Balanced Amp Topology

[kingneb]

Hello,

I am exploring fully balanced driver topologies for my KT88 push pull parallel project. Building balanced circuits with op amps is not hard but I am not finding much info for tube amps. A concept drawing I have posted below shows you an idea that I have. Any suggestions?

[the_manta]

Basically I see no problem with that topology.
But....
Power OPT's with a symmetric output for speakers are rare. Unless you already have OPT's or know a supplier, I would be careful. Just using the 4Ohms tap as "new ground" and the "old ground" as ode side and the 8Ohms tap as the other side is a big fail. These aren't symmetric ! Neither in winding topology nor in impedance ! Lets assume Volatge on 4Ohms is: Sqrt[4] = 2V. Then on between 4 and 8 Ohms it is: Sqrt[8/4] = Sqrt[2] =1.414V. You see ? You would have to use the the 16Ohms tap for symmetric output voltages. But As I said, due to interleaving these aren't physical symmetric in terms of leakage capacitance and inductance and resistance. Best would be a special OPT with symmetric secondaries with a dual-chamber-bobbin (dunno if that's the right word )

Next thing: Since you apply feedback directly to the driver tube it seems that your circuit is already your complete amplifier. But one single driver stage isn't really enough for KT88 with feedback. (In case you don't use a very high-gm tube with anode choke and µ~300)
But maybe you don't use global feedback and just posted only the last stages of your amp. Pls let us know.

Regards, Simon

[hidnplayr]

If you need more gain, use pentodes or a cascode in the preamp stage.
For a good CMRR, you might want to use a CCS in the tail of the LTP and/or an input transfo.
I have no experience with the latest, but have read many times that if you decide to do so, better spend some money on it.

For the feedback, you can use a fully symetrically wound trafo, I know of at least one trafomaker who does that (Automatic Electric Europe)
You can also use another simple trafo, and use two resistors (100 ohm or so?) to connect the 0 and 8 ohm tap to ground.
Or, you can use a transforer with a separate feedback winding. (this option will result in a lower DF I think)

Also, I think it's a good idea to add a source-follower or kathode-follower, to drive the KT-88's

just my 2c

EDIT: I just noticed you didnt ground the OPT secondaries, and there is no need for it, it should work as drawn..

[kingneb]

My transformer's will not cut it then as it only has an 8, 4, and common tap. Its more of an idea at this point, something that will go into mothballs until another amp project.

[Frank Berry]

You can always use a capacitor and series resistor to create a balanced feedback path from the plates of the upper output tubes to the cathode of the lower driver and from the plates of the lower output tubes to the cathode of the upper driver.

[kingneb]

How is an SRPP stage at CMRR?

[DavesNotHere]

you getting there, feedback is not really used commonly. mostly used as a fix in these circuits.


think of it as two se stages through: one from the + signal and the other as the - signal.

here is an example of a balanced amp:

[the_manta]

This is a nice circuit
Applying feedback on the cathode of a diffamp doesn't work properly since in theory the current should be constant. But when it's a See-Saw / Paraphase splitter this seems to work well when I look at that circuit.

But I wouldn't call that symmetric since the lower driver tube is extremely feedbacked to follow the upper tubes operation (which describes paraphase very well). A symmetric approach without symmetric feedback from the secondaries is standard diff-amp topology used in solid state amplifiers. Like in that driver stage: Treiber Old

[jazz]

Hi,

Did you look at the vacuum state electronics website? Allen has posted such a schematic there.

Link: Schematics

grtz
Joris

[Frank Berry]

That is really not a balanced feedback circuit. Balanced feedback requires two feedback paths. One path from each output tube or from each end of a center-tapped feedback winding on the output transformer.