Re: AF Osc AGC alternatives to lamp/thermistor/FET

On Mon, 21 Oct 2024 19:45:28 -0000 (UTC), Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

john larkin <jl@glen--canyon.com> wrote:

On Sun, 20 Oct 2024 19:25:08 -0000 (UTC), Cursitor Doom
<cd999666@notformail.com> wrote:
 

On Sun, 20 Oct 2024 11:42:27 -0700, john larkin wrote:
 

On Sun, 20 Oct 2024 14:25:33 -0400, bitrex <user@example.net> wrote:
 

On 10/20/2024 11:38 AM, john larkin wrote:

On Sun, 20 Oct 2024 14:15:19 +0100, piglet <erichpwagner@hotmail.com>
wrote:
 

Yesterday in the thread "Random thoughts on sinewave oscillators" I
posted a LT Spice wirelist showing amplitude levelling without fet,
ntc thermistor or ptc filament instead using current steering long
tail pair - kinda poor mans multiplier.
 
Here is another alternative, using diode conductance bridge
"variolosser".
 
For the full 1960s vibe replace U2 U3 with bjt current sources.

 
That's still limiting a Wein bridge, essentially a single-element
resonator.
 
If you do the double-integrator type oscillator, you can do gain
control on the first integrator, and then the second one adds 6
dB/octave harmonic reduction.
 
I think you can even get 12 dB/octave harmonic reduction if you do the
amplitude limiter right. 12*3 = 36!
 
 
 

There's probably a way to finagle the 4th section of an LM13700 to apply
a bias current inversely proportional to the output level into pin IB1,
here:
 
<https://imgur.com/a/xRVx2PC>

 
At some point, opamp distortion and slew rate will usually dominate THD.
I'd guess that the LM13700 isn't very good.
 
It might be OK as the small-influence AGC gain tweak, ahead of two good
integrating opamps.

 
At the cost of adding another component, how about adding another opamp
solely to regulate the output signal level? Say you want 2V p-p output,
sample some of the output, rectify it and compare it to a reference level
which is applied to the other opamp input.
Just my 2c.

 
You still need a multiplier of some sort to twaek loop gain. And
multipliers are nonlinear!
 
 

>
My vote would go to a comparator sensing the peak of the AC waveform, and
reducing the DC bias current a smidge each time it fires.
>
Changing the current on the peaks causes minimal change to the phase,
assuming that you do it right, and the tiny nonlinearity goes away in
between, rather than causing distortion all through the cycle.
>
Some months back, we talked about an audio amp design of Jim T’s that set
it’s quiescent bias that way, except switching near the zero crossing.
Turns out to be a startlingly good approach.

.<https://people.engr.tamu.edu/spalermo/ecen620/general_pn_theory_hajimiri_jssc_1998.pdf>

People don't much use Hajimiri for phase-noise estimates (Leeson being
adequate and simpler), but this is relevant to the audio-amp question.

Joe Gwinn