NFB and first stage headroom

[thorpej]

I've been fretting over the first stage headroom in the 6L6GC UL amp I'm working on. Trying to balance quiescent voltage of the 12AX7 AF amp stage against the needs of the DC-coupled differential inverter and finding the most linear operating region of the 12AX7 has led me to setting the bias of that stage to -1V, which is only 700mVrms of input signal headroom.

Obviously, I don't want to overdrive that first stage, which on the surface seems like what's going to happen with basically any input source I connect.

But I do plan to use about -12dB of feedback, which, as I thought about it, seems to suggest that, because the closed loop gain of the stage is only 25% of that the open loop gain is, I can actually manage 2.8Vrms of input signal before the stage overdrives (which would be plenty of headroom for my needs).

Am I correct?

 

[kward]

[erroneous post, content deleted]

 

[dcgillespie]

If I may -- you absolutely must consider the NFB applied to the input stage in determining if overload conditions will exist. As an excellent example, take the venerable Dynaco MK III amplifier. The static bias voltage on the pentode input stage is 1.0 vdc under quiescent conditions -- which will only drop as B+ levels to the stage drop as power output is increased. Against this, the amplifier requires 1.6 vac RMS be applied to the grid of this stage (2.26 volts peak) for the amplifier to develop full power output. Taken at face value, this would imply that gross overload exists, but of course it doesn't, as the NFB network elevates the bias by (basically) 1.97 volts peak under full power conditions. This, added to the static bias produces nearly 3 volts of bias for the tube, against the 2.26 volt signal applied to the grid. In other words, there is no overdrive.

In all of my work, I have found that as long as the peak input signal at the stage where NFB is inserted represents no more than 80% of the total bias applied under full power conditions (static + peak NFB applied) -- AT 20 kHz -- then an adequate bias reserve is present to prevent overloading. In the case of the MK III, the peak input signal required for full power output at 20 kHz represents about 85% of the total bias applied to the input stage under that condition.

Dave

 

[thorpej]

If I may -- you absolutely must consider the NFB applied to the input stage in determining if overload conditions will exist.

Thanks for chiming in, Dave. I have looked at several example circuits, which is what led me to have that impression. The Eico HF-87 is another example of a -1V biased input stage, and those are commonly driven with pre-amps that produce more than 700mVrms.

So, based on your explanation, it sounds like I was on the right track, but for the wrong reason I.e. it's not the gain reduction ratio that I need to think about, but rather the actual voltage being impressed on the AF amp cathode by the feedback loop.

 

[thorpej]

Related... Of course, I also need to be concerned about the signal being fed to the inverter and where it is biased... I'm guessing that the differential inverter behaves somewhat like a cathodyne in terms of its headroom, but of course I want to avoid overdriving that stage as well. However, it seems like the feedback loop technically only affects the gain of the first stage, thus the output swing of that stage would be reduced by the same ratio as the overall feedback (i.e. if -12dB of feedback is applied, the output swing of the loop's first stage would be 25% of what is shown on the load line), and that swing is what needs to be taken into account when biasing the next stage.

Have I got that right?

 

[dcgillespie]

As long as your differential stage has enough poop to drive the output stage with a low distortion signal, then over-driving the inverter stage is highly unlikely.

By circuit action, the voltage gain of the grid driven section of a differential inverter is 1/2 that which it would normally be were the otherwise same stage used in a non-differential configuration. As a result, it can be shown that the AC voltage developed at the cathode connection of the differential stage is an in phase signal that is 1/2 that applied to the driven grid. When this is accounted for, then as a result, the top section is really only effectively receiving 1/2 of the signal applied to it, producing the resulting 50% reduction in gain. Of course, with half of the signal appearing at the cathode connection, and the grid of the bottom connection grounded AC wise, then that section is effectively receiving the same level of drive signal applied to it as the top section is -- forcing a balanced output from each section if true constant current operation is achieved.

The point is then, with what amounts to being NFB applied to the grid driven section of a differential inverter, it inherently has built in overdrive protection, over that which the static bias level would indicate.

I hope this helps!

Dave

 

[thorpej]

The point is then, with what amounts to being NFB applied to the grid driven section of a differential inverter, it inherently has built in overdrive protection, over that which the static bias level would indicate.

I hope this helps!

Dave

It does, thanks very much!