> article that explains it
Site seems to be down. Maybe we crashed it.
> the tube would be operating out-of-spec
Specs are for cowards.
> maximum grid voltage as 0V.
Yeah, but that does not mean the sky falls down at +0.000,001 volts.
At -0.1V, the grid impedance starts dropping from "infinity" to less than a Meg. Around zero it may be 100K, at +0.1 it may be 10K.
At +10V it may be 1K, at which point we have Grid Dissipation of 10^2/1K= 0.1 Watts. That may be enough power in a small grid to warp it, at least enough to short to the nearby cathode. The official rating is "Don't do that!".
But back at +0.1V and 10K, the grid dissipation is 0.1V^2/10K= 0.000,001 Watts. That won't hurt the grid at all.
Oh, this plan shows 10K in series with the grid, so you probably can't melt the grid even with 10V input. THD may be 432%, but the tube won't be ruined.
The overall amp is scaled for a hi-Z dynamic (or crystal) mike in a small room (I mean, 10 Watts output, pretty low-powered PA system). The mike output impedance might be 20K, so a 10K grid impedance at 0.1V would be a heavy and nonlinear load. But such mikes and applications rarely work above 0.01V (10mV). The grid is 100K-50K at the lowest, distortion is not too bad. And you also see this type of scheme in phono preamps, where the cartridge impedance is 2K-5K (except at 20KHz resonance where signals are weak anyway). The low grid impedance may be a non-problem when source impedance is lower.
But wait, there is more. The 2.2Meg grid resistor is (I think) "out of spec" for the 12AX7. Why? Because grid current in a resistor higher than 1Meg will induce enough grid voltage to shift the operating point. "Which way?" is kinda tricky. If you use it that way with a high B+ and small plate resistor, it is possible for the cathode current to increase until the tube melts. If you do it with a modest B+ and a large plate resistor, the grid tends to swing negative. One way, grid emission rises, the other way contact potential dominates. You will probably find the grid sitting about -0.3V to -0.5V (you need a VERY hi-Z voltmeter to prove it).
I should comment that this plan is a toss-off experimenter's starting point, and would be made better if it were a Product.
While a 2.2Meg grid-leak will let the grid float down negative, as shown a dynamic mike at the input shorts-out the 2.2Megs. You end up with about 15K of DC resistance, and the grid will sit mighty close to zero. You can get maybe a tenth of a volt of swing, maybe less, before distortion gets gross.
Yes, a loud talker can probably overwhelm the first stage. It would be better if the grid-leaked input were cap-coupled: then the first clipped positive peak would charge the cap and shift the DC grid voltage negative, allowing most of loud sounds to come through unmolested. (Using a crystal mike would give some of this effect, and they were popular for small PA.)
> Why not just add the cathode resistor?
A waste of a good $0.12!!! Actually, the problem is that then you need a Capacitor, and those cost even more. Or take less gain, but this is a mimimal amplifier so we want all the gain/bottle we can get. And adding a resistor without a bypass adds thermal noise (though maybe not so we care) and allows more heater-hum leakage (enough to care about). And this grid-leak trick tends to put the tube at the highest possible gain and near a low-noise point. And it saves many-many pennies. And used with care, the users won't know how cheap you are. (Though I think a mike input really does need more headroom than this... did Bogen ever grid-leak a mike-input stage? If they didn't, then users must have objected.)
