> way to avoid filter interaction,
They are sure to interact. Screening won't fix the fact that all the tuned circuits are on the same busses.
Anyway in audio we often want some interaction.
> How to to calculate the Q bell... close enough to ideal 3db bell
Why is "-3dB bell" somehow "ideal"? That's just a handy way to express a curve.
> What's the formula to compute R in relation of db for oct.
Simplify. Just look at the Cut section: the input 10K resistor, one pot and tuned circuit, and an infinite load. You can lash this up on the bench without the opamp, just a signal generator, an AC voltmeter, and some passive parts. Use one pot, with one cap and resistor (no coil). When the pot wiper is at the far end of the pot, there is little effect (in the complete cut/boost plan, a centered wiper has no effect, or rather its effects on cut and boos cancel out).
With the pot wiper at the other end: At very low frequency, the cap has very high impedance, so the "cut" network is unity gain. When the capacitor's reactance equals 10K, the response is down 3dB. At very high frequency, the cap is a very low impedance. If the resistor were zero, you would have very large cut. If the resistor is say 2K, you have a flat shelf at (2K/(10K+2K)) or -15dB. If the resistor is 10K, the shelf is -6dB. So the "Q" resistor actually sets the maximum depth. Note however that the 10K pot adds to this, so when the pot wiper is not at the end the depth decreases.
Now try the same thing with a coil instead of the cap. Now the high frequencies are unaffected, the lows fall off to the depth set by the Q resistor.
The term "Q" is not good here. In general audio EQ we use Qs from 0.5 to maybe as much as 2, mostly around 1. A Q of 1 is a VERY low Q, and combined with the limited depth (less than 15dB) you just do not get the same things as radio tuned circuits with Q of 10, 50, even 200. Everything is sloppy and you don't get the intense interaction of a high-Q circuit.
For our low Q equalizers, you can pretty nearly estimate where the flat part of the curve starts to dip (or boost) just by looking at where the L or C equals 10K, without considering the other reactance.
When you do the full cut/boost plan with opamp, the boosts are the mirror image of the cut (assuming the opamp has much more gain than the maximum boost).
For this multi-band EQ all on one bus, you also have to figure the effect of all the other pots. Five 10K pots with centered wipers gives about a 1K impedance for the reactances to suck against, not the 10K of a single network. It also increases noise: the overall gain is still unity, but internally it is attenuating by 10 and then amplifying by 10. This would not be the first-choice plan for a 27-band EQ.
> 350V B+ does it matter? may I use 250 V Jakob PSU?
They are vacuum triodes. Feed them whatever you have. All voltages change, nearly in proportion, and it still works fine. Maximum output level drops a little faster than supply voltage, but 250V should be fine. (And 350V is really a little high for heater-cathode insulation in the upper triode in the output stage.)
> A over load detector topology for tube operation
Audio is audio. Any level detector will tell when you have "high level". You usually do not want to wait for an OVERLOAD light, you usually want some light to come on -before- overload.
For a very rude indication that this particular amp IS overloading, wire an LED across the input grids. As long as the opamp is doing its job, the voltage across the two input grids will be a fraction of a volt. But when the opamp overloads, it can't force its inputs to be equal. Increased signal will put a couple volts across the grids, which will light the LED (dimly).
> you know guys....seriously I'm fear of electric shocks
I've taken 120V shocks all my life, but I would be afraid to live where you have 230V coming out of wall outlets. Every time you plug in a lamp, you put your finger next to 320V peaks 100 times a second.
High voltage is no more dangerous than poison snakes or power saws. Keep the snake in a tight box. If you have to play inside, unplug the saw and be sure it is completely stopped before put a finger near it.
This plan will work JUST as well with an IC opamp powered from 9V batteries. It won't have quite the same sound or output level, but you can perfect your filter components fearlessly. Then build the tube opamp separately, and be sure there is zero voltage on the input and output terminals (you do need the 1Meg grid resistors for this testing, and maybe a 10K bleeder on the output cap). Once you have the two parts working individually, you can bring it all together.