Input grid leak bias

[Tobyk]

Do I have this correct:
With input grid leak bias and say a 5M grid leak resistor, the voltage on the grid may be around -0.5 volts.
A guitar signal is around 0.2 volts. Would a smaller grid leak resistor (say 470K like used in some Gibson GA20) render less negative voltage on the grid?
And if the grid voltage got down to -0.1 volts, would that easily cause the grid to go positive when playing? And would that in turn make the tube turn off, OR clip in asymmetric fashion?

Thanks!

[pdf64]

No, within the range useful to us, the grid leak resistor value has little effect on the operating point, eg Vg about -0.9V when it’s 10M, about -0.8V when it’s 470k.
And a grid leak biased stage doesn’t really grid clip, large signals just get rectified and push the bias colder, the same mechanism as blocking distortion.
Turning anode current off requires the grid voltage to be more negative than usual, not more positive.

[Tobyk]

Thanks pdf64, I guess all those years they were using 5M+ grid leaks were pointless then, if it only differs 0.1 V?
And about the clipping, it’s strange – I’ve read many times grid leak bias (especially mic channels with no grid stoppers) is prone to asymmetric clipping of the input tube when overloaded, resulting in a strong one octave above signal..

[roberto]

No, within the range useful to us, the grid leak resistor value has little effect on the operating point, eg Vg about -0.9V when it’s 10M

That is where usually 12AX7 start to exhibit grid current, that's why a great value brings the voltage there.

[Tobyk]

No, within the range useful to us, the grid leak resistor value has little effect on the operating point, eg Vg about -0.9V when it’s 10M, about -0.8V when it’s 470k.
And a grid leak biased stage doesn’t really grid clip, large signals just get rectified and push the bias colder, the same mechanism as blocking distortion.
Turning anode current off requires the grid voltage to be more negative than usual, not more positive.

Here's what HotBluePlates wrote on the subject back on Gearpage:

"Why "octave up"? Well the grid-leak biased Instrument channel can be prone to distorting by lopping-off the positive-going part of the input signal applied to that stage. And every octave pedal schematic I can find seems to full-wave rectify the signal (NOTE: this lops off only one side of the waveform).

Rectification is super-strong even harmonic distortion, with the 2nd harmonic (one octave up) being the dominant component. If the GA20T is prone to octave-up, it's probably because the input stages are easily overdriven until they rectify at the grid of the grid-leak biased 12AY7.​

Note the 12AY7's grid-leak bias is shown as -0.8v. This being the real Mic channel, it aligns with the observed effect of "more octave effect" because the input is more easily overdriven.

Grid overdrive of a preamp tube tends to clip the positive-going side of the signal if the stage is biased too far in that direction, because most tubes requires a larger voltage-change to turn-off than they do to turn-on. That's a recipe for strong second-harmonic, and so octave-up effects.​"

[pdf64]

HotBluePlates is a top chap, but bear in mind the context of that post, ie attempting to explain possible causes of the octave up effect in a particular vintage Gibson design.
It’s not like he’d made an in depth investigation of an amp on his bench.
I’ll recheck my findings, but using a higher source impedance. It may be that the bias shift mechanism becomes less effective as that increases.

[Tobyk]

HotBluePlates is a top chap, but bear in mind the context of that post, ie attempting to explain possible causes of the octave up effect in a particular vintage Gibson design.
It’s not like he’d made an in depth investigation of an amp on his bench.
I’ll recheck my findings, but using a higher source impedance. It may be that the bias shift mechanism becomes less effective as that increases.

Sorry to keep beating this but I find it interesting, and something that I think hasn’t been analyzed much before, if at all.
I feel it can’t be a coincidence that it only (to my knowledge) happens in this Gibson, which is the only amp I’ve seen using this ’no gridstopper 470K grid leak bias’ setup. Also, the resistors, being about 65 years of age, surely have drifted to maybe an even lower value.

[dorrisant]

Also, the resistors, being about 65 years of age, surely have drifted to maybe an even lower value.

They usually drift toward open.

[pdf64]

...
I feel it can’t be a coincidence that it only (to my knowledge) happens in this Gibson, which is the only amp I’ve seen using this ’no gridstopper 470K grid leak bias’ setup. Also, the resistors, being about 65 years of age, surely have drifted to maybe an even lower value.

Yes, as per previous comment, resistor value typically drifts higher with age, especially in the case of 1/2W CC type.

I think that whatever's going on with that Gibson amp video you're referring to probably isn't related to the input stage. Other than in that the higher the gain / volume, the more pronounced the effect. It may not even be working right, the effect may be caused by something having gone wonky with the phase splitter or output stage. Hence another example of the same model, or a new build of the same schematic, may not do the octave up effect.

I've done some more testing on a grid leak biased common cathode 12AX7 stage.
With a guitar, other than a slight loading effect on the guitar pickup reducing the treble response, there's not a significant difference between a grid leak of 470k, 3M9 and 10M.
If the subsequent stages of the signal chain are kept clean, there's no obvious overdrive / clipping, even with my Les Paul, which can put out a Vpeak of nearly 2V on the initial pick attack.
Whereas it causes obvious clipping with a cahode biased input stage if the resistor is too low value, eg 680ohms bypassed, with a typical 300V HT and 100k anode load.
Hence I suspect that a grid leak biased input stage may be more resistant to harsh clipping overdrive than a DC coupled cathode biased input.

With a sig gen and scope, using a 56k resistor in series with the sign gen output to simulate a guitar's source impedance, even with a very large continuous input of 2Vpeak, there's no harsh clipping. Rather waveforms are smoothly rounded off.
The grid leak value affects the symmetry of the output under that condition, with 10M giving a colder biased effect (top rounds off first), 470k a hotter biased effect (bottom rounds off first), around 3M9 giving about equal rounding of the upper and lower peaks.
Using the scope's spectrum anaylser, 2nd and 3rd harmonics are well below the fundemental, at least -20dB whichever the grid leak value. There's certainly not the very strong 2nd harmonic that's apparant in the Gibson video.

https://www.thegearpage.net/board/index ... t-32597257

[Tobyk]

Thanks for trying it out! I guess it will remain a mystery then. Btw, did you omit the grid stopper, and what input cap value did you use? These might be if importance too.
And yes, I’m sure the amp didn’t sound this way out of the factory.

[pdf64]

0.1uF coupling cap to input.
I added a 56k grid stopper with the recent sig gen testing, and didn't remove it when testing with a guitar.