Shunt NFB and gain Q

[teemuk]

What's next? Sites that specialize in "NOS" SS devices? With the same sort of "Manufactured by Texas Instruments for US Telecom." descriptions like you see supporting NOS tubes.

Well... the pedal geeks have already wondered knee deep into that zone.

[pdf64]

Wombaticus, yes the Vox AC50 used that arrangement; it works very well.
Pete.

[J M Fahey]

@ wombaticus.
Yes, it's true.
And I like it.
Lots of advantages: symmetry is guaranteed, you do not lose DC voltage (headroom) in the long tail resistor, etc.
Since you aim for 1:1 gain, meaning you are applying 100% feedback, having , say, 50X open loop gain available is not bad at all.
The error between that and the theoretical infinite gain of the "perfect" Op Amp in practice translates to a 1% or 2% error in gain, less than resistor tolerance.

[Tage]

This is the Phase inverter used in most ampeg amps right? Like V4's and such? I always wondered how it performed compared to the long tailed pair? It nver occured to me that it was using negative feed back to generate the G2 referance, I think I may need to try it out in a build.

[pdf64]

Yes, the V4 http://www.webphix.com/schematic%20h...v4poweramp.pdf uses the floating paraphase type of phase splitter.
My experience is that it's beneficial to either bypass the shared cathode resistor or have individual cathode resistors, bypassed or not; otherwise, when heavily overdriven, the NFB can flip into 'blips' of positive feedback somehow, via the shared cathode.
Pete.

[J M Fahey]

Fully agree and add: in the long tailed PI we *need* an unbypassed cathode resistor to force them to share a tight current source ; on the paraphase we need the exact opposite, meaning separate (or perfectly decoupled) cathodes.
Why? : because here we need a "perfect" (or as good as possible) 1:1 feedback network making the 2nd plate follow the 1st one perfectly, only out of phase.
Adding other uncontrolled feedback or signal injection network through the cathodes messes with our calculations.

[Chuck H]

And if you've ever had an uncontrolled injection you know how painful THAT can be.

[Tage]

So... diode biasing could be advantageous in this situation?

[Wombaticus]

So... diode biasing could be advantageous in this situation?

Sounds reasonable...