Inverter question

[catalin gramada]

Hello. Here is how Ampeg designed their nfb network around paraphase inverter (top). If I move the common OT lead from the ground over shunt nfb resistor (bottom) the decoupling cap over cathode resistor is still necessary please ? And I talking here about possible positive feedback issue... Thanks.

[loudthud]

Unless there is lots of shielding within a transformer, there will be some capacitance between windings. Even if the capacitance is small, because the Voltage is so large on the primary, significant current wants to flow. Normally this is not a problem when one side of the secondary is grounded. When you lift the ground on the secondary, it can be a problem because the Voltage from the parasitic coupling can easily be bigger than the Voltage across the secondary, not to mention it will be higher at high frequencies.

Do you need any more reasons not to try this ? I'm guessing you don't own an oscilloscope.

[pdf64]

I think the decoupling cap is still beneficial, as those common cathode stages are in cascade and share cathodes, and a (weak) positive feedback loop would form if unbypassed.

[Helmholtz]

At first glance and without knowing circuit values,
I expect lower PI gain and poor balance.
But could be wrong because of several interacting NFB paths.

[Helmholtz]

Unless there is lots of shielding within a transformer, there will be some capacitance between windings. Even if the capacitance is small, because the Voltage is so large on the primary, significant current wants to flow. Normally this is not a problem when one side of the secondary is grounded. When you lift the ground on the secondary, it can be a problem because the Voltage from the parasitic coupling can easily be bigger than the Voltage across the secondary, not to mention it will be higher at high frequencies.

Yes, that could cause stability issues.
Primary to secondary capacitance can be between 300pF and 3nF.

[catalin gramada]

Thank you gents. Clearly understand. Just a additional question I wonder jn this case please: Even the common nfb return wil be linked to ground (like in top sample) then the grounding point choosed make sense to be close to power stage supply node more than the stage the nfb network it belong.This should be correct please ? Thanks.

[Helmholtz]

From circuit logic the NFB return should use the same ground as the PI.

[catalin gramada]

Ok. I installed separate cathode resistors without bypass caps. 6db of global nfb used but did a small experiment using different shunt resistors into a string. To be honest there is a effective diference jn sound comparing 100k/2.1k than 470ohm/10ohm despite the fact the global nfb ratio is the same. I much prefer 470/10 ohm. Why such of difference please ? Thanks

[Helmholtz]

To be honest there is a effective diference jn sound comparing 100k/2.1k than 470ohm/10ohm despite the fact the global nfb ratio is the same. I much prefer 470/10 ohm.

Is that drawn on toilet paper?

I guess you measured/compared the closed loop gain?
Actually I wouldn't expect a difference.
In both cases the NFB signal reaching the cathode should be roughly the OT secondary voltage divided by 150.
Also the DCV at the cathode shouldn't change noticeably.

Might be a parasitic effect.
Did you mount the NFB series resistors directly to the output jack?

"Sound difference" is rather vague.