Tango lessons. From Ihquam's infinite damping factor to Nelson's own load vanishing from the equation, their wordings at least on their bold faces suggested that within its limited power envelope, the F7's +/- feedback created less load-variant behaviour. With negative feedback out of phase, subtractive and canceling, positive feedback in phase, additive and enlarging, the balance between anti-phase loops appeared to create an ideal state of zero output Ω. But since fluctuating signal/load impedance meant that this interplay had to be dynamic, how could one maintain a static zero Ω condition? I asked Nelson for some more insight into this dance and what it means to the end user. As usual, there was a gap been idealized theory and assumptions based on it; and the real world with its actual meaning.

"At the equilibrium point—where you can cancel the load value out of both sides of the equation—the output impedance is zero so the damping factor would be infinite. However, that's the simplified story, the one using 'ideal components'. It's tempting to use the infinite damping factor phrase but it has more value for marketing than actual performance. Being merely the reciprocal of the output impedance, in this circuit it can range through a continuum from positive numbers through zero into negative numbers. At zero, you have infinite damping factor. Below zero, you have negative damping factor. The difference between a damping factor of +1000 and –1000 is about 0.2% as far as the loudspeaker performance goes. That's how the use of the term can be misleading. Plus, a literal zero output Ω is not a realistic expectation anyway even if you sat there constantly twiddling the adjustment pots on the circuit. Any little thing will move it away from having infinite damping. So you settle for reasonable values knowing that there's very little difference. The summed impedance of the load plus the output impedance of the amp and cable is the actual value to consider.

"The small amount of positive current feedback in the F7 sets the output impedance much nearer to zero. It gives the amplifier some of the authority of a big amp without the additional hardware. This represents about a 20% improvement in the summed impedance of amplifier/speaker/cables over the original circuit. That's audible. In these modest quantities, it has very little downside. It gives this simple Class A circuit flatter response and better control into reactive loads. If you wanted more damping factor via the conventional route, you'd raise the open loop gain and use it for more negative voltage feedback. For example, a damping factor of 100 for this circuit would require a feedback factor of 100 (40dB) and likely an additional gain stage.


"Is it load invariant? No, it responds to the load in an interesting way. The native damping factor of the F7 is about 5, reflecting a negative voltage feedback factor of about 14dB. This value is proportional to the impedance of the load. When the load impedance goes down, so does the amplifier gain. Positive current feedback replaces that lost gain. It does not otherwise lower the distortion of the amplifier, which mostly reflects the output current. With a high-impedance load, there's less positive feedback. With no load, there is none. I have attached some graphics to illustrate some of this. The first is my 'tough load' whose impedance variations simulate some of your tougher (real) loudspeakers. It dips down below 3Ω at a couple of points and inductively goes to about 40Ω at 20kHz. When you use an F7 to drive this load, you get results like the second and third graphics.

"The second picture is the frequency response. The amplifier is set at 14dB output without a load and the blue curves represent the output response without positive current feedback. The flat ones are for an 8Ω load and the wiggly ones are the tough load. The two red curves at 14dB are with the positive current feedback and are barely distinguishable from each other. The third picture is an F7 driving a 50uF+1MH coil in series. On the left you can see the ringing of this resonant circuit at the output of the amplifier without positive current feedback; and on the right you see it with. The last picture shows the distortion vs frequency at 1-watt figures for both loads. The slightly higher distortion figure on the red curve (with positive current feedback) reflects the greater amount of current being delivered into this load at that frequency. Regeneration of distortion? It’s on the order of the order of 0.1% times 0.1%, which is about  0.0001% distortion at 1 watt even with the tough load; and it's still mostly 2nd harmonic. Again, this is due to the circuit having very good intrinsic linearity and a simple characteristic. It's tempting also to use even more positive current feedback for a negative output impedance (negative damping factor) so as to create a motional control amplifier. However that veers in the direction of 'special effects' and starts making the circuit more specific to particular loads both in terms of sound and stability. This is not the goal of the F7, which is to be a good-sounding general purpose amplifier. As to Bruno Putzeys' class D, I doubt he uses so much feedback just for damping factor as like me, he understands the diminishing returns of high DF numbers. Besides, he already uses positive feedback to establish oscillation in his switchers."