SE sound from a P-P amplifier

[smoking-amp]

A triode with a constant current obeys the simple rule:
Vp = Mu x Vgk

Usually one thinks of Vp as the output and Vgk as the input. But this can be reversed as well. By grounding the grid and applying signal to the plate (Vp), the cathode (Vgk) will have an (AC) voltage of Vp attenuated by Mu. Since the usual resistive feedback network for a P-P amplifier is an attenuator, one can replace it with the triode attenuator. See attached diagram:

The usual feedback loop operation controls the amplifier output so as to maintain nearly zero difference between the amplifier input signal and the attenuated feedback signal. So, Vgk will nearly equal the amplifier input signal, and since Vgk = Vp/Mu , Vp (or the amplifiers output) will be Vinput x Mu. This is what the ideal one tube SE amplifier would produce, but with the power of a P-P amplifier in this case. Anyone try this?

Variations on this idea are presented at this website:
http://www.ne.jp/asahi/evo/amp/el86/report.htm

[janneman]

Very, very smart! One of the most original idea in audio I saw in a long time. You owe it to yourself to try it. For me there is one unknown: in what way would the forward path influence the feedback characteristics. I mean, your feedback mimics the response of a SE stage, meaning this generates a certain amount and a certain spectrum of harmonics, which is what you want. Now putting this spectrum through a conventional PP amp could generate some IMD, but you may not hear that. Do it, let us know how it works!

Jan Didden

[tschrama]

Heee that's indeed a nice creative idea! I'm not a real tube-guy but this looks really nice. Do you have any idea what a nice , relative low voltage, tube would be? High Mu or Low mu .. I'm going to my simulator right away .... build a virtual GC with a 12ax7......

regards,
Thijs


PS
Maybe the guys at TubeCad.com would like a copy of the link, I think they would like the idea too.....

[tschrama]

Would this work....

I gues it will need a very silent B+ supply?

Will the 100K bias resistor work....


Regards,
thijs

[smoking-amp]

re: Janneman
The more gain in the loop and the lower the distortion in the P-P amp part, then the less effect on the feedback triode's gain rule. One interesting idea would be to put the same type triode used in the feedback into the gain section of the P-P amp too. This way the P-P amp maybe naturally assisted in conforming to the triode's feedback rule. I myself favor using an output stage in the P-P amp with some partial cathode feedback, to keep the most distorting stage(output) relatively clean. (See my earlier posts on the Elliptron, an easy to build variation of the circlotron with about 28% CFB. Uses standard 40% UL P-P output transformer - say Hammond 1650T- and can be built conveniently with a single power supply.)

re: Tschrama
The Mu of the triode tube determines the forward gain of the total amplifier unless some additional resistive attenuation is configured in too. Need to use a triode with not too high an Rp or will not have much drive capability at its cathode. Suggest making R3 in your diagram bigger and connect to minus V voltage to make better approx. to current source and maybe can get power supply noise to cancel out too.
100K for R2 looks OK as long as inputs to the gain clone don't draw bias current. The driving impedance from the 12au7 cathode should be about Rp/Mu I think (or is it just Rp, should check on this).

[janneman]

You know, not many people seeing Thijs's diagram will be able to figure out how it works! I like that.

Jan Didden

[fdegrove]

Hi,

You know, not many people seeing Thijs's diagram will be able to figure out how it works! I like that.

Slecht karakter...

It took me one second flat to figure it out, honestly.

Cheers,

[smoking-amp]

One more thought: the output impedance of the NonInvGainClone ampl. needs to be considerably less than the Rp of V2. Or else some signal may feed thru R2 and V2 in the wrong direction to the output. Seeing as the gainclone is feeding an 8 Ohm load I would guess this is the case.
I would also suggest using a triode tube, V2, with a more constant Mu versus varying plate voltage for a better SET sound (ie. minimal "island effect" due to grid wire spacing), the 12AU7 is pretty bad. Try 6SN7 or 5687 maybe.

regards,
Don

[smoking-amp]

Just realized that the gain clone is a solid state version with DC gain. Some modifications are in order to handle SS amplifers: There will be no feedback thru the tube until the tube warms up, which will cause excursion to one of the rails until the tube conducts. And, DC signals must be handled too, for zero output offset. My original schematic was intended as a concept of operation schematic for an AC coupled P-P tube amplifier. Affixed in this post is one suitable for a solid state amplifier. (see diagram)
Values given are for approx. 2.5 mA operation thru tube V2. TR1 gets adjusted for zero output offset after the tube has warmed up thoroughly. Vb is the bias voltage for the tube at the design current. Diode V1 is included to delay application of the minus current source until V2 has warmed up. It could be a damper tube or possibly a second section of tube V1 wired as a diode. Or, one could just wait until V2 warms up before applying HV, a relay maybe. R3 provides feedback until V2 warms up, its then over-ridden by the lower cathode impedance of V2 when it warms up.
Just as a side note, the cathode drive impedance from V2 is
Rk = 1/(gm+1/Rp) and since gm=Mu/Rp Rk= Rp/(1+Mu) so V2's cathode drive capability is pretty good.
Oops, for safety, should put a 1 meg. resistor from -2Vb side of TR1 to center of TR1 in case pot. opens. Si diode strings on inv. ampl. input are for safety of the amplifier and are not functional in the circuit under normal operation. Pulling V2 out during operation may blow the speaker, so be carefull. Could include some circuitry to sense V2 filament current or even plate current (maybe across V1) that would shut down HV, but probably would be too slow to save the speaker.

[smoking-amp]

possibly a section of tube V1 wired --> V2

for safety, should put a 1 meg. --> 100K