies so it would live, and get much the same result. The 220ks and 470K form a simple inverting adder feedback amp with a gain of something like one or two. It's a rudimentary feedback gain amplifier.

As such, it doesn't compress - it HIDES any compression or distortion that the tube might be doing, as long as it has open loop gain to do so. It does "control gain" in that feedback amps have much more stable gains than non-feedback amps.

[lowell]

Ive started toying around again and reread this thread. I understand this stuff a bit more I think. ANyone ever direct couple a cathode follower to a common cathode stage? I'd think this could drive the grid more positive for smoother grid current limiting effect. Only issue is biasing the following stage. If the cathode follower cathode is at say 30v, then the cathode of the proceeding stage would have to be around 31v or higher. Maybe also using a supply that is 30v higher for the proceeding stage as to maintain headroom. Anyone tried this? You may ask "why?". Just humor me. Oh and why use a cathode follower at all in a capacitively coupled tube amp??

[jpfamps]

If you undersize everything and then apply a bass boost towards the end, to put the low end back in, that's how you make what Bruce Collins called the "wanker woofie drone tone", as modelled by James Hetfield, Dimebag Darrell and so on.

And Dimebag Darrell achieved this with nary a valve in site.

[tubeswell]

Ive started toying around again and reread this thread. I understand this stuff a bit more I think. ANyone ever direct couple a cathode follower to a common cathode stage?

Hi Lowell.

Since its your prerogative if you want to re-open an old thread you started... ;-)

The 5F6A direct DC-coupled pair is the classic example I can think of... (unless you are talking about putting the CF stage in front, and then direct-coupling to the grid of an ordinary inverting gain stage; in which case, if the following stage is cathode biased, you need to use a fairly largish Rk on the following stage to get the grid naturally sitting at the idle voltage of the prior CF stage, which in turn - assuming the HT remains the same - will mean the plate resistor of the following stage has to be correspondingly reduced, thereby reducing gain in the following stage. But I'm not sure whether that's what you were asking??)

I'd think this could drive the grid more positive for smoother grid current limiting effect. Only issue is biasing the following stage. If the cathode follower cathode is at say 30v, then the cathode of the proceeding stage would have to be around 31v or higher. Maybe also using a supply that is 30v higher for the proceeding stage as to maintain headroom. Anyone tried this? You may ask "why?". Just humor me.

As R.G.s prior post in this thread explained @ Sustain - Grid Blockin or Anode Clipping ; "the bigger the unbypassed Rk is, the more that the cathode follows the grid up as Vgs rises. The logical conclusion to this is a cathode follower, where the cathode is alway only a volt or two higher than the grid, no matter where the grid goes. This happens when the plate resistor is so low that the plate and cathode don't "meet in the middle" by pseudo-saturating the tube current to B+ divided by the some of Rk, Rp, and rp."

Oh and why use a cathode follower at all in a capacitively coupled tube amp??

I have used an LFO stage followed by an AC-coupled, cathode-biased CF, in order to bias the CF in such a way as to keep the CF's h-k voltage within manufacturer's recommended specs. The fact I was losing voltage swing didn't matter in that application, because the aim was to produce a tremolo wave form with enough current to happily influence a couple of 6L6 grids. I could've used a voltage divider to level-shift a DC-coupled signal going into the CF grid, which wouldn't matter for a trem circuit, but AC coupling to a cathode biased stage just seemed simpler and more automatic at the time.

If however, the level shifting is used in a signal path, that would also introduce some inter-stage HF roll-off, which then leads one into a finicky spiral of voltage-divider-resistor/bypass-cap combinations, and ...